Epistemological Debate

This is a summarized version of a long debate I had on the epistemology of conscious experience, or the debate between Direct Perception v.s. Representationalism on the PSYCHE-D mailing list in April - June, 2005. I have selected choice threads that led to interesting exchanges, with links to the original messages.

This debate is interesting not only as a discussion of the nature of consciousness, but also as a prime example of a paradigmatic debate. In fact, this issue is perhaps the ultimate paradigm debate, because the alternative paradigms present such radically different inside-out-inverted views of the world relative to each other, no wonder the opposite camps could never reach agreement even on the meaning of terminology!

See also the Cartoon Epistemology

Steve Lehar


Index

Initial message
Lehar: refutation of Gibsonian concept
Brook: Proof of direct perception: Kick a chair
Rickert: Gibson's epistemological error?
Lehar: Theories v.s. Paradigms
Lehar: What would it take to convince you?
Sizemore: Conceptual v.s. Empirical Issues
Sizemore: Meaning of representation
Lehar: Ontology of Experience
Lehar: Summary Direct Percepton v.s. Representationalism
Brook: What more do you want?
Rickert: No clear meaning of Representation
Rickert: Computers don't compute
Chalmers: Terminological dispute
Lehar & Brooks: offlist debate
Lehar: Analogical or analytical?
Dalton: Representation is analogical?
Reason: Is the word "red" red?
Seager: How big is your experience?
Trehub: We do not perceive our perceptions
Lehar: More paradigmatic stuff
Lehar: Still more paradigmatic stuff
Brook: Hostile to sense data theory
Lehar: Epistemology of exeperience
Lehar: Concluding exchange



[Original Message]
Date: Wed, 6 Apr 2005 09:41:13 -0700
From: Stuart Hameroff
Subject: The geometry of conscious experience

Hi everyone

Regarding the most interesting discussion on projection geometry related to visual conscious experience between (primarily) Alex Green and Brian Flanagan, I highly recommend the work of Steven Lehar. See http://cns-alumni.bu.edu/~slehar/Lehar.html

He has written several books and a target article in Behavioral and Brain Sciences on this and related topics, and his work is not only insightful but (as you can see on his website) extremely well illustrated with artistic color cartoons.

Steven will be speaking at the next Tucson conference a year from now. See http://consciousness.arizona.edu/Tucson2006.htm

cheers

Stuart



[Original Message]
Date: Thu, 7 Apr 2005 07:09:55 -0700
From: Glen Sizemore
Subject: Re: The geometry of conscious experience

From Lehar's web site:

In fact there is no need for an internal observer of the scene, since the internal representation is simply a data structure like any other data in a computer, except that this data is expressed in spatial form (Earle 1998, Singh & Hoffman 1998). For if a picture in the head required a homunculus to view it, then the same argument would hold for any other form of information in the brain, which would also require a homunculus to read or interpret that information. In fact any information encoded in the brain needs only to be available to other internal processes rather than to a miniature copy of the whole brain. The fact that the brain does go to the trouble of constructing a full spatial analog of the external environment merely suggests that it has ways to make use of this spatial data.

Sizemore:
I do not think that the absurdities of representationalism go away so easily. It is becoming widely known that saying we see a representation immediately raises the issue of how seriously to take it. If it is a metaphor, then it is useless (unless we can show that it is likely to somehow be literal), but if it is taken literally, then it raises the issue of infinite regress.

So, the ploy is simply to say that the representation is information and information is "used" by some other part of the nervous system. By substituting some other word for "see" (here "use" is substituted) the glaring absurdity is obscured. But the notion of "using information" needs to be critically examined, and when it is, it can be seen to raise the same issues as "garden-variety" representationalism. In what sense do nervous systems "use information"? Does it mean any more than events cause receptors to "fire" which cause other neurons to "fire," and so on? If not, then we are simply saying that somehow physiology mediates the behavioral functions that we wind up calling "seeing," and "hearing," etc. and this only raises issues of infinite regress and homunculi if representationalism is assumed. If it means more than that, then we are probably back where we started. That is, in the conventional view, we say that we see the world, and that we do so by creating an inner copy that is seen. But we may easily say that animals (human and non-human)"use information" in the environment (colors, sounds, etc.) in order to behave in the ways that we observe and that they do this by creating an inner copy of the information which is what is "used." If "using" information in the environment requires further "usage" than why doesn't "using" the internal information?



[Original Message]
Date: Fri, 8 Apr 2005 09:42:17 -0500
From: Steven Lehar
Subject: Re: The geometry of conscious experience

The key issue is not the word "see" or "use", to describe how the brain makes use of internally represented information, it is the question of whether the internal representation is processed by other internal processes in the brain, or whether it requires a copy of the *whole brain* in order to "see" that represented data. Only the latter formulation leads to the infinite regress.

If Sizemore claims that a picture in your brain requires a miniature copy of your whole brain to "see" that picture, then surely this objection would apply to *any* information represented in the brain, including verbal, linguistic, and cognitive knowledge, all of which would require a miniature copy of the whole brain to interpret or process that cognitive information. Why is it that the homunculus objection is only raised against pictorial data? What is so special about image data that requires a homunculus to "see", when other data do not?

Now I acknowledge a profound philosophical issue here, that applies to *any* explanation of mental function, that is, after we are done explaining the mechanism of perception or cognition, there is the question of how come we get to *experience* that information processing. The mechanistic explanation of neural signals and sensory processing says nothing of the experience of perception and why we have it. Even if we have actual pictures in our brain, how is it that we *experience* those pictures? Why doesn't a computer experience the image data that it processes? That is indeed a profound issue, but again, it is one that applies to cognitive and verbal information processing just as much as to pictorial processing.

But whatever the explanation might be for the "Ultimate question" of consciousness, the indisputable fact remains that we *DO* in fact experience the operation of our brain during mental processing.

Steve



[Original Message]
Date: Sat, 9 Apr 2005 06:03:02 -0700
From: Glen Sizemore
Subject: Re: The geometry of conscious experience

Lehar >
The key issue is not the word "see" or "use", to describe how the brain makes use of internally represented information, it is the question of whether the internal representation is processed by other internal processes in the brain, or whether it requires a copy of the *whole brain* in order to "see" that represented data. Only the latter formulation leads to the infinite regress.
< Lehar

GS:
I disagree. The key issue is whether or not what has to be explained is simply placed inside the head (as a category error). We ask "What is 'seeing'"? and representationalism answers that "it is the 'seeing' of a representation." It doesn't matter if the answer is "it is one part of the brain 'seeing' the representation located in another part." Or: "it is one part of the brain 'making use of' a representation in another." Or: "it is one part of the brain "processing" the representation in another part."

Lehar >
If Sizemore claims that a picture in your brain requires a miniature copy of your whole brain to "see" that picture, then surely this objection would apply to *any* information represented in the brain, including verbal, linguistic, and cognitive knowledge, all of which would require a miniature copy of the whole brain to interpret or process that cognitive information.
< Lehar

Glen Sizemore >
I do not only raise the issue with respect to "pictoral data." I am critical of all metaphorical uses of 'representatation.' I put no stock in 'information' being 'represented' in the brain.
< Glen Sizemore



[Original Message]
Date: Sat, 9 Apr 2005 11:12:10 -0500
From: Steven Lehar
Subject: Re: The geometry of conscious experience

Lehar >>
The key issue is not the word "see" or "use", to describe how the brain makes use of internally represented information...
<< Lehar

Sizemore >
I disagree. The key issue is whether or not what has to be explained is simply placed inside the head...
< Sizemore

I acknowledge that placing a representation in the head does not solve the whole problem of how we see. It *is* however a *prerequisite* for being able to see something that that something must be represented in your brain. Is it not?

Sizemore >
I do not only raise the issue with respect to "pictoral data." I am critical of all metaphorical uses of "representatation." I put no stock in "information" being "represented" in the brain.
< Sizemore

Yow! That's pretty radical man! I'm not sure I can debate you if we don't even share this much in common!

Sizemore >
I claim that explanatory fictions like "representations," and earlier versions like "beliefs" and "knowledge," shed no light on what is going on in the nervous system ... What does it mean for "information" to be "processed?" ...
< Sizemore

Wow! Thats pretty bizarre!

Do you agree with the robot metaphor, that people are like a robot that receives sensory input, stores it in internal representations, and processes that information in order to compute an appropriate behavioral response? Do we at least agree on that metaphorical image?

If not, then I don't think we can have a meaningful exchange beyond just agreeing to disagree.

Steve



[Original Message]
Date: Mon, 11 Apr 2005 08:11:06 -0500
From: Steven Lehar
Subject: Direct Perception v.s. Representationalism

Ok, NOW I understand where you guys are coming from! Its the Gibsonian / O'Regan, "organism interacting with the environment" idea.

Well, in the first place, Representationalism has never been shown to be false. Au contraire, mon ami, the *principle* of representationalism has been actually demonstrated in robots that use video cameras for sensory input, computations in a computer brain, and behavior by way of servo actuators. Now admittedly current robots are pretty primitive and "robotic", and very different from animal perception. But if anyone has any ambiguity about the meaning of terms like "information", "representation", and "processing", just look at a robot and the concepts become perfecty clear.

Robots offer an *existence proof* that the *concept* of representationalism is *feasable* at least in principle.

So lets have no more talk about not knowing the meaning of information or representation or processing. Those terms are perfectly clear, and obviously workable in principle.

Furthermore, in the absence of *compelling* evidence to the contrary, a representationalist assumption is the most *reasonable* understanding of perception, given the eye that appears to work like a video camera, the optic nerve that sends data to the brain, given the complex wiring suggestive of computation in the brain, and motor neurons from the brain back out to the muscles as if to produce behavior. The representationalist thesis comes directly from inspection of the wiring of animal bodies, that looks for all the world *AS IF* it were a representationalist system! It may not be, but until the strong contradictory evidence comes in, that *IS* the most *REASONABLE* initial assumption!

...o0o...

Now, the Gibsonian / O'Regan / direct percepton concept, on the other hand, has *NEVER* been demonstrated in ANY kind of artificial system, and there is very good reason to believe that the whole concept is totally incoherent and impossible *IN PRINCIPLE*!

How would you even build a robot that works by direct perception??? Would you equip it with video cameras for eyes? If so, what do you do with the data generated by those cameras? You can't send it to the brain, because that would be computation and representation again, so we can save ourselves the expense of video cable and computer brain. But what would drive the servo-actuators? Where does that signal come from? And how does the direct-perception robot project its experience out of its body into the world so as to produce behavior *as if* it were actually *seeing* that environment *without* representations and computations???

HOW WOULD IT BE DONE????

The concept is so vague as to be totally meaningless! Gibson himself refused to discuss what gets sent from the eye to the brain, or what kind of computations might occur in the brain. In fact Gibson even denied that the retina records anything like an image! But he NEVER OFFERED ANY ALTERNATIVE EXPLANATION for how behavior works besides a few vague mumblings about being tuned to invariants in the environment. Gibson spoke as if the computation of perception occurs OUT IN THE WORLD, rather than in the brain. But there is no computational or representational *machinery* out in the world, the only machinery is found in the brain, right where the sensory nerves terminate, exactly AS IF the sensory organs were sending data to the brain for processing!

As for O'Regans concept of using the external environment as a representation of itself, that idea is DEMONSTRABLY FALSE, because probing the world with visual saccades, especially in the monocular case, is **nothing like** accessing a memory, internal or external, because every saccade presents only a two-dimensional pattern of light. The three- dimensional spatial information of the external world is **by no means** immediately available from glimpses of the world, but requires **the most sophisticated** and **as-yet undiscovered** algorithm to decipher that spatial information from the retinal input.

Secondly, O'Regan's concept of direct perception is totally inconsistent with the *phenomenal experience* of vision, where we do in fact experience the world as a spatial structure, and we perceive individual saccades to be located at the location in the global framework of space that we perceive that saccade to be located.

Central v.s. Peripheral Vision

Thirdly, the absurdity of O'Regan's concept is highlighted by the condition of *visual agnosia* (specifically, *apperceptive* agnosia) which is a visual integration failure. An agnosic patient can see individual features, but cannot distinguish a picture of a bicycle from a picture of disassembled *parts* of a bicycle. They can see a wheel here, and a wheel again, but they cannot tell whether the second wheel is not just a second glance at the first wheel, or if it is a separate wheel, they cannot see the spatial relation between the two wheels. In fact, the experience of visual agnosia is *exactly as if* vision worked as O'Regan proposes, with individual glances picking out features in the world in the absence of a global framework or spatial representation to store or record those features.

The condition of apperceptive agnosia is the absence of a visual function whose existence O'Regan effectively denies!

So don't be complaining about the vagueness of terms like information, representation, and processing. Those terms are **perfectly clear**, and are demonstrable in **actual physical robots** that receive sensory input and compute behavioral responses.

Instead, the real concern is the vagueness of terms like "active interaction with the environment" and "responsive to invariants in the environment" IF NOT by way of sensory input and internal representations. What does that even ***MEAN***?

Until the *principle* of direct perception is demonstrated in a simple robotic model, the concept is totally incoherent and ill-defined!

Steve Lehar



[Original Message]
Date: Mon, 11 Apr 2005 11:13:32 -0400
From: Andrew Brook
Subject: Re: Direct Perception v.s. Representationalism

I have been lurking in the recent discussion but I want to say that, modulo a few excess asterisks, Steve Lehar's message on anti-representationalism seems to me to be exactly right. Theorists who present themselves as denying the existence of representations almost always, on closer inspection, turn out to be denying that certain kinds of representations exist (Brooks) or to be denying that representations are as plentiful or play as big a role as the tradition would have it (O'Regan, Noe, Clark), or whatever. Some just refuse to talk about central issues at all (Gibson). Mounting a flat-out denial that there is anything in the brain that stands for, indexes, refers to, even pictures items other than itself is a pretty tough task. (Actually, Lehar might be a bit hard on O'Regan on this score, though some of his rhetoric invites Lehar's kind of response.)

Andrew



[Original Message]
Date: Sat, 9 Apr 2005 15:37:46 -0700
From: Augustin Carreno [stincarr@GMAIL.COM]
Subject: Re: The geometry of conscious experience

I think the argument against representation should be made based on physics. Namely, what is the nature of the environment-organism interaction allowing the brain to make a copy of it, how exactly can a copy of this environment be made? Is it a tape-like recording? A CD? And despite its enormous capacity, can the brain really afford to hold a copy of the entire universe together with whatever extra brain power is needed to analyze it? Judging by nature's preference for parsimony, the answer is no.



[Original Message]
Date: Tue, 12 Apr 2005 09:08:14 -0500
From: Steven Lehar
Subject: Re: Direct Perception v.s. Representationalism

Andrew Brook >
Steve Lehar's message on anti-representationalism seems to me to be exactly right. Theorists who present themselves as denying the existence of representations almost always, on closer inspection, turn out to be denying that certain kinds of representations exist
< Andrew Brook

Yes, I am more accustomed to debating people who contest the notion of *spatial* representations in the brain, given that neurophysiology has not (yet) found "pictures" in the brain. It threw me for a loop to find people like Glen Sizemore who deny representationalism altogether! Even the most ardent supporters of Gibson's theories generally take care to disclaim his most radical views (Bruce & Green 1987 p. 190, 203-204, Pessoa et al. 1998, O'Regan 1992 p. 473) although they present no viable alternative explanation to account for our experience of the world beyond the sensory surface.

Many of Gibson's observations on environmental affordances and invariants were very valuable and insightful, even from a representationalist viewpoint. In fact, it was Gibson's refusal to discuss physiology and computation that released him from the burden of having to consider issues of "neural plausibility", and that is why he dared to make such bold and generally valid observations on the nature of perception.

But Gibson's profound epistemological error backed him into a corner which is ultimately indefensible, which made him get defensive and dogmatic in his later years, as often happens to those who commit themselves to defending the indefensible.

Carreno >
I think the argument against representation should be made based on physics. ... can the brain really afford to hold a copy of the entire universe together with whatever extra brain power is needed to analyze it? Judging by nature's preference for parsimony, the answer is no.
< Carreno

I am viscerally sympathetic with this argument. Given what we know about neurophysiology, it seems totally implausible that the brain could construct a model of the world as rich and complex as our experience of it, and maintain it in real time as we move about in the world. Carreno is right: that does ideed stretch credulity to its elastic limit. But before deploying Occam's razor we must first balance the scales, and take a full accounting of the alternatives under consideration. For the alternative is that we experience the world directly, as if bypassing the causal chain of sensory processing. As incredible as representationalism might seem, the alternative is even more incredible.



[Original Message]
Date: Tue, 12 Apr 2005 12:16:30 -0400
From: Andrew Brook
Subject: Re: Direct Perception v.s. Representationalism

Lehar >

But before deploying Occam's razor we must first balance the scales, and take a full accounting of the alternatives under consideration. For the alternative is that we experience the world directly, as if bypassing the causal chain of sensory processing. As incredible as representationalism might seem, the alternative is even more incredible.

< Lehar

Fallacy of the excluded middle. There are other options. One is the one I sketched in the message from which Steve quotes: that our representations give us access to the world itself, not just to the end point of the representing process. Our brain has the capacity to work its way down the causal chain to experience the kickoff point. How we can do this is a wonderful mystery but that we do it seems, to me at least, pretty much beyond question. When I open my eyes, I see the world around me. This is to be in epistemic contact -- to see, know, experience, be conscious of -- the world, the part of it in my immediate vicinity anyway, not any representation or construct of mine.



[Original Message]
Date: Tue, 12 Apr 2005 14:46:32 -0500
From: Steven Lehar
Subject: Re: Direct Perception v.s. Representationalism

Brook >
How we can do this is a wonderful mystery but that we do it seems, to me at least, pretty much beyond question.
< Brook

Wonderful mystery indeed! Downright *miraculous*, wouldn't you say?

I mean, how would you demonstrate this in a simple robot model?

Experience or awareness of an object means having posession of information about that object, color, shape, location, etc. Exactly *as if* one had a miniature colored model of the object right there inside your brain. Except we *DON'T* ?

So how would this work in a robot model? The video camera picks up a 2-D image which is sent to the computer brain, that extracts a few features and performs a few computations, and suddenly, miraculously, a three- dimensional colored data structure appears--not in the computer brain in some kind of holographic 3-D imaging mechanism at the end of the causal chain, but right back out there in the world! With NO high-tech holographic imaging machinery involved!!! The image just appears out there, disconnected computationally from any of the hardware of the robot.

And when the robot closes its lens covers, the model out there DISAPPEARS! As if it were causally connected to the computational hardware downstream of the video signal, except it *ISN'T!*

Projection

I say again: Until the *principle* of direct percepton can be demonstrated in an artificial sensory system, the whole idea is completely implausible and ill-defined. I don't mean anything fancy, just a simple demo like the representationalist robot. *HOW* does the robot "work its way down the causal chain to experience the kickoff point" ???



[Original Message]
Date: Tue, 12 Apr 2005 21:27:14 -0400
From: Andrew Brook
Subject: Re: Direct Perception v.s. Representationalism

Steve,

I'm not sure what's going on here. We should be on the same side. Here is why I think that some form of direct realism is virtually unrejectable. When I open my eyes, I see a chair just as directly as when I kick it, I kick a chair. In both cases, it is the chair that I am in contact with. In neither case am I in (interesting, relevant) contact with any intermediary. And -- here is why the view is essentially unrejectable -- if you say, in either case, 'the contact is not direct', then I will invite you to tell me what you mean by 'direct'? What could be more direct than seeing or kicking an object in plain view in front of me? If this is not direct, what would be?

Simple robot model? I sketched how at the end of my message. In the same way that vision systems since Marr's have been able to extract three dimensional objects from two dimensional arrays, our vision system not only extracts threre dimensional objects but allows us 'reverse infer' down the causal chain to be directly aware of them. What more do you want?



[Original Message]
Date: Wed, 13 Apr 2005 10:19:40 -0500
From: Steven Lehar
Subject: Re: Direct Perception v.s. Representationalism

Brook >
I'm not sure what's going on here. We should be on the same side.
< Brook

Yeah, thats what *I* thought when you chimed in to reject the extreme Gibsonian (Sidemorian) version of direct perception! I guess that this issue is not a clean binary choice, but there are a number of intermediate positions between direct perception and representationalism.

Brook >
Here is why I think that some form of direct realism is virtually unrejectable. When I open my eyes, I see a chair just as directly as when I kick it, I kick a chair. In both cases, it is the chair that I am in contact with. In neither case am I in ... contact with any intermediary. What could be more direct than seeing or kicking an object in plain view in front of me? If this is not direct, what would be?
< Brook

Ok, stand in front of a chair, and before you kick it, touch your finger to one eyeball (through the eye lid) and push it gently to one side, until you see a double image. Now KICK! Now you see TWO chairs, and TWO feet kicking them! Which one is the "real" chair and the "real" foot? And what is the actual objective location of that chair? If this is not IN-direct, what would be?

Brook >
1. Simple robot model? I sketched how at the end of my message. In the same way that vision systems since Marr's have been able to extract three dimensional objects from two dimensional arrays, our vision system not only extracts three dimensional objects but allows us 'reverse infer' down the causal chain to be directly aware of them. What more do you want?
< Brook

Au contraire! Marr's vision model is entirely representational. From the moment the image registers on the (synthetic) retina, all computational processing of that image occurs inside the computer, or inside a brain. Nothing gets projected out into the world again! There is *NO* "reverse inference" going on here, but a feed-forward progression of *forward* inference, from input stimulus to internal mental model. The computer never has access to *any* information that is not explicitly represented in the machine. That is an indirect representational algorithm!



[Original Message]
Date: Thu, 14 Apr 2005 14:20:55 -0500
From: Neil W Rickert
Subject: Re: Direct Perception v.s. Representationalism

Steven Lehar wrote on Apr 12, 2005:

But Gibson's profound epistemological error backed him into a corner which is ultimately indefensible, which made him get defensive and dogmatic in his later years, as often happens to those who commit themselves to defending the indefensible.

Yet I could nowhere find a clear statement of what Gibson's "profound epistemological error" is presumed to be.

For sure, Gibson avoided questions of mechanism. But that hardly seems to be an epistemological error.

I am quite sure that Gibson understood the physics and biology, and recognized that stimulation of retinal cells and the transmission of signals on the optic nerve were part of the causal processes involved in that direct perception.



[Original Message]
Date: Sun, 17 Apr 2005 09:37:47 -0500
From: Steven Lehar
Subject: Re: Direct Perception v.s. Representationalism

Rickert >
Yet I could nowhere find a clear statement of what Gibson's "profound epistemological error" is presumed to be.
< Rickert

For a summary of the epistemological issue and the various alternatives, see

The Epistemology of Conscious Experience

Direct perception embodies a profound epistemological error.

Rickert >
But I am quite sure that Gibson understood the physics and biology, and recognized that stimulation of retinal cells and the transmission of signals on the optic nerve were part of the causal processes involved in that direct perception.
< Rickert

Actually, Gibson made his views on the role of the retina perfectly clear, and they were very much at odds with the consensus view on it.

Gibson, J. J. (1966) The Senses Considered as Perceptual Systems. Boston: Houghton Mifflin.

p. 263:

"The very idea of a retinal pattern sensation that can be impressed on the neural tissue of the brain is a misconception, for the neural pattern never even existed in the retinal mosaic. There can be no anatomical engram in the brain if there was no anatomical image in the retina. The retina jerks about. It has a rapid tremor. It even has a gap in it (the blind spot). It is a scintillation, not an image. ... The whole idea stems from the persistent myth that there has to be something in the brain that is visible, and from Johannes Müller's assumption that the nerves telegraph messages to the brain."

The reason why Gibson denied that the retina records an image and transmits it to the brain, is that to even allow this much representationalism in the visual process is to acknowledge that the principle behind representationalism is perfectly feasable, and that the first stage of visual processing is apparently representational.



[Original Message]
Date: Tue, 19 Apr 2005 15:33:56 -0400
From: Andrew Brook
Subject: Re: Direct Perception v.s. Representationalism

Steve Lehar said (among much else):
Ok, stand in front of a chair, and before you kick it, touch your finger to one eyeball (through the eye lid) and push it gently to one side, until you see a double image. Now KICK! Now you see TWO chairs, and TWO feet kicking them! Which one is the "real" chair and the "real" foot? And what is the actual objective location of that chair? If this is not IN-direct, what would be?

I respond:
Well, it is still the chair that we see -- and we have extremely reliable means for sorting out when the perceptual medium is working well and when it is distorting things. All I want to insist on is that when we see, we see the world (most of the time). We see *via* various media, but we not aware of these media (except when things go wrong or we otherwise pay attention to them), we are aware of the things in the world that kick them into action. That's all.

Try another way:
of course there is intermediate *machinery" but that does not mean that the resulting *perception* or *consciousness* is indirect, i.e., by inference from something in the head.

Steve:
Unless, that is, you mistake our experience for a direct view of the world itself, bypassing the chain of sensory processing, which in impossible in principle!

Me:
Why should a 'direct view of the world' have to bypass sensory processing? It proceeds via, though, sensory processing. What makes it direct is that it makes us aware of the world, not the processing. How it does this is, at this point in history, mostly wide-open.

Steve:
It is extremely difficult to banish the last vestiges of naive realism from our philosophy.

Me:
I don't know what naive realism is but I certainly do not want to banish realism because it is true! (IMHO, of course.)

Andrew



[Original Message]
Date: Tue, 19 Apr 2005 10:09:43 -0500
From: Steven Lehar
Subject: Theories v.s. Paradigms

Ok, we've been round and round the direct perception v.s. representationalism debate enough times to see that nobody is about to change their minds, no matter HOW eloquent or persuasive the arguments are on either side. Why is this so?

This is a sure sign of a *paradigm* debate!

As Kuhn explained, (Kuhn 1970 The Structure of Scientific Revolutions) there is a profound difference between theories and paradigms, and how to debate them. The problem is that what we are debating is not a question of theory, like the question of whether the representation in the brain is digital or analog, or whether time-to-collision information is available from the optic array, which would be resolved by the normal rules of logic and evidence. But in this case what we are debating are the *foundational assumptions* with which we come to the debate in the first place. Debates between paradigms tend to go round and round in futile circles, because the participants are debating from different foundational assumptions. We take our foundational assumptions as a *given*, as obviously self-evident *fact*, and from that perspective the opposing paradigm appears patently absurd, it makes us wonder how intelligent, educated people could possibly defend such an absurd and indefensible view.

But if alternative paradigms are to be fairly evaluated, it is necessary to temporarily and provisionally suspend one's own paradigmatic assumptions, (a feat that many find impossible to do) and accept the assumptions of the alternative paradigm **as if they could actually be true**. Only then can the competing paradigms be fairly compared, not on the basis of the perceived incredibility of their initial assumptions, but on the overall coherence and self-consistency of the world view that they implicate in total.

In the case of our debate, we hear some state that it is "obvious" that we experience things directly, while others state that it is "obvious" that perception is indirect. If we begin with either of those assumptions, we are sure never to reach agreement. In the case of the historical debate between an earth-centered or sun-centered cosmos, the earth-centered people argued that the idea of the whole earth with all its mountains and forests and oceans spinning and flying through space is so absurd and incredible on the face of it, that it does not matter what evidence you might cite, they would never be convinced!

There is a parallel with the current discussion, because as in that ancient debate, there is a certain asymmetry in the two views: one alternative is the "naive" view, in the sense that that is the view that we adopt by default, even before giving the question any serious thought, while the other view appears initially to be patently absurd.

In paradigmatic debates, the argument that one view "seems incredible" is no valid argument. Many of the greatest discoveries of science seemed initially to be so incredible that it took decades or even centuries before they were generally accepted. But accepted they were, eventually. And the reason why they were accepted was not because they had become any less incredible. Facts such as the immensity of the universe, and its cataclysmic genesis from a singularity in space and time, as well as the smallness of the atom, or the bizarre properties of quantum phenomena, are just as incredible today as they were when they were first discovered. And yet all of these incredible theories have taken their place in the realm of accepted scientific knowledge, not because they have become any less incredible since they were first proposed, but because the evidence for them has been irrefutable. In science, irrefutable evidence triumphs over incredibility, and this is exactly what gives science the power to discover unexpected or incredible truth.

There is an asymmetry in this debate: all representationalists were once naive realists, whereas most direct perceptionists have never been representationalists. I can tell you that I find representationalism **just as incredible** as any of you do. It is absolutely incredible that my physical skull should be larger than the dome of the sky. And it is incredible that the brain can construct and maintain a real-time volumetric moving image of the world with the rich detail and fidelity of the world I see around me. Given current knowledge of neurophysiology, that appears to be absolutely incredible!

But before we deploy Occam's Razor, we must first balance the scales and take a full accounting of the alternatives under consideration. For the alternative is that we can somehow become aware of objects and surfaces in the external world *without* the mediation of sensory processing and internal representations. That, in my view, is not just incredible, it is incoherent, as demonstrated by the fact that nobody has ever, or could ever possibly build a robot that can demonstrate the *principle* behind direct perception in a simple model.

I am therefore suspicious of direct perceptionists who focus exclusively on the incredible aspects of representationalism, without also acknowledging the incredible aspects of direct perception. I can accept someone who argues that both views appear incredible, but that they consider this view to be somewhat less incredible than that one. That is a valid and reasonable position. But anyone who does not see the profound problems in direct perception (as I see the profound problems in representationalism) is suspect of being a paradigmatic partisan, that they accept one view as plainly obvious, thus requiring no further proof, while the other appears patently absurd no matter what the evidence. One suspects of such people that they never really understood the alternative position enough to give it any serious consideration.

To those people I implore that they entertain the *possibility* that they may perhaps be mistaken.

"There are more things in heaven and earth, Horatio, than are dreamt of in your philosophy."

-William Shakespeare, Hamlet, Act I, Scene 5.

Steve Lehar



[Original Message]
Date: Tue, 19 Apr 2005 15:03:40 -0500
From: Neil W Rickert
Subject: Re: Theories v.s. Paradigms

Steven Lehar wrote on Apr 19, 2005:

Ok, we've been round and round the direct perception v.s. representationalism debate enough times to see that nobody is about to change their minds, no matter HOW eloquent or persuasive the arguments are on either side. Why is this so?

Steven perhaps sees this as a problem. I don't. While I tilt toward the direct perception side, I am not all all concerned that Steven and Alex are strongly committed to representationalism. We don't need to put all of our eggs in the one basket. Science is best served when a problem is studied from several different perspectives. And may the best perspective (whichever that is) win.

There is an asymmetry in this debate: all representationalists were once naive realists, whereas most direct perceptionists have never been representationalists.

I don't think I agree with Steven's assessment.

No doubt naive realism is pretty much the received view. But direct perception is not the same thing as naive realism. I suspect that if people with an ordinary science education were asked to choose between Gibson's account of vision and Marr's account of vision, most would prefer Marr's account. So the default view would be closer to that of the representationalists than to that of the direct perceptionists. There are subtleties to Gibson's theory that make it a little difficult to appreciate.

But before we deploy Occam's Razor, we must first balance the scales and take a full accounting of the alternatives under consideration. For the alternative is that we can somehow become aware of objects and surfaces in the external world *without* the mediation of sensory processing and internal representations.

However, that is not the alternative. It seems that Steven can only see one side of the paradigm shift.

The alternative, or one alternative, is that we can become aware of objects and surfaces *with* the mediation of sensory processing, but *without* the mediation of internal representations.

Incidently, I don't doubt that some of our interactions with the world are mediated by internal representations.

-NWR



[Original Message]
Date: Wed, 20 Apr 2005 09:52:59 -0500
From: Steven Lehar
Subject: Re: Theories v.s. Paradigms

Reply to Neil Rickert:

Rickert>

Steven perhaps sees this as a problem. I don't. While I tilt toward the direct perception side, I am not all all concerned that Steven and Alex are strongly committed to representationalism. We don't need to put all of our eggs in the one basket.
< Rickert

Are you serious? It does not bother you in the least that people believe two mutually contradictory theories of perception? Surely the goal of science is to discover which of those two views is right, and which is wrong. I am passionately interested in that question!

slehar >>

But before we deploy Occam's Razor, we must first balance the scales and take a full accounting of the alternatives under consideration. For the alternative is that we can somehow become aware of objects and surfaces in the external world *without* the mediation of sensory processing and internal representations.
<< slehar

Rickert >
However, that is not the alternative. It seems that Steven can only see one side of the paradigm shift. The alternative, or one alternative, is that we can become aware of objects and surfaces *with* the mediation of sensory processing, but *without* the mediation of internal representations.
< Rickert

And exactly *HOW* would this be implemented in a simple robotic model? How can a robot "become aware" (aquire immediate parallel access) to environmental information *without* the mediation of internal representations?

We hear these *words* loud and clear, but the *concept* behind the words remains as clear as mud. Again, until you can demonstrate the *principle* behind direct perception in a simple model, this concept is so vague as to be virtually meaningless.

...oOo...

Given my comments at the launching of this "Theories v.s. Paradigms" thread, isn't it perfectly clear what is happening here?

Isn't it perfectly clear that Rickert himself does not have a clear idea of what he means by direct perception, or at least not clear enough to tell us how to build the robot to demonstrate the principle. And yet he is strangely blind to this gaping hole in his concept of perception. He appears to be strangely blind to the profound problems inherent in the concept of direct perception, he does not even acknowledge that any kind of problem exists.

Indeed, Rickert exhibits all the signs of a paradigmatic partisan. He is arguing from the initial assumption that perception is direct, but he is unable to question that initial assumption itself. To him that assumption appears so manifestly obvious that it requires no explanation.

I believe that the whole notion of direct perception is an elaborate rationalization to try to make some kind of logical sense out of the profoundly paradoxical observation that experience is outside of our head, as if in the world itself, and yet vision is clearly representational, from the retina on in to the brain.

But at a very deep level there is one small part of Rickert's mind that is aware of this paradox, although his conscious mind is in denial over that recognition. So, like Gibson, his response is to get more dogmatic and emphatic about his certainty that his view is right, even though he cannot explain it to us in any kind of detail, all he can do is to emphasize again and again that you can perceive the world directly without representations.

Until Rickert can find it in himself to acknowledge the *problem* that we are discussing, that is, the difficulty of actually implementing direct perception as he conceives it, further debate is simply useless, because we are arguing from different foundational assumptions. Rickert BEGINS with the assumption that perception is direct, so he cannot contemplate the possibility that it might not be.

Steve



[Original Message]
Date: Tue, 19 Apr 2005 20:47:36 -0400
From: Andrew Brook
Subject: Re: Theories v.s. Paradigms

Steve Lahar represents direct realism, a position I hold, this way:

"We can somehow become aware of objects and surfaces in the external world *without* the mediation of sensory processing"

But that is not my view at all!!!! As I have said over and over and .... OVER! (The caps may get me censored.) Nor, given that it is an utterly implausible view, is it the view of most other direct realists. A rough approximation of my view would be:

"we can somehow become directly, i.e., noninferentially, aware of objects and surfaces in the external world *through* the medium of sensory processing"

Furthermore, I don't think the difference between us is a paradigm shift difference, I think it is a difference induced by an implacable belief, totally untouchable by evidence or argument, that *if* there is a medium of perception, *then* the resulting perceptions cannot be direct. To which I respond:

You can define the word *direct* this way if you want, but if you define it as it is usually defined, that we are aware of objects around us, not just end-products of sensory processing in our brains, then the inference needs argument and evidence to support it. And we have seen not a single bit that the direct realists in the crowd have not been able to deal with handily.

Andrew



[Not released on PSYCHE-D (censored?)]
Date: Wed, 20 Apr 2005
From: Steven Lehar
Subject: Re: Theories v.s. Paradigms

Lehar >>
"We can somehow become aware of objects and surfaces in the external world *without* the mediation of sensory processing"
<< Lehar

Brook >
But that is not my view at all!!!! As I have said over and over and .... OVER! (The caps may get me censored.)
< Brook

Well in that case you are not the target of that criticism.

Brook >
we are aware of objects around us, not just end-products of sensory processing in our brains
< Brook

And exactly *HOW* would that occur??? If you cannot explain how this concept might be implemented in an artificial robot, then the idea is so vague as to be *meaningless*! Simply stating "over and over and .... OVER!" that perception is direct is just not going to cut it!

Steve



[Original Message]
Date: Wed, 20 Apr 2005 11:17:11 -0700
From: Glen Sizemore
Subject: Re: Theories v.s. Paradigms

Lehar:
There is an asymmetry in this debate: all representationalists were once naive realists, whereas most direct perceptionists have never been representationalists.

Sizemore:
Obviously, you are asserting that "representationalism" is a view battling for acceptance, but nothing could be further from the truth. The fact is that it is one of the first academic positions concerning human behavior to be taken seriously. It wasn't really until the 20th century with behaviorism (and I include later Wittgenstein here) that any serious challenge was mounted. Mental "theories" of representation simply morphed seamlessly into neurobiological "theories".



[Original Message]
Date: Wed, 20 Apr 2005 15:32:56 -0400
From: Andrew Brook
Subject: Which came first

I agree with Glen as against Steve on which view comes first. Though I agree with Steve that some kind of inchoate realism would seem true to most people who have never studied cognition, I was thoroughly indoctrinated in indirect representationalism in school as though it was the obvious and even the only conceivable point of view (Steve?). It then took me years to battle my way out of that bewitchment and back to a view that allowed me to accept what seems to me obviously true -- that I see, touch, taste, feel, and smell things other than myself, especially including other people. One good name for my view (he said being deliberately provocative) would be: direct representationalism.



[Original Message]
Date: Fri, 22 Apr 2005 10:27:21 -0500
From: Steven Lehar
Subject: Re: Which came first

Which comes first? When we are children we begin with the most naive of naive realism--that the world of experience is the world itself, viewed directly where it lies.

Then we learn about the eye with retina and optic nerve that project to the brain, an obviously representational system. At this point our mental image of the problem goes into a bistable state. We understand the causal chain of vision, but at the same time we observe that the experience, from the end of the causal chain, jumps back out of your head again and appears back out in the world, "like a stretched rope when it snaps", as Bertrand Russell expressed it. This is the state in which direct realists become permanently stuck.

Andrew Brook>
One good name for my view ... would be: direct representationalism.
< Andrew Brook

A very apt name for a befuddled epistemology! Experience is both at the end of the causal chain, and it is also back out at its beginning. Vision is obviously representational, at least as far as the retina (except for extremists like Gibson and Sizemore) and yet at the same time perception is direct, as if bypassing the retina and viewing the world directly again. Paradoxically, both appear to be true simultaneously. "Direct representationalism" indeed!

It is only by really taking representationalism seriously that the epistemological paradox is resolved, although that comes at the cost of an almost *incredible* neurophysiological hypothesis, that the brain is capable of constructing a three-dimensional volumetric real-time moving model of the external world with as much rich spatial detail as you see in the world around you.

Even if you continue to find that hypothesis too incredible to swallow, will you not at least admit the profound paradox inherent in the direct perception view? As I said in the "Theories v.s. Paradigms" thread, if you cannot even see that there is a paradox at all, then further debate is really a waste of time.



[Original Message]
Date: Fri, 22 Apr 2005 05:51:30 -0500
From: Steven Lehar
Subject: What would it take to convince you?

We've beaten this direct perception v.s. representationalism debate nearly to death, without much progress. And yet one view is right, and the other is wrong. One day the conclusive evidence will come in that will finally settle the issue once and for all. Maybe even in our own lifetimes! That raises the question:

What would it take to convince you? What kind of evidence can you possibly imagine that would settle the issue finally and conclusively? Or are we all so dogmatic that we will go to our graves in obstinate denial no matter what the evidence?

Just for the fun of it, let me hypotheticalize two alternative future scenarios that once and for all *PROVE* the correctness of representationalism, and of direct perception, respectively. Would the paradigmatic partizans among us be convinced by these? I suspect the responses might be illuminating.

H1: REPRESENTATIONALISM IS PROVEN!

It is discovered in the future that the brain is one giant resonator, just humming with one global resonance through all its tissues, right down to the spinal cord. This sets up volumetric spatial standing waves in the various cortical areas, in patterns like three-dimensional Chladni figures.

Chladni figures

More Chladni figures

Although the different cortical areas are connected, they are also independent resonators, each one sustaining its own spatial standing wave, as in a separate Chladni plate. And yet they are also coupled, so that the pattern in one cortical area is coupled to the patterns in all the other areas. If you modulate the resonance in one area, it has an influence on the pattern in all the other areas simultaneously. This immediate parallel coupling between resonating areas turns out to be the solution to the "binding problem". If you *were* the standing wave in your brain (which in fact you *are*), there would be no way for you to tell that your resonance is distributed over different resonators, because the pattern of your resonance would be unified.

Images in Visual Cortex

Careful neurophysiological recordings reveal that the experience of volumetric substance and void, solid matter surrounded by empty space, is encoded by the *phase* of the vibration: positive phase within perceived objects, and negative phase in the surrounding void. Experienced colors correlate with a cylic phase representation (as in the NTSC color television standard, for those who are familiar) which finally explains why phenomenal color defines a circular space (the color circle) even though the spectrum is linear. When a subject views a three-dimensional scene, neurophysiologists can sample that volumetric experience in various parts of the visual cortex, and actually "read" what the subject is experiencing in any portion of his visual space by the resonance in that part of his cortex.

Image in Visual Cortex

When a person turns their head, or moves about in the world, the spatial image in each of the cortical areas rotates and translates in synchrony, like the images in an array of television sets in a shop display that are all tuned to the same channel.

If the mapping of experience in the brain were decoded so that *every aspect* of experience could be deciphered from the outside with appropriate probes, which revealed a complete world of experience all encoded inside your brain, and if neurophysiologists could transmit signals into the visual cortex and predictably cause the subject to experience a red square, or a blue circle, or whatever, by using the right Fourier code, would THAT be sufficient to finally convince the doubting direct perceptionists out there?

I suspect not.

Ok, then lets switch to hypotheticalization #2

H2: DIRECT PERCEPTION IS PROVEN!

Uh, here I have a little more difficulty imagining the "experimentum crucis" (Oops! My paradigmatic partizanship is showing!) But lets give it a try regardless.

In the future it is discovered that there are in fact NO representations in the brain! Even the image on the retina is not really an image as such, transmitted from the eye to the brain, but instead, the eye is a tool for active exploration of the environment that detects environmental invariances OUT IN THE WORLD where the objects of perception reside.

This bizarre notion is finally proven beyond a shadow of doubt with the invention of the "experience meter", a device tuned to some quantum- mechanical cat-in-the-box state of existence in the world, so that it can detect when a material object is being experienced by someone. If you point the experience meter at an object in front of a person, the meter lights up when that person's eyes are open, and blinks out when their eyes are closed.

Blink on, blink off

The meter has a screen that shows the information content of the subject's experience--monochrome for people who are color blind, full color for people with normal vision. When a subject with visual agnosia views an object, the experience meter displays a chaotic jumble of ever shifting visual features instead of a coherent image. There is a peculiar void, or loss of signal from the rear faces of objects that are not exposed to the subject's direct view, as well as for background objects that are occluded by foreground objects. This finally proves that experience is not some mysterious non-existent entity, but a real physically measurable quantity that exists out in the physical world, not in a person's brain.

The experience meter also records *affordances*. When a subject views a chair, there is a measurable "sit-onable" affordance detected that appears superimposed on the chair, and that affordance has the mysterious property that it has a causal influence on the subject's motor system, so as to tend (when circumstances are right) to make the subject actually move toward the chair and attempt to sit on it. Furniture manufacturers use the experience meter to objectively measure the "sit-onable" appeal of various furniture designs. Neurophysiologists map out in exquisite detail the relation between the measured affordances, and the spatial influence that they exert on the various muscles of the body, by some kind of mysterious action-at-a- distance from the affordance out in the world, to the muscles inside the body.

Would THAT be sufficient to finally convince the doubting representationalists out there?

I gotta say, *I* would be convinced!

Would you?

Steve



[Original Message]
Date: Fri, 22 Apr 2005 13:39:50 -0400
From: Andrew Brook
Subject: Re: What would it take to convince you?

Steve, absolutely the right approach, ingeniously carried out. Trouble is, it does not address the issue between, for example, you and me. I can't imagine anything that would 'prove' direct perception *as you describe it*, anything compatible with what we now know about the brain, anyway. But within representationalism, there are two houses.

There is the house of those who think 'If a representational medium is present, the results can only be indirect perception/consciousness.'

And there is the house of those who think, 'The right kind of representational medium lets us go right through it all the way to the world itself, so that the result is direct perception/consciousness.'

For me, this debate between direct and indirect representationalists, which is within representationalism, is the interesting one.

Andrew



[Original Message]
Date: Fri, 22 Apr 2005 15:30:11 -0500
From: Steven Lehar
Subject: Re: What would it take to convince you?

Andrew Brook >

Steve, absolutely the right approach, ingeniously carried out. Trouble is, it does not address the issue between, for example, you and me. I can't imagine anything that would 'prove' direct perception *as you describe it*, ... But within representationalism, there are two houses.

[1]:
There is the house of those who think 'If a representational medium is present, the results can only be indirect perception/consciousness.'

[2]:
And there is the house of those who think, 'The right kind of representational medium lets us go right through it all the way to the world itself, so that the result is direct perception/consciousness.'


< Andrew Brook

Well then can you imagine the "experimentum crucis" that would select between THOSE two alternatives?

If not, then what does the second alternative actually *mean*?



[Original Message]
Date: Fri, 22 Apr 2005 12:02:14 -0700
From: Glen Sizemore
Subject: Re: What would it take to convince you?

Lehar has conflated conceptual and empirical issues. The only slight difference is that he considers future experiments, whereas most cognitivists claim that representationalism has already been proven. Once again, the notion that psychological phenomena require current representations and stored and retrieved representations (as well as a host of related notions like expectation, unconscious inference, unconscious rule-following, etc. etc.) for their explanation is an assumption, as is the view that they do not.

As long as these views remain assumptions, and it is not clear that they can ever be anything else, philosophical debate and conceptual analysis is the only avenue via which they can be compared. Did Lehar not, in fact, argue this himself when he invoked Kuhn a couple of days ago?

No one has this stuff worked out by any conceivable stretch of the imagination, but, quite bluntly, the representationalists think that they do and this faith is, I assert, largely a product of the ubiquitous, literal, usage of real representations such as photographs and paintings etc. It is bewitchment by metaphor.



[Original Message]
Date: Fri, 22 Apr 2005 16:10:43 -0500
From: Steven Lehar
Subject: Re: What would it take to convince you?

But even paradigmatic issues can be determined by experiment, at least in principle. The earth-centered cosmos has now been conclusively rejected by the "experiment" of flying around the back side of the moon without colliding with the crystal sphere that supposedly supports it in space. Although this experiment was technologically not feasable in Ptolomy's time, Ptolomy and Copernicus could have predicted the different outcomes of this "experimentum crucis" based on the two theories.

Any "theory" that is *NOT* testable by any experiment, even in principle, is not a scientific theory at all, but a pure *belief*. For example the existence of immaterial souls, if they are by definition undetectable by physical means.

If you cannot describe an experiement that could prove direct perception at least *in principle* with some kind of future technology, then direct perception is a *belief* not a theory, since it predicts nothing different than the alternative representationalist hypothesis.

But the truth is that the principle of representationalism *is* demonstrable in a simple robotic system, while the principle of direct perception remains as mysterious as the immaterial soul! After all these rounds of debate, we still have no idea how such a system could possibly be built in a real physical system.

Steve



[Original Message]
Date: Fri, 22 Apr 2005 14:12:59 -0500
From: Neil W Rickert
Subject: Re: What would it take to convince you?

Lehar >

H1: REPRESENTATIONALISM IS PROVEN!

...If the mapping of experience in the brain were decoded so that *every aspect* of experience could be deciphered from the outside with appropriate probes, which revealed a complete world of experience all encoded inside your brain ... would THAT be sufficient to finally convince the doubting direct perceptionists out there?

I suspect not.

< Lehar

You are right. That would not convince me.

Lehar >

H2: DIRECT PERCEPTION IS PROVEN!

[some of the details snipped]

This bizarre notion is finally proven beyond a shadow of doubt with the invention of the "experience meter", a device tuned to some quantum- mechanical cat-in-the-box state of existence in the world, so that it can detect when a material object is being experienced by someone. If you point the experience meter at an object in front of a person, the meter lights up when that person's eyes are open, and blinks out when their eyes are closed.

< Lehar

This would convince me that I should consult the Amazing Randi, so that he can uncover and debunk the trickery involved in the demonstration.

The trouble with Steven's two examples is that they are both quite implausible. They look like sleight of hand parlor tricks.

A scientist needs better evidence than that.

To convince me, I need a detailed account of the relevant processes. This should, preferably, be at the level of information processing. It should plausibly account for all of the unresolved questions. And it should be supported by empirical evidence.

-NWR



[Not published on PSYCHE-D. Censored?]
From: Steve Lehar
Date: 4/22/2005
Subject: What would it take to convince you?

Neil Rickert >
You are right. That would not convince me.
< Neil Rickert

Well then is there *ANY* possible future experiment that would prove it to your satisfaction? If not, then you are a hopeless paradigmatic partisan.

Rickert >
The trouble with Steven's two examples is that they are both quite implausible. They look like sleight of hand parlor tricks. A scientist needs better evidence than that.
< Rickert

What *kind* of evidence are you talking about? Describe the experiment!

Rickert >
To convince me, I need a detailed account of the relevant processes. This should, preferably, be at the level of information processing. It should plausibly account for all of the unresolved questions. And it should be supported by empirical evidence.
< Rickert

Notice that the direct perceptionists are always long on high sounding verbiage and vague concepts, but very short on specific mechanisms and mechanical details.

They cannot describe a simple robotic system that would demonstrate the *principle* of direct perception in an actual physical system, and they cannot describe a future experiment that would prove it one way or another.

All this just confirms my original suspicion that the concept of direct perception is no *theory* at all, but merely a series of vague rationalizations used to justify their naive realist intuitions. It is impossible to pin them down on *any* kind of specifics.

Steve



[Original Message]
Date: Sat, 23 Apr 2005 08:33:27 -0400
From: "Glen M. Sizemore"
Subject: Conceptual vs. Empirical Issues

Lehar:
But even paradigmatic issues can be determined by experiment, at least in principle. The earth-centered cosmos has now been conclusively rejected by the "experiment" of flying around the back side of the moon without colliding with the crystal sphere

Sizemore:
This only shows that "paradigmatic issues" are not necessarily "conceptual issues." Besides, you yourself emphasized the incommensurability of "opposing paradigms."

Lehar:
Any "theory" that is *NOT* testable by any experiment, even in principle, is not a scientific theory at all, but a pure *belief*.

Sizemore:
But, as I have pointed out, the issues in question are NOT theoretical issues. They are conceptual issues. They concern the ASSUMPTIONS that underlie theory. And the empirical side of things is frequently distinct from both theory and concepts

Lehar:
If you cannot describe an experiement that could prove direct perception at least *in principle* with some kind of future technology, then direct perception is a *belief* not a theory, since it predicts nothing different than the alternative representationalist hypothesis.

Sizemore: Yes. Whether or not there are representations that are stored and retrieved etc. etc. are [not] beliefs, they are assumptions.

Lehar:
But the truth is that the principle of representationalism *is* demonstrable in a simple robotic system, while the principle of direct perception remains as mysterious as the immaterial soul! After all these rounds of debate, we still have no idea how such a system could possibly be built in a real physical system.

Sizemore:
I don't think that representationalism is "demonstrable in a simple robotic system". ... showing orderly relations between "the world" and a "pattern of neural activity" is not "proof" of representation. Representation, like "computation," is defined by its functional relation to the behavior of an animal, human or otherwise. To invoke "representation" is to invoke behavioral characteristics of whole animals, hence the charge of homunculism.

An alternative view is that perception is behavior, and that behavioral function is mediated by physiology. Unless all "mediation" is "representation" (and I argue it is not) this simple statement constitutes the beginning of a scientific approach. You are correct that "we still have no idea how such a system could possibly be built in a real physical system," and we have representationalism to thank for the wild-goose chase that constitutes much of "cognitive neuroscience." We understand, in some complete sense, behavior at about the level of habituation of the gill-withdrawal reflex in Aplysia, and maybe some classical conditioning of the system (and despite the language that Kandel uses, there is little to be called "representation" here, unless everything orderly is representation). Little wonder that "we still have no idea how such a system could possibly be built in a real physical system," if the system in question is the behavior we call seeing and hearing etc., and the other behavior of which it is a part. We know some of the behavioral observations that need to be explained, at least to the extent that some portions of psychology actually demonstrate behavioral regularities in individual subjects, but we do not yet know how physiology mediates such behavior. But to depend on the obviously category-error-ridden conceptual muddle that constitutes representationalism is no solution.



[Original Message]
Date: Sat, 23 Apr 2005 15:25:20 -0500
From: Steven Lehar
Subject: Re: Conceptual vs. Empirical Issues

The question whether psychological phenomena require internal representations is not just a conceptual question, as Sizemore suggests, it is also an empirical question that will one day be confirmed one way or the other experimentally, as soon as we figure out the code of the brain. Because the theory of direct perception states that there is no need for internal representations, or in the softer version defended by Andrew Brook, the internal representations do not need to encode ALL of the information of experience, because SOME of that information can be perceived directly *through* the representation (whatever that means). In any case, if it is discovered that the brain actually *does* explicitly encode *ALL* of the information in our experience, that will pretty much prove direct perception to be false.

Of course there will still be dogmatic defenders of direct perception even after that discovery, such as Glen Sizemore who tells us that he would not be convinced even if representations *are* found in the brain, because we would still not understand how we *see* those representations. (But then neither does Direct Perception tell us how we can *see* the world *without* representations, so the mystery of experience remains regardless.)

But you can never convince *everyone*. After all, there are still people who believe in God and intelligent design, as opposed to evolution, despite the overwhelming evidence to the contrary.

The *only* kind of paradigm that remains purely conceptual, in Sizemore's usage, are theories that make no testable predictions whatsoever. For example a version of direct perception that posits that there *are* complete and explicit representations in the brain that encode *all* the information of experience, but that perception is still direct, and does not actually *use* those representations. Or the theory that God did design the world, but that he used evolution has his mechanism of creation. Those kinds of theories are indeed purely conceptual, not at all empirical, because the experimental evidence is identical for both alternatives. But such theories are not theories at all, they are *beliefs*, and thus fall outside the realm of science.

Steve



[Original Message]
Date: Sat, 23 Apr 2005 15:35:20 -0500
From: Steven Lehar
Subject: Re: Conceptual vs. Empirical Issues

The question of Conceptual v.s. Empirical issues raises a question about Andrew Brook's softer concept of direct perception *through* representations.

Question to Andrew Brooks:

Is there *any* information in our experience that is *not* explicitly represented in the brain? Like the information about the world that is experienced *through* the representation instead of *in* it? If so, then your concept of direct perception *through* representations could be tested in principle by seeing whether that information is in fact encoded in the brain or not.

If on the other hand you posit that *all* of the information in our experience *is* explicitly encoded in the brain, but we still view the world "directly" *through* that representation, then we are in perfect agreement, that in veridical (non-illusory) perception we are viewing the world "directly" in that sense. But in that case your theory becomes indistinguishable from represntationalism. (Viewing "directly" *through* a representation strikes me as a contradiction in terms.)

Steve



[Original Message]
Date: Sun, 24 Apr 2005 15:59:44 -0400
From: Andrew Brook
Subject: Re: Conceptual vs. Empirical IssuesE

Maybe this will help situate my position. I agree with Glen Sizemore that what is between us is mainly conceptual, not empirical. Another way to put this is that we are not disagreeing about the facts of human information processing and behaviour, we are disagreeing about how these facts should be interpreted, what they entail for theory.



[Original Message]
Date: Mon, 25 Apr 2005 09:44:18 -0500
From: Steven Lehar
Subject: Re: Conceptual vs. Empirical Issues

It is hard to tell whether your "direct representationist" position really is conceptual, or whether it has empirical implications. It is very slippery the way you describe it. But in any case, if it *is* purely conceptual, and makes no predictions, then it is entirely vacuous. It is like saying that "the representation is so good, it really feels as if I am seeing the world directly". Well we all agree with that! But if your theory does not make any predictions, then it is no theory at all! It is a belief, and a rather vague and ill-definable one at that.

Glen Sizemore calls them "assumptions" rather than "beliefs". But the issue is what are they assumptions about? If they are *untestable* assumptions, such as the existence of immaterial souls, then they are not scientific assumptions at all. If they *are* scientific assumptions they *must* make some kind of predictions, even if they are only testable in principle.

In fact there is a very profound empirical issue that is wrapped up in this debate, and that is the question of how one would construct a robot that operates by the same principle as our own brain. Either it needs to be equipped with a representation of the world, or it needs to be designed to extract that information from the world directly. And the debate also has profound implications for neuroscience, that is, should we even be *looking for* representations in the brain, and should we expect those representations to encode *all* of our experience, or only part of it?

It is exactly the profound implications of this debate for artificial intelligence, philosophy, and neuroscience, that make the epistemological question interesting in the first place!



[Original Message]
Date: Mon, 25 Apr 2005 15:56:21 -0400
From: Andrew Brook
Subject: Re: Conceptual vs. Empirical Issues

Steve,

if your theory/interpretation/whatever and mine are both compatible with all the facts, they are *equally* untestable, at least at the bar of the facts. As I said yesterday, I suspect that that is the case. Andrew



[Original Message]
Date: Mon, 25 Apr 2005 16:13:12 -0500
From: Steven Lehar
Subject: Re: Conceptual vs. Empirical Issues

*IF* our theories are indeed both identical with respect to predictions and future discovery of facts, then they are identical, because you would claim, as I do, that every aspect of experience must necessarily be explicitly encoded in the brain, and thus your theory is indistinguishable from representationalism. Welcome to the most reasonable explanation!

But then your rhetoric about seeing things directly, *through* the representation amounts to no more than that the representation is so damned convincing that it really *seems* as if we are viewing the world directly, although you acknowledge that you are seeing it "through" the representation, which is tantamount to saying that you are seeing the representation, while having the vivid impression that you see the world itself, as when watching through a television monitor.

If on the other hand you continue to insist that seeing *through* the representation is something different than observing the state of your brain, while believing that you are seeing the world directly, then you will have to explain what that actually *means* in real down-to-earth terms with sufficient specificity that we would know how to build a robot that sees directly by that same principle, or that we could describe the experiment that would distinguish between the two--for example that there is information encoded in the representationalist view that is not encoded in the direct representatioanlist view. We still do not know what you actually mean by this term.

Doesn't it bother you to defend a view so adamantly and with such conviction, while being unable to specify exactly what you mean by it? It sure has me confused!

But then so do all the other attempts at rationalizing direct perception!

Steve



[Original Message]
Date: Tue, 26 Apr 2005 17:10:23 -0400
From: Andrew Brook
Subject: Re: Conceptual vs. Empirical Issues

Steve,

1. Stop the aggressive bombast. I am not intimidated.

2. I have put a lot of effort into specifying my position in detail, much more than you have put into specifying yours. That you either do not understand what I am saying or do not believe that I could mean is another matter.

Andrew



[Original Message}
Date: Wed, 27 Apr 2005 09:58:33 -0500
From: Steven Lehar
Subject: Re: Conceptual vs. Empirical Issues

Brook >
1. Stop the aggressive bombast. I am not intimidated.
< Brook

I am sorry, the intent was not to intimidate, nor to belittle your contribution. You have done a masterful job of articulating a very slippery issue, and you have my deepest respect for your willingness to explain your thinking very honestly and in excruciating detail. I am really very interested to hear what you have to say.

My point was that I am truly baffled at how you perceive your own position. You freely admit that you cannot describe a robot that operates by direct perception, and that you cannot conceive of an experiment that would differentiate between our two views. And now in the most recent exchange you raise the level of ambiguity of your concept of direct perception one more level by saying that you **don't know** whether your view suggests that the representation in the brain encodes *all* of the information in your experience or not. Given all that, can you understand my puzzlement about exactly what it is you are proposing?

Contrast that with the powerful conviction and certainty with which you argue your position, and it makes me wonder just what it is that you are so powerfully certain *of*? As far as I can tell, you acknowledge *everything* about representationalism, except for the one single statement that you know for a fact that your experience is *direct*, *not* an experience of a representation. Can you not see a contradiction here? The concept of "direct representationalism" is a contradiction in terms, because representationalism is by definition indirect, by way of a representation. In all our long debate, I have never once heard you acknowledge this as any kind of paradox. It need not be fatal; you might still argue that you consider that paradox more tolerable than the incredible notion that the whole world is represented in your head. That would be a reasonable and understandable position. But do you not even see what we representationalists see as a problem with your view?

And how can you maintain your view with such supreme conviction when you cannot describe an experiment that would differentiate our views? Isn't it in the very nature of scientific theories that they make predictions? Shouldn't this at least dampen the magnitude of your conviction, or at least your expectation that others be persuaded to join your faith? As I explained earlier, I can see the profound problems in the representationalist view, but in my view they are outweighed by the deeper paradoxes of the direct perception view. But do you even see where we see a problem in your position? Why do we not hear you at least acknowledge that much?

Your arguments just confirm my initial suspicion that the theory of direct perception is logically indefensible, but that people hold it with great conviction simply because it *SEEMS* so obviously to be true, and the whole theory of direct perception is just an elaborate rationalization of that initial assumption. Can you disabuse me of that suspicion?

Brooks >
2. I have put a lot of effort into specifying my position in detail, much more than you have put into specifying yours.
< Brooks

Really? Are you counting THIS?

A Cartoon Epistemology

and THIS?

Gestalt Isomorphism and the Primacy of Subjective Conscious Experience

and THIS?

The World In Your Head: A Gestalt view of the mechanism of conscious experience

I have invited you to read Chapter 1 of my book (available on-line) but I have not hear your commentary on it. I would be interested to hear your reaction to it. What do you make of the "introspective retrogression"? I would really like to hear what you have to say.

Steve



[Original Message]
Date: Mon, 25 Apr 2005 18:24:46 -0700
From: Glen Sizemore
Subject: Re: Conceptual vs. Empirical Issues

Lehar:
Because the theory of direct perception states that there is no need for internal representations, or in the softer version defended by Andrew Brook, the internal representations do not need to encode ALL of the information of experience, because SOME of that information can be perceived directly *through* the representation (whatever that means).

Sizemore:
Here is where Lehar ignores a point I have made several times. We cannot identify "representation" with some observation of a correspondence between two domains. If we did, then all correspondences would be a matter of representation. That would make all of science representation. ... Whole persons look at representations and they "use" their eyes and their brain. How does a "part of the brain" do this, and how do we justify "our" belief that this could possibly make sense? ... Lehar may say that we don't see the representation ... but the problem is that that is what we do with literal visual representations, we see them. If the things we say about literal representations do not apply to "brain representations," what does that say about "our" cavalier use of language?

Lehar:
The *only* kind of paradigm that remains purely conceptual, in Sidemore's usage, are theories that make no testable predictions whatsoever. ... But such theories are not theories at all, they are *beliefs*, and thus fall outside the realm of science.

Sizemore >
As long as these views remain assumptions, and it is not clear that they can ever be anything else, philosophical debate and conceptual analysis is the only avenue via which they can be compared. Did Lehar not, in fact, argue this himself when he invoked Kuhn a couple of days ago?
< Sizemore



[Original Message]
Date: Wed, 27 Apr 2005 09:05:28 -0500
From: Steven Lehar
Subject: Re: Conceptual vs. Empirical Issues

Sizemore >
Here is where Lehar ignores a point I have made several times. We cannot identify "representation" with some observation of a correspondence between two domains. If we did, then all correspondences would be a matter of representation. That would make all of science representation.
< Sizemore

Whether or not something is a representation depends on whether somebody or something uses it as a representation. A picture in a newspaper is only representational when viewed as a depiction of something else, not when the newspaper is used to line your bird cage. Voltages in a computer memory, or patterns of activation in a retina, are not representations unless or until some process further downstream interprets them as such, at which point they become representations only to that process.

Sizemore >
Whole persons look at representations and they "use" their eyes and their brain. How does a "part of the brain" do this, and how do we justify "our" belief that this could possibly make sense?
< Sizemore

The retina simply responds to light, it does not experience any kind of representation, just its own state. Lower brain functions interpret the signal from the retina as a pattern of light, like the meaningless patterns in an abstract hallucination. The next higher brain functions interpret those patterns of light as meaningless objects in an illuminated scene, like the view of an abstract sculpture. The next higher levels interpret the pattern of objects as a meaningful scene, for example a view of a face. And the highest levels of brain function interpret the face as someone you know, and call up the appropriate response such as nodding or greeting the experienced person. Only the last stage involves the *whole* visual brain, although not necessarily the auditory, olfactory, or limbic functions, which may or may not engage in any particular experience. The lower level functions do not involve the whole brain, and in fact there are an array of visual deficits caused by failures of various regions of the brain (e.g. visual agnosia) that clearly demonstrate that the whole brain does not always have to be involved in every kind of experience.

And likewise in synthetic vision. An image on a photodiode array is just a pattern of voltages, it does not represent anything until the video camera is plugged into a computer that interprets that signal as a pattern of light, or an image. Further algorithms work on that data on the assumption that it is a pattern of light, to detect presumed features in the scene, and then further algorithms work on that feature data to extract presumed objects from that presumed scene. Those are all representations, whether or not there is any correspondence between two domains. If the lens cover is on, then there is no pattern of light, although the rest of the algorithm mistakenly interprets the signal as an image of light nonetheless.

Sizemore >
Lehar may say that we don't see the representation ... but the problem is that that is what we do with literal visual representations, we see them. If the things we say about literal representations do not apply to "brain representations," what does that say about "our" cavalier use of language?
< Sizemore

The word "see" implies a viewer, and that does not apply inside the brain, because the viewer is the whole brain. Instead, we experience the states of our own brain, and when the whole brain directs its attention to an external object (by way of its internal representations) then we call that process "seeing". It is a fallacy to insist that experience necessarily involves eyes and a separate viewer. But we've been over this ground once before already.

Sizemore >
Some questions are empirical questions and some are conceptual. It may be that conceptual questions may become empirical, but the fact of the matter is that the notion that "representations" are necessary to explain perception is simply an assumption.
< Sizemore

True enough, but it *is* a *testable* assumption, at least in principle, and a good way to test it is by demonstration with a simple model system. Representationalism is easy enough to demonstrate. Direct perception is more difficult to demonstrate, because nobody has ever articulated the concept with enough specificity to either build a model, or to make predictions.



[Original Message]
Date: Mon, 25 Apr 2005 10:06:44 -0500
From: Steven Lehar
Subject: Ontology of Experience

Ok, lets try the *ontological* approach to the Direct Perception v.s. Representationalist debate. What is the ontology of experience?

First let us agree what we mean by experience. For example visual experience is the colored three-dimensional volumetric world you see around you when you open your eyes. This is distinct from the objective external world in the fact that when you close your eyes, the visual world disappears, or rather, it is transformed into a foggy brownish space of indefinite extent, while the real world continues to exist unaffected by the blinking of your eyes. Whether you are a direct or indirect perception advocate, we can agree on the definition of visual experience, even though representationalists believe it to be located inside the brain, while direct perceptionists locate experience out in the world beyond the retina.

In either case, visual experience takes the form of modulations of color qualia across a volumetric space. For example a checkerboard pattern is experienced as an alternating modulation of black and white squares. What is the *ontology* of those alternating qualia? What is it that flips from black to white and back again? We know that it is an experience, and that experience is spatially extended in a pictorial fashion, but is there anything in the external physical world that corresponds to that experienced alternation? What is its substance? Does it even have one?

The representationalist answer is that qualia are different states of the physical brain, and thus they are located inside the brain. In other words the brain must posess some continuous spatial medium across which extends a pattern of alternating states. Whether these states correleate with voltages, spiking frequencies, or some standing wave representation, remains an open question at this point. But experience has physical presence in the physical universe known to science.

But what is the direct perceptionist's answer? What is the "stuff" that changes color across space that you experience? And where is it located? And is it in principle detectable by scientific means at that location?

If my experience of a chess board is out there where the chessboard exists, is the black and white pattern I experience the alternation of the pigment in the paint on its surface, perceived directly? Or is it the reflectivity of the surface experienced directly? Or is it the intensity of reflected light experienced directly? Or is it a pattern of activation in my retina? What is its ontology?

I think that direct perceptionists are uncertain about the ontology of experience, they perceive it in a bistable manner, as being both an external objective, and an internal subjective entity, and their answers, when probed, flip back and forth between these two as if a pattern in your brain could somehow be also outside of your head. The whole concept of direct percption is founded on a profound epistemological error, that we can in principle be conscious of things which are not explicitly represented in our brain. Direct perception states the *problem* of experience, it does not offer a *solution* to it that can either guide the construction of a model of the concept, or even an experiment to test the concept. The theory of direct perception is every bit as mysterious as the property of consciousness that it is supposed to explain.

The representationalist position is more coherent because it posits a single ontology, that the modulations of the qualia of visual experience are modulations of the physical state of your brain across some spatial representational medium. And it makes the testable prediction that that medium and its modulations will one day be discovered and decoded in the brain. My vote is for a standing wave representation using a Fourier code to produce moving volumetric holographic images in the brain, and that those images correspond directly to our experience.

Glen Sizemore will complain that there remains the problem of experience, and why it is we have it when our brain is in certain states. But if you accept that mind is a physical process taking place in the physical mechanism of the brain, and you acknowledge that the brain is conscious, then that already is an admission that a physical process taking place in a physical system can under certain conditions be conscious.

Besides, the mystery of experience, or why consciousness exists in the brain, is by no means unique to representationalism. Direct perception cannot resolve that one either. Representationalism at least offers an account of the functional aspects of experience that can be expressed in actual models and make testable predictions. Direct perception does not even offer that much.

Steve



[Original Message]
Date: Tue, 26 Apr 2005 16:45:36 -0500
From: Neil W Rickert
Subject: Re: Ontology of Experience

It's an ingenious argument. It seems that you could use that method to prove that we don't eat food, we eat representations of food. It makes you wonder how we get our nutrients.

You seems to be assuming the kind of Cartesian theater that Dennett criticised. I don't agree with that. But even assuming a Cartesian theater, you are misusing "experience". For example, my experience in watching a movie includes my emotions and thought. It isn't just what was played on the screen. Contrary to your assertion, I don't locate experience out in the world. I don't consider it a thing. It seems to me that treating experience as a thing is a category mistake.

The argument about blinking the eyes is interesting, because I think that actually argues for direct perception. If there is some sort of volumetric representation, then you would think your visual experience would persist during a blink, perhaps slowly fading out. May I suggest that the "foggy brownish space of indefinite extent" is closer to what is represented.

SL >
But what is the direct perceptionist's answer? What is the "stuff" that changes color across space that you experience?
< SL

There is no such stuff. You have confused the issue by your misuse of "experience".

...

I seriously doubt that there is enough DNA in the human genome to encode the hardware specifications that would be required for the proposed system.

In another message ("Conceptual vs. Empirical Issues"), Steven wrote:

SL >
For you direct perception offers the "simplicity" of being intuitively believable, whereas for me representationalism offers the "simplicity" of being causally explanatory.
< SL

There is nothing explanatory about representationalism. Most representationalists admit that they are unable to explain conscious experience. The argument about an infinite regression of homunculuses keeps coming up precisely because representationalism explains nothing.

Whether a system uses direct perception, or is based on representations, in an implementation issue, not an explanatory issue. So lets stop the arguing, and wait for until there is enough empirical evidence to answer questions about implementation details in homo sapiens.



[Original Message]
Date: Wed, 27 Apr 2005 13:27:42 -0500
From: Steven Lehar
Subject: Re: Ontology of Experience

Rickert >
It's an ingenious argument. It seems that you could use that method to prove that we don't eat food, we eat representations of food. It makes you wonder how we get our nutrients.
< Rickert

No, just that we experience internal representations of the external "nouminal" food that nourishes us.

Rickert >
you are misusing "experience". ... I don't consider it a thing. It seems to me that treating experience as a thing is a category mistake.
Rickert

It seems there are a lot of these "category mistakes" in direct perception. I see a spatial structure that is my experience, it is distinct from the world itself, and yet we are not permitted to think of that spatial structure as anything we can talk about. It seems that a lot of direct perception involves prohibitions against certain concepts, as in behaviorism, that fobade discussion of conscious experience. Curiously, all of these forbidden concepts are the very things that reveal the incoherency of direct perception.

Rickert >
The argument about blinking the eyes is interesting, because I think that actually argues for direct perception.
< Rickert

I see. Because closing the eyelids makes the real world out there cease to exist momentarily. Hmmmm...

Rickert >
If there is some sort of volumetric representation, then you would think your visual experience would persist during a blink, perhaps slowly fading out.
< Rickert

No, you would expect it to blink out immediately when the input data stream is blocked, like the image on a photodiode array when you put the lens cover on.

Rickert >
I seriously doubt that there is enough DNA in the human genome to encode the hardware specifications that would be required for the proposed system.
< Rickert

There I believe you put your finger on what I believe is the principal reason why representationalism is generally not given serious consideration.

Rickert >
There is nothing explanatory about representationalism. Most representationalists admit that they are unable to explain conscious experience. The argument about an infinite regression of homunculuses keeps coming up precisely because representationalism explains nothing.
< Rickert

Well it does explain the *functional* aspect of vision, that is, how the information of the world gets in to the computational hardware of the brain. Direct perception does not even explain that much. And direct perception does not explain experience either, it merely prohibits discussion of it.

Rickert >
So lets stop the arguing, and wait for until there is enough empirical evidence to answer questions about implementation details in homo sapiens.
< Rickert

I think that is wise. We are not making any further progress in understanding each other. With this last ontological argument I have expended my last big arrow from my quiver. I will provide a summary overview of all the arguments we have covered in these various threads under the Subject:

Summary: Direct Perception v.s. Representationalism

Steve



[Original Message]
Date: Thu, 28 Apr 2005 08:13:24 -0500
From: Steven Lehar
Subject: Summary: Direct Perception v.s. Representationalism

We have had a full and informative exchange on the question of Direct Perception v.s. Representationalism, and it seems we have pretty much exhausted that topic by now, as we are now just repeating the same arguments over and over again. So I will bow out of the discussion in a major way at this point, although I will answer any residual questions that people might have. I have no more major arguments to make on the subject. Most sincere thanks to all who have contributed, and to all those lurkers out there who have found this debate worth following. If nothing else, it is in my view a most fascinating topic, and resolving this issue once and for all would be of the greatest significance for philosophy, psychology, and neuroscience. I hope we have advanced that cause if only by a small notch.

I will take this opportunity to summarize the whole debate as I see it from the representationalist perspective. The origin of direct perception is naive realism. It is almost impossible to shake the very vivid impression that what we see in experience is the world itself. But this concept is profoundly at odds with the neurophysiology of perception, with sense organs that transmit information to the brain, an obviously representational system. One of these views must be right, and the other is wrong, unless as Andrew Brook claims, the truth lies somewhere in between.

It turns out however that the direct perception view is incoherent, whether in its pure or partial form, because it involves the organism having knowledge of things which are not explicitly represented in its brain. Proponents of direct perception, from Gibson onwards, must sense some kind of a problem at least subconsciously, as seen in their supreme confidence that their view is right, even though they cannot articulate their position with sufficient specificity as to explain how an artificial robotic system could possibly be built that operates by direct perception. Even more surprising, they cannot seem to formulate any possible future experiment that would resolve the issue definitively one way or the other. And strangest of all, they appear to have a peculiar blindness to this gaping hole in their view of perception, as if they simply cannot understand the objections of the representationalists. It would be one thing if they acknowledged that there is some kind of paradox involved, but that they consider that paradox more tolerable than the incredible representationalist view. But direct perceptionists cannot seem to even bring themselves to acknowledge that there is any kind of problem at all.

The more extreme form of direct perception espoused by Gibson is like behaviorism: it is full of prohibited concepts and so-called "category errors" that forbid one to acknowledge experience as a "thing", or to recognize our experience as pictorial, or that it has spatial extent, when those facts are plain for all to see as soon as they just open their eyes. "Seeing is behavior", Glen Sizemore tells us, and there is supposedly no meaning to words like "information" and "representation" and "processing", concepts which have become part of our everyday lexicon since the arrival of computer technology in our lives. One gets the sense that direct perception is a religion rather than a scientific hypothesis, as seen in Gibson's stark refusal to even discuss sensory processing at all.

Andrew Brook's more moderate concept of "direct representationalism" sounds at first more reasonable, because he allows for representations in the brain, but paradoxically, he insists that we view the world "directly" *through* the representations in our brain, a concept that strikes me as a contradiction in terms. If our brain uses representations, then of course our experience consists of those representations, and we cannot view the world directly except by way of them. When questioned more closely, Brook wriggles and squirms until his explanation becomes almost identical to a representationalist thesis, with the sole exception that he continues to insist that our experience of the world is direct, although he cannot even explain exactly what he means by that term.

Until the *principle* of direct perception can be demonstrated in a simple robot model, the concept is so vague and incoherent as to be essentially meaningless.

Conscious experience has an information content, and information cannot exist without a physical medium or carrier to carry that information. That medium can only be the brain, where the sensory nerves terminate, and from whence the motor nerves originate.

A theory that makes no predictions about possible future discoveries in the brain, is no scientific theory at all, but is more like an article of faith.

Direct perception cannot explain the *ontology* of the vivid spatial structure of visual experience, those volumetric objects bounded by continuous colored surfaces that disappear when we close our eyes. These are obviously a product of the brain, and yet they appear out in the world. They can only be states of our own brain, and thus be located in our brain.

Like Behaviorism, direct perception only survives by simply prohibiting discussion of concepts like the information, representation, and processing in the brain, and by prohibiting discussion of the manifest properties of the vivid spatial structure of experience.

The chief argument raised against representationalism is the question of experience, and why we have it when our brain processes sensory information. This is indeed a deep philosophical quandary. However we know for a fact that experience does exist, and that the brain is the organ of conscious experience. The only reasonable location for experience is inside the brain as a representation.

We do not need "internal eyes" to "see" the representations in our brain, we simply experience the structure of certain patterns of energy in our brain directly. While this concept may seem deeply troubling to some, it is nowhere near as troubling as the concept of awareness of the world out beyond the brain, where there is no computational or representational hardware to do the experiencing. Direct perception does not offer any better explanation for the question of experience, except by prohibiting discussion of experience as a "category error".

O'Regans concept of probing the external environment as if it were an internal memory is demonstrably false, because the three-dimensional spatial information of the external world is by no means immediately available from glimpses of the world, but requires the most sophisticated and as-yet undiscovered algorithm to decipher that spatial information from the retinal input.

The absurdity of O'Regan's concept is highlighted by the condition of visual agnosia, a visual integration failure, because the condition of apperceptive agnosia is the failure of a visual function whose existence O'Regan effectively denies.

The question of direct perception v.s. representationalism is not a conceptual issue that is beyond the reach of science, it is a very significant empirical issue with profound implications for the nature of perceptual representation in the brain, and for our attempts to replicate the principle of perception in artificial robot models.

The phenomena of dreams and hallucinations clearly demonstrate that the brain is capable of constructing vivid spatial experiences in the absence of an external world available for direct inspection. Perception is a guided hallucination, constrained by sensory input.

In the face of this overwhelming array of indisputable evidence, how can intelligent, educated people continue to insist that our experience of the world is direct? The answer is the very vivid impression that our experience simply *appears* direct. Some people just cannot bring themselves to accept the view towards which all of the evidence inevitably points. This issue is the ultimate example of a paradigm debate, because seeing things from the representationalist perspective requires that one inverts one's entire epistemology to recognize that the world which appears outside is actually inside one's head. It is admittedly a difficult concept to swallow.

I have prepared an exerpted summary of this whole debate on-line, which can be found at:

http://cns-alumni.bu.edu/~slehar/cartoonepist/EpistDebate.html

Steve Lehar



[Original Message]
Date: Thu, 28 Apr 2005 12:19:14 -0700
From: Augustin Carreno
Subject: Re: Summary: Direct Perception v.s. Representationalism

Steve Lehar has presented a strong defense of representation in his summary. Unfortunately, as he himself has acknowledged previously, it is unlikely to cause too many minds to change. Many people have suggested that maybe there is a way to reconcile the opposing views of direct vs indirect perception, and this seems desirable since the absence of empirical evidence that would clearly determine the correct position suggests that there is a possibility that both approaches could be wrong, or both could be right. I submit that both could be right.

...

In sum, yes there is a picture, but it is not a copy of the world. It is, for all intents and purposes, the world.



[Original Message]
Date: Thu, 5 May 2005 11:32:41 -0500
From: Steven Lehar
Subject: Re: Summary: Direct Perception v.s. Representationalism

Carreno >
there is a possibility that both approaches could be wrong, or both could be right. I submit that both could be right.
< Carreno

The only form of direct perception that might be right is one that is experimentally indistinguishable from representationalism, and thus makes no testable predictions to distinguish between the two. That form of direct perception however is indistinguishable from representationalism, because it posits that every aspect of experience is necessarily replicated in the brain.

The more extreme version of direct perception that prohibits representations of any sort, cannot be right if representationalism is also right.

Carreno >
In sum, yes there is a picture, but it is not a copy of the world. It is, for all intents and purposes, the world.
< Carreno

But there are two aspects of the "picture", one that disappears when you close your eyes, and the other that continues to exist unchanged. That clearly places one world on the outside beyond your eyelids, and the other one inside on this side of your eyelids. The failure to distinguish these fundamentally different worlds is exactly the theory of direct perception.

Carreno >
if we stipulate that the world is indeed already there, then the advantage of making a copy of it over taking it at face value is difficult to elucidate.
< Carreno

It is very easy to elucidate. If "taking it at face value" means making a sensory image of it and sending it to the brain for processing, then that is representationalism, and "direct perception" would be wrong. If "taking it at face value" means behavior and experience *as if* we had a copy of the world in our brain, but we actually *don't*, then the causal loop between perception and behavior is critically broken, and perceptual motor function remains forever a deep dark mystery.

Steve



[Original Message]
Date: Thu, 28 Apr 2005 15:26:48 -0500
From: Neil W Rickert
Subject: Re: Summary: Direct Perception v.s. Representationalism

Steven indeed gave the representationalist perspective. It was perhaps more a polemic than a summary.

SL>Until the *principle* of direct perception can be demonstrated in a
SL>simple robot model, the concept is so vague and incoherent as to be
SL>essentially meaningless.

We should put this in perspective. There is no robot model that demonstrates representational perception. There is no robot model that credibly demonstrates any kind of perception.

-NWR



[Original Message]
Date: Thu, 5 May 2005 12:10:25 -0500
From: Steven Lehar
Subject: Re: Summary: Direct Perception v.s. Representationalism

Reply to Neil Rickert:

Rickert >
Steven indeed gave the representationalist perspective. It was perhaps more a polemic than a summary.
< Rickert

Obviously it was a view from the representationalist perspective; feel free to compose a summary from the "direct perception" perspective. And there was an overall message to that polemic, which is that direct perception may seem plausible enough when debated issue by issue, but it falls apart when viewed in the aggregate, when one takes a "big picture" view as revealed in the summary. Advocates of direct perception cannot even agree on what their theory states, whether there are any representations in the brain or not, or if there are, whether they encode all of experience or only some of it. They cannot make testable predictions of future experiments that would resolve the matter one way or the other, nor can they build a functioning robot to demonstrate the *principle* behind the concept.

The motivation behind direct percepton on the other hand is perfectly clear, it is motivated by the very vivid naive impression that what we experience is the world itself, rather than an internal representation. In that sense it is very much like the "animism" of the turn of the last century whose advocates insisted that life is "something more" than just chemical reactions, although they were unable to define what the added ingredient might be, or how it would be detected in principle, or how it could be implemented in a simple model.

Lehar >>
Until the *principle* of direct perception can be demonstrated in a simple robot model...
<< Lehar

Rickert >
We should put this in perspective. There is no robot model that demonstrates representational perception. There is no robot model that credibly demonstrates any kind of perception.
< Rickert

Any robot model with a camera, computer, and servos, demonstrates the *principle* behind representationalism, and thus clarifies concepts such as "information", "representation", and "processing" in terms that are perfectly clear to anyone who has used a computer. There is *no* such simple demonstration of the *principle* behind direct perception, nor is there even consensus among advocates what that term actually means, or what it says with any specificity about the perceptual process, or how it would be implemented in a robot model. It is quite extraordinary how dogmatically and with such supreme confidence the advocates defend what is a pretty vague concept.

Steve



[Original Message]
Date: Thu, 28 Apr 2005 16:40:50 -0400
From: Andrew Brook
Subject: Re: Summary: Direct Perception v.s. Representationalism

Direct perception is entirely consistent with the neurophysiology perception. It is inconsistent merely with a certain ideology of what follows from these facts. Steve, saying what you do here shows as serious an unwillingness to come to grips with what we are actually saying as your earlier statement that we believe that representation is outside the head. We believe that representation can be of or about thing outside the head. But your statement is about ... the nature of the vehicle, mine is about .... what it is about, what it makes us conscious of. If you refuse to even try to grasp this distinction, I hypothesis that you are letting ideology substitute for evidence and argument.

Lehar >
It turns out however that the direct perception view is incoherent, whether in its pure or partial form, **because it involves the organism having knowledge of things which are not explicitly represented in its brain.**
< Lehar

[The clause in asterisks] No, it does not ... and I have said why over and over and over. What does it take to get you at least to acknowledge that we believe what we say we believe? Go back over my messages. I have addressed this point at least half a dozen times. Again, I suspect you are letting unshakeable ideology go proxy for looking at the arguments.



[Original Message]
Date: Thu, 5 May 2005 12:29:28 -0500
From: Steven Lehar
Subject: Re: Summary: Direct Perception v.s. Representationalism

Brook>
Steve, saying what you do here shows as serious an unwillingness to come to grips with what we are actually saying as your earlier statement that we believe that representation is outside the head. If you refuse to even try to grasp this distinction, I hypothesis that you are letting ideology substitute for evidence and argument.
< Brook

You have yet to tell us whether the representations in the brain necessarily encode *all* of experience or just *some* of it. This may seem like an irrelevant detail to you, but for me that issue is *everything*. Because if you acknowledge that some experience is not explicitly represented, then you actually *ARE* claiming that part of your experience is out in the world, not in your head, and I challenge you to demonstrate *that* in a robot model. So are you herewith now denying that *any* aspect of experience can exist without being explicitly represented in your brain?

Lehar >>
**because it involves the organism having knowledge of things which are not explicitly represented in its brain.**
<< Lehar

Brook >
[The clause in asterisks] No, it does not ... and I have said why over and over and over. What does it take to get you at least to acknowledge that we believe what we say we believe?
< Brook

Is that really what you believe? That all of experience *IS* necessarily represented in the brain? But in that case your "theory" is indistinguishable from straight representationalism. You can't have it both ways, and then complain when I point out the weakness in either one of the two versions which you refuse to choose between.

I say again, how can you feel such supreme confidence that your view is right, when you cannot even explain to us what your view is? Could it be that it is you who are letting ideology substitute for evidence and argument? Steve



[Original Message]
Date: Sat, 7 May 2005 10:53:16 -0400
From: Andrew Brook
Subject: Re: Summary: Direct Perception v.s. Representationalism

Steve once again says that we have not explained our position. So let me once again say what I have done. I have:

1. ... defined the position. Direct perception is being aware of objects in the world, nothing short of them, and not by inference from anything else of which I am aware.

2. ... given examples. If Steve is reading this, the state he is in a a perfect example of direct awareness. He is aware of the words on the screen directly, not by inference from anything else of which he is aware. And he is aware of the words on the screen, not any intermediary. He won't accept this as an example but has given no reason for this refusal.

3. ... said why this position works, indeed is really the only game in town -- no notion of anything more direct can even be articulated.

4. ... said that all this is perfectly compatible with there being a complex information-process connecting brain to objects, with there being representations, and so on. What matters here is that it is crucial to distinguish between the representational vehicle, some state of the brain I expect, and what that vehicle represents, what it makes us conscious *of*, its object.

What more do you want?

If experiments cannot distinguish direct from indirect representationalism, that most emphatically does *not* mean that the indirect view is supported. It means that on this issue both views are unscientific -- not selectable by evidence -- and the issue has to be settled other ways. If A cannot be distinguished from B by evidence, B cannot be distinguished from A either. 'Distinguish' is a symmetrical relationship.

It should be obvious that robot models, as with Bach-y-Rita's prosthetic devices, demonstrate direct and indirect representationalism equally well -- at best. More likely, they tip the scales in the direction of direct representationalism. Steve's reading of the transition in B-y-R patients from experiencing the tickles to seeing objects in the world is this:

Final comment: Steve is entitled to say,

"The motivation behind direct percepton on the other hand is perfectly clear, it is motivated by the very vivid naive impression that what we experience is the world itself, rather than an internal representation. In that sense it is very much like the "animism" of the turn of the last century ...",

only if he accepts the same judgment on his own view. My own view is that civilized debate should avoid speculations about people's motives. It is clear that Steve refuses to consider any position other than his own but I wouldn't dream of speculating about why that is so.

Let me repeat. I have said all these things before. Nothing in this message is new. I am not supremely confident that our position is right but I sure would like it if Steve and others convinced that we are wrong would try to understand the position, rather than attributing to us over and over views that we don't hold, don't need to hold, and would be silly to hold. Our position is as articulated in this message. If you don't like it, say what you find to be wrong about it.

Andrew



[Original Message]
Date: Mon, 9 May 2005 10:27:02 -0500
From: Steven Lehar
Subject: Re: Summary: Direct Perception v.s. Representationalism

Reply to Andrew Brook:

Brook >
What more do you want?
...
Our position is as articulated in this message. If you don't like it, say what you find to be wrong about it.
< Brook

This is the more that I want:

1: Predictions of possible experiments that would differentiate the two positions. If there are no such experiments, then it is no theory at all, and the default is representationalism, because...

2: Define what "direct representationalism" means in terms of an actual mechanism--how would a direct representation differ from an indirect one? We all understand indirect representationalism, that is how robots work. But how would you build the direct representationalist robot? How would it differ?

So far the only "definition" you have given us is "being aware of objects in the world, nothing short of them, and not by inference from anything else of which I am aware.". We understand the *words*, but what does that actually *mean*? What difference would it make to our *experience* if representationalism *was* indirect? What difference would it make for neurophysiology? Your definition is perfectly consistent with pure representationalism, that also gives a vivid impression of perceiving directly.

Brook >
What matters here is that it is crucial to distinguish between the representational vehicle, some state of the brain I expect, and what that vehicle represents, what it makes us conscious *of*, its object.
< Brook

But that is just as true in indirect representationalism. The voltage in a photodiode that represents the brightness of light, or the electrical state of a retinal cone that represents a color component of a point of light. The voltage gives information *of* the brightness of light. But that information is nevertheless necessarily inside the computer, or at least downstream of the photodiode where the transduction occurs. What is the difference between saying 1: the voltage gives indirect information about the brightness of light, and 2: that the voltage gives *direct* information *of* the brightness of light measured *through* the representation. Is there a difference that makes a difference? Or are we just playing with words? The onus is on YOU to demonstrate that direct representationalism is different in any way from indirect representationalism.

Brook >
If experiments cannot distinguish direct from indirect representationalism, that most emphatically does *not* mean that the indirect view is supported. It means that on this issue both views are unscientific -- not selectable by evidence
< Brook

Wrong. Representationalism is the "obvious" concept that is thoroughly understood, to the point that we have built representationalist systems whose principles of operation are clear to all. Not only has direct perception never been demonstrated, it has never even been clearly defined sufficient to make predictions about possible future experiments. If it makes no such predictions, then the default returns to the concept that we *do* thoroughly understand, and that *does* make specific predictions.

Brook >
It should be obvious that robot models, as with Bach-y-Rita's prosthetic devices, demonstrate direct and indirect representationalism equally well
< Brook

Wrong. The Bach-y-Rita device has a human being in the loop to do the perceptual integration, so the fact that they report having a spatial experience outside of themselves, as also with any ordinary visual experience, can mean either that experience is escaping their heads and projecting into the world around them, *or* that a spatial percept is being constructed in their internal representation of surrounding space.

When I'm talking about a robot model I don't mean anything more complex than, for example, a single photodiode and a circuit that interprets its voltage as a brightness of light. That there is already a representational system. How would it differ if it were built as a "direct representationalist" system? Does it not already perceive brightness *through* voltage? If there would be no difference, then the words "direct representationalist" are meaningless.

Brook >
My own view is that civilized debate should avoid speculations about people's motives. It is clear that Steve refuses to consider any position other than his own but I wouldn't dream of speculating about why that is so.
< Brook

I am sorry, I was aware that that might sound a little offensive. But the point I was making is that the two sides of this debate are not symmetrical, one is the "naive" view that we understand from earliest days of childhood, while the other is admittedly incredible, or seemingly so, which is why it remains so controversial. I can understand your position, having been there myself, whereas I don't think you can even fully conceptualize my position, it sounds so self-evidently absurd to you, that it sounds like a theory that "black is white" or "up is down". And that asymmetry in the debate is a significant factor, even if it remains (of course) invisible to you. For example:

Brook >
... said why this [direct representationalism] work, indeed is really the only game in town -- no notion of anything more direct can even be articulated.
< Brook

and

Brook >
If Steve is reading this, the state he is in a a perfect example of direct awareness. He is aware of the words on the screen directly, not by inference from anything else of which he is aware.
< Brook

That does not distinguish the two alternatives, indirect perception also predicts an experience *as if* it were direct. How does Andrew think experience would appear if representationalism were indirect? It would appear just like that.

In summary, we need

1: predictions of possible experiments, and

2: a definition of what the concept means that differentiates it from indirect perception in a way that would have implications for biology or robotics.

Othwise there is no "there" there.

Steve



[Original Message]
Date: Tue, 10 May 2005 12:07:56 -0400
From: Andrew Brook
Subject: Re: Summary: Direct Perception v.s. Representationalism

Lehar:
Even if we found representations for *every last* experience in the brain, that would not disprove direct perception. But then again, what would? If there is no imaginable experiment that would distinguish between direct and indirect perception, then they are indistinguishable.

Me:
As I have said repeatedly, if this is so, then neither theory is empirical on this point. If evidence does not distinguish A from B, then it does not distinguish B from A either, and Steve has no evidence whatsoever that perception is indirect rather than direct. 'Distinguish' is a symmetrical relationship. Since I think that indirect representationalism is at bottom not a theory at all, just an utterly confused misinterpretation of the scientific evidence, that result is fine with me.

Andrew



[Original Message]
Date: Sat, 7 May 2005 13:18:36 -0500
From: Neil W Rickert
Subject: Re: Summary: Direct Perception v.s. Representationalism

SL>Obviously it was a view from the representationalist perspective;
SL>feel free to compose a summary from the "direct perception"
SL>perspective.

I don't see any need for a one-sided summary. I'm involved because of scientific interest, not because I want to push a particular ideology. I welcome a diversity of methodologies.

SL> Advocates of direct perception cannot even agree on what their
SL> theory states, whether there are any representations in the brain
SL> or not, or if there are, whether they encode all of experience or
SL> only some of it.

There is no clear meaning of "representation". Different people take it to mean different things.

SL> They cannot make testable predictions of future experiments that
SL> would resolve the matter one way or the other, nor can they build
SL> a functioning robot to demonstrate the *principle* behind the
SL> concept.

Steven keeps repeating this mantra.

Both sides are in the same predicament with respect to predictions. Both can make vague predictions about what will eventually be discovered in neurophysiology, but neither side can give specifics sufficient to guide the neuro-scientists in such a determination.

No functioning robot has yet persuasively demonstrated any cognitive or conscious capabilities, so both sides are on a par here too.

SL> The motivation behind direct percepton on the other hand is
SL> perfectly clear, it is motivated by the very vivid naive
SL> impression that what we experience is the world itself, rather
SL> than an internal representation.

This is where Steven makes a serious mistake. He is not a mind reader, and he should stop trying to attribute motives to others.

In my case, I am not all that concerned with conscious experience. My main interest has been cognition -- how we acquire and use knowledge. My original view was the received view of representationalism, which is the view most likely to arise out of naive realism. I dropped that view when I realized how implausible it is for a biological system.

I find nothing in representationalism that contributes to a study of cognition. When we examine robots with representationalist designs, the source of knowledge is clear -- it is "knowledge" designed into the system. These systems are built on nativist principles, and cannot account for the acquisition of knowledge such as we see in human children.

SL> Any robot model with a camera, computer, and servos, demonstrates
SL> the *principle* behind representationalism, and thus clarifies
SL> concepts such as "information", "representation", and "processing"
SL> in terms that are perfectly clear to anyone who has used a
SL> computer.

No, it doesn't do any of those things. I don't know of anybody who really believes that a camera, computer and servos has any conscious experience, nor that it demonstrates any cognitive abilities comparable to the learning we see in human children. The terms "information" and "representation" are still quite muddy, with a great deal of disagreement on what they main.

The camera/computer system uses representations in the sense that it represents data for us humans. It does not represent anything for the computer itself, for the computer has no self.

What is represented in the computer is information only in the sense that it informs us humans. The computer is not itself informed. It is merely a mechanical data processor, following preprogrammed procedures on our behalf.

If roboticists are ever able to produce something comparable to Asimov's R. Daneel Olivaw, that might indeed clarify some of these concepts. But I won't hold my breath while waiting.

-NWR



[Original Message]
Date: Sat, 7 May 2005 12:11:15 -0400
From: Andrew Brook
Subject: Re: Summary: Direct Perception v.s. Representationalism

I read Steve Lehar's 'summary' of the debate nearly last of the long series of messages on direct realism, etc., in my inbox. A most annoying document! Without rehashing all the details, let me say that virtually nothing he says about direct realism is true of the position that I have articulated, as I am sure he knows. In fact, in no sense has he summarized the debate. What he does is merely to summarize is all the charges he has levelled against us. I will leave it to others to judge whether I have 'wiggled and squirmed' more than Steve. At least I try to respond to his claims.

Andrew



[Original Message]
Date: Mon, 9 May 2005 12:10:03 -0500
From: Steven Lehar
Subject: Re: Summary: Direct Perception v.s. Representationalism

Reply to Neil Rickert:

Rickert >
There is no clear meaning of "representation". Different people take it to mean different things.
< Rickert

There's a lot more consensus on the meaning of "representation" than there is of "direct perception". Now there's a term that means different things to different people! But most anyone would recognize a photodiode array as a representation that represents patterns of brightness across an image with a pattern of voltages in certain registers. It does not matter whether there is anyone there to "see" or experience the representation, it is still a representation that represents one thing with another.

SL> They cannot make testable predictions of future experiments that
SL> would resolve the matter one way or the other, nor can they build
SL> a functioning robot to demonstrate the *principle* behind the
SL> concept.

Rickert >
Steven keeps repeating this mantra. ...
Both sides are in the same predicament with respect to predictions. Both can make vague predictions about what will eventually be discovered in neurophysiology, but neither side can give specifics sufficient to guide the neuro-scientists in such a determination.
< Rickert

Not so! Representationalism makes the very concrete prediction that every aspect of the world that is experienced, will eventually be identified as various states of the brain. That there is no aspect of experience that is *not* explicitly represented as some state in the brain. That thesis will one day be proven right, or proven wrong. It cannot be both.

Direct perception on the other hand seems to make no testable predictions. Even if we found representations for *every last* experience in the brain, that would not disprove direct perception. But then again, what would? If there is no imaginable experiment that would distinguish between direct and indirect perception, then they are indistinguishable. That is, direct perception would be identical to representationalism.

So we toss a coin to choose between them? No! One way is a reasonable explanation in terms that everyone understands, such as information and representation and computation, that can be built in real hardware, whereas the other is a vague statement of words including a lot of redefinitions of ordinary words to new meanings, and a host of "category errors" that forbid certain conceptualizations, but with *no* specific implications for how to build a robot, or what kind of representations (if any) we should expect to find in the brain.

SL> Any robot model with a camera, computer, and servos, demonstrates
SL> the *principle* behind representationalism, and thus clarifies
SL> concepts such as "information", "representation", and "processing"
SL> in terms that are perfectly clear to anyone who has used a
SL> computer.

Rickert >
No, it doesn't do any of those things. I don't know of anybody who really believes that a camera, computer and servos has any conscious experience,
< Rickert

Redefinition: "representation" now means "has conscious experience"

Rickert>
nor that it demonstrates any cognitive abilities comparable to the learning we see in human children.
< Rickert

Redefinition: "representation" now ALSO means "with cognitive abilities comparable to human children".

Rickert >
The terms "information" and "representation" are still quite muddy, with a great deal of disagreement on what they main.
< Rickert

Only to direct perceptionists. To the rest of the world these terms are perfectly clear and unambiguous. Now if you think "representation" necessarily implies consciousness and cognitive ability comparable to children, then I can see how it would seem to be muddy.

Rickert >
The camera/computer system uses representations in the sense that it represents data for us humans. It does not represent anything for the computer itself, for the computer has no self.
< Rickert

That is one narrow and non-standard re-definition of a commonly understood concept. Most any reasonable person would consider a camera/computer system a representation whether or not there was a human around for it to represent to. Instead of redefining every other word to new and obscure meanings, it would be helpful if direct perceptionists used the ordinary meanings of words, and explain their theory in words we already understand.

The principle of representationalism *is* demonstrated in robot models, whereas the principle of direct perception is not even specified sufficient to devise such a model.

Steve



[Original Message]
Date: Tue, 10 May 2005 15:50:38 -0500
From: Neil W Rickert
Subject: Re: Summary: Direct Perception v.s. Representationalism

SL> There's a lot more consensus on the meaning of "representation"
SL> than there is of "direct perception". Now there's a term that
SL> means different things to different people!

There is a computer science meaning of represent, and there is a cognitive meaning of represent. They are not the same.

It is one thing to say that the output of individual diodes represents the intensity of light striking that diode. It is a quite different thing to say that the array as a whole represents an image.

SL> They cannot make testable predictions of future experiments that
SL> would resolve the matter one way or the other, nor can they build
SL> a functioning robot to demonstrate the *principle* behind the
SL> concept.

Rickert >>
NR>> Steven keeps repeating this mantra. ...
NR>> Both sides are in the same predicament with respect to predictions.
<< Rickert

SL> Not so! Representationalism makes the very concrete prediction
SL> that every aspect of the world that is experienced, will
SL> eventually be identified as various states of the brain.

There is nothing concrete about that prediction. There is no suggested way of testing it. There is no known metric for "every aspect of the world that is experienced". The prediction claims an identity between two things. One of these is knowable in principle, but unknowable in practice, due to the complexity of the brain. The other is unknowable in principle for it refers to what is private and subjective.

SL>> Any robot model with a camera, computer, and servos, demonstrates
SL>> the *principle* behind representationalism, and thus clarifies
SL>> concepts such as "information", "representation", and
SL>> "processing" in terms that are perfectly clear to anyone who has
SL>> used a computer.

Rickert >>
NR>> No, it doesn't do any of those things. I don't know of anybody NR>> who really believes that a camera, computer and servos has any NR>> conscious experience,
<< Rickert

SL> Redefinition: "representation" now means
SL> "has conscious experience"

The redefinition is due to Steven. A few paragraphs up, he makes what he calls a concrete prediction, which equates what he calls representations with experience. But now he apparently wants to deny that representation has anything to do with experience.

I can only repeat my point. Representationalism, at least according to some of the claims of its proponents, is purported to be an explanation of conscious experience. Yet the robots that are said to demonstrate the principle of representationalism do not actually demonstrate any conscious experience at all.

-NWR



[Original Message]
Date: Wed, 11 May 2005 07:09:50 +0100
From: Jeff Dalton
Subject: Re: Summary: Direct Perception v.s. Representationalism

Reply to Neil Rickert:

SL>> Representationalism makes the very concrete prediction that every
SL>> aspect of the world that is experienced, will eventually be
SL>> identified as various states of the brain.

Rickert >
There is nothing concrete about that prediction. There is no suggested way of testing it.
< Rickert

That's unduly pessimistic. For example, I'm aware of the colours of various objects. That's reasonably knowable. Sure, there are skeptical reasons for saying we don't really know it, but skepticism applies much more generally. We needn't make it especially important in this case.

Then, it's unlikely that the brain is so complex that we can never work out whether colours are represented (in the CS sense) in the brain.

Rickert >
I can only repeat my point. Representationalism, at least according to some of the claims of its proponents, is purported to be an explanation of conscious experience. Yet the robots that are said to demonstrate the principle of representationalism do not actually demonstrate any conscious experience at all.
< Rickert

Most representationalist (at least) don't think they're explaining conscious experience. They're explaining the content of conscuious experience (or something like that). How / why various goings-on in the brain are conscious experience is left as a mystery by their representationalism.

Do the advocates of direct perception think that conscious experience is somehow (magic?) able to directly connect with the world, without involving the sorts of mechanisms that would be implemented in a robot?

-- Jeff



[Original Message]
Date: Wed, 11 May 2005 17:22:40 -0500
From: Neil W Rickert
Subject: Re: Summary: Direct Perception v.s. Representationalism

I don't claim to speak for all direct perceptionists. So I will speak only for myself.

I don't expect that there is any magic. On the other hand, I also don't expect that we will build robots capable of having experience, or at least I don't expect that we will build them any time soon.

For sure, I do not deny that there are mechanisms, and I don't expect any new physics to be required.

The trouble with the mechanisms used by robots, as we build them today, is that they lead to a brittleness of behavior. I expect biological systems will use different mechanisms more resistant to this problem.

At least part of the disagreement between me and a representationalist such as Steven, is conceptual. We disagree on how things should be described.

Let me start by stating a couple of positions where I think Steven would disagree.

1: Computers don't actually compute. People compute, and use computers as tools to assist them in that computation.

My reason for this view is that I take computation to be action on numbers (or other mathematical objects), rather than action on numerals (the marks we make to represent numbers). To be sure, we carry out our computation by operating on numerals, which we use as proxies for the numbers. But the term "computation" refers to the action on the numbers, not to the proxy action on the numerals.

2: An ordinary mercury thermometer does not carry a representation of the temperature. Nor is it a symbol that denotes the current temperature.

The thermometer reading is strongly correlated with temperature, and we use that correlation in our determination of the temperature. But we also have to consider whether the thermometer is miscalibrated, or whether there is a local source of heat that makes the thermometer reading unreliable for a particular reading.

This possibility of error in the thermometer is one of the considerations that leads me to the view that the temperature does not represent or denote.

I expect that the representationalist does take the thermometer reading as a representation. He assumes that all we need is to copy that external representation, to form an internal representation.

The direct perceptionist view, or at least my own view, is that we cannot copy. Rather, we must *read* the thermometer. And reading is not just copying, for it involves judgement as to whether conditions (such as a local source of heat) could affect the reliability of the thermometer.

There, I think, is the main difference. The representationalist sees perception as a mainly passive activity, copying representions that exist in the sensors. The direct perceptionist view is that perception is not passive, but involves activity (as in reading the thermometer). The direct perceptionist is emphasizing the interactive aspects of perception.

-NWR



[Not published on PSYCHE-D. Censored?]
Date: Thurs May 12 2005
From: Steven Lehar
Subject: Re: Summary: Direct Perception v.s. Representationalism

Rickert >
Let me start by stating a couple of positions where I think Steven would disagree.
< Rickert

You are absolutely right in that assessment.

Rickert >
1: Computers don't actually compute. People compute, and use computers as tools to assist them in that computation.

1(a): Computers don't make inferences.

2: An ordinary mercury thermometer does not carry a representation of the temperature. Nor is it a symbol that denotes the current temperature.
< Rickert

Very revealing pattern! Do you believe that computers will *ever* be able to compute, make inferences, and represent? Or are these functions in principle beyond the capacity of mere machines?

And if some future computer *will* one day be able to perform these functions, what essential element or vital function would it be in these super advanced computers that would allow them to perform those functions? What is the missing ingredient, or magic component, that would allow them to do what current computers cannot?

Steve



[Original Message]
Date: Thu, 12 May 2005 23:04:59 +1000
From: David Chalmers
Subject: Direct perception and terminological disputes

A couple of observations from a fairly neutral standpoint.

It's become reasonably clear that the debate between Andy Brook and Steve Lehar is terminological. That is, they don't disagree in any relevant respects about what's going on in the brain, in experience, or in the world. They just differ in how they use the terms "direct perception", "representationalism", and maybe "perceive". As far as I can tell, what Andy calls "directly perceiving the world through the use of a representation", and what Steve calls "indirectly perceiving the world through perceiving a representation" are exactly the same thing.

So it's better to get past the terminological issues and concentrate on matters of substance. It may be that there is some other substantive, non-terminological respect in which the two views disagree. But if so, I haven't seen it in this discussion.

Suggestion: to find a substantive thesis on which the sides differ, one must state it without using any of the contested terms above. If one can find a neutrally-stated thesis about which the two sides disagree (for non-terminological reasons), that will be progress. If one can't, so that the two sides turn out to agree on matters of substance, that will be progress too.

--David Chalmers.



[Original Message]
Date: Thu, 12 May 2005 10:25:57 -0500
From: Steven Lehar
Subject: Re: Direct perception and terminological disputes

Reply to David Chalmers:

Chalmers >
It's become reasonably clear that the debate between Andy Brook and Steve Lehar is terminological. ... As far as I can tell, what Andy calls "directly perceiving the world through the use of a representation", and what Steve calls "indirectly perceiving the world through perceiving a representation" are exactly the same thing.
< Chalmers

No, there is something more fundamental at stake here than mere terminology. In an off-list discussion I have discovered that Brook believes that the spatial structure of our experience is the structure of the world itself, whereas representationalism states that the structure of visual experience is a structure in our brain, and only in secondary fashion is that structure also representative of a more remote external world.

It comes down to the question of whether it is valid to make observations about the principles of visual representation by examining the nature of visual experience. According to Brook, visual experience gives us knowledge *of* the external world. According to representationalism the dimensions of conscious experience necessarily map directly to the dimensions of the representational machinery in the brain, and therefore it is valid to make observations on the representational machinery of the brain by observation of experience. For example the first and most obvious observation is that the visual representation is *analogical*, that is, objects and surfaces are represented explicitly by objects and surfaces in a spatial representation.

Chalmers >
Suggestion: to find a substantive thesis on which the sides differ, one must state it without using any of the contested terms above. If one can find a neutrally-stated thesis about which the two sides disagree (for non-terminological reasons), that will be progress. If one can't, so that the two sides turn out to agree on matters of substance, that will be progress too.
< Chalmers

I think the core issue behind this disagreement is whether or not there are analogical or pictorial representations in the brain that explicitly encode every aspect of our experience. And whether or not the principles of the organism interacting with the environment necessarily involves an analog replica of the organism, its environment, and the forces in the world, as suggested here.

http://cns-alumni.bu.edu/~slehar/cartoonepist/cartoonepist44.html

It is clear that it is *that* idea which the direct perceptionists find to be incredible. And it is not just a question of terminology; the visual brain is either analogical/representational, or it is not.

Steve



[Original Message]
Date: Fri, 13 May 2005 09:23:16 -0400
From: Andrew Brook
Subject: Re: Direct perception and terminological disputes

I think I have spotted what leads Steve to keep insisting that I hold that part of the structure of experience is out in the world. He holds as self-evident that if X represents Y, then X has to have the same structure as Y. As he said yesterday, if X represents corners and colours, then X has to have corners and colours. (So Steve was right when he said that this is one issue between us. Not the crucial one but it is an issue.) Then when I insist that representations are of things in the world, he thinks it follows from this that part of the structure of the representation is not only *of* but *in* the world, a patently crazy view and a reductio of any form of direct realism. Well, a reductio it would be -- if we held any such view. But we don't. Nor do most philosophers. As Bill Seager said, the word 'red' does not have to *be* red to *mean* redness.

Certainly information about the world is encoded in some way in the brain, information falling on the foveal spot on the retina at least (and analogously for the other senses), but it would be astonishing if it is encoded in any analogue form. When I see a tangerine, nothing in my brain is tangerine (and if it were, I could not sense that, everything being pitch-black in my brain).

Andrew



[Original Message]
Date: Fri, 13 May 2005 11:34:00 -0500
From: Steven Lehar
Subject: Analogical or Analytical?

Response to Andrew Brook, in "Direct percepton and terminological disputes" thread, I thought it appropriate to re-name the thread, as it seems that we have identified the key difference between our viewpoints, and it is not a mere question of terminology, but a profound paradigmatic issue. A way we view the world. People have difficulty even imagining the world of the alternative paradigm, so they talk at cross-purposes, every word being interpreted in subtly different ways. We've seen enough of that here, but I believe we have broken through that cycle.

Brook >
He holds as self-evident that if X represents Y, then X has to have the same structure as Y. As he said yesterday, if X represents corners and colours, then X has to have corners and colours.
< Brook

This here, I believe, is the center of our disagreement. But I would state it this way: If you accept that perception is indirect (as some of us can't help believing) then it follows that the dimensions of conscious experience must necessarily reflect the dimensions of the information encoded in the brain. I know you direct perceptionists don't believe this, but just imagine for a minute *if* it were actually true, that the world around you was a kind of holographic 3-D image created by resonances in your visual brain. If that were your understanding of your experience, would it not *then* (hypothetically speaking) be self-evident that the representational principle of the brain is analogical?

Our *experience* of the world is analogical. Only a bizarre re-definition of experience would deny to it its spatial aspect. You direct perceptionists believe that the reason that your experience is analogical (spatially extended, volumetric manifold) is because it is an experience *of* an analogical world. We representationalists believe that experience is analogical because the representational mechanism of our brain is analogical, and the truth of that statement can be verified by inspection, as long as one does not get confused about what one is actually "seeing". (Note the ""quote"" marks) Of course we believe the world to be analogical also. But that is why an analogical imaging mechanism is the best representation for spatial interaction with a spatial world.

If the representation were not analogical, how would the "analogicality" of the world get through to our experience of it? Wouldn't it be filtered out by the first non-analogical sensor, just as a photodiode array filters out the polarity and phase of detected light?

And why do direct perceptionists insist on denying the spatial aspect of experience? Why do they have to "define it out of existence"? Because if I talk about the spatial aspect **of your experience** (not of the world) everyone but a direct perceptionist knows exactly what you are talking about. It is that three-dimensional volumetric space full of colored objects, that follows you around wherever you go, filling in a 3-D picture of the world for you wherever you point your eyes. Why do we have to deny its existence? Can direct percepton not survive if our experience *is* spatial? Can't we admit "Yeah, we have a problem with spatial experience, but it is still more credible than representationalism". Wouldn't that be the honest response? Why does one get a sense that direct perceptions are forever redefining words and banning concepts? Is that really necessary?

Brook >
but it would be astonishing if it is encoded in any analogue form. When I see a tangerine, nothing in my brain is tangerine (and if it were, I could not sense that, everything being pitch-black in my brain).
< Brook

Au contraire! It would be astonishing if this rich spatially structured analog experience were represented in the brain any other than by a three- dimensional volumetric imaging system with as much fidelity and resolution as you see in the world around you now. To deny that mechanism is to leave the rich spatial information in our experience in a peculiar kind of limbo, where it is experienced, but does not actually exist as spatial structure anywhere, especially in the case of dreams and hallucinations. But information cannot exist without a physical substrate to store or register that information, and spatial experience requires a spatial representation.

The fact that the world of visual experience appears in the form of a spatial structure is itself conclusive evidence that information in the visual brain is spatially structured.

This is not an esoteric philosophical question, but a very significant concrete question that will one day be proven one way or the other experimentally. I propose we start looking *for* an imaging mechanism in the brain. Its probably right before our eyes, if only we knew what to be looking for in the brain.

Steve



[Original Message]
Date: Fri, 13 May 2005 16:15:31 -0400
From: Andrew Brook
Subject: Re: Analogical or Analytical?

Steve, you don't get it. We *deny* that experience has a structure analogous to the structure of what it represents -- as do philosophers of all realist and irrealist stripes, both connectionist and classical AI researchers, most cognitivists, and at least a great many neuroscientists. Do you really believe that the word 'red' has to *be* red to *represent* redness?

As to your claim that analogical structure is obvious, um, a bit of evidence that we are doing more than representing structure, that our representations have the structure they represent, would be nice. Not that there is -- or perhaps even could be -- any.

Andrew



[Original Message]
Date: Fri, 13 May 2005 21:52:05 -0500
From: Steven Lehar
Subject: Re: Analogical or Analytical?

Brook >
Steve, you don't get it. We *deny* that experience has a structure analogous to the structure of what it represents -- as do philosophers of all realist and irrealist stripes, both connectionist and classical AI researchers, most cognitivists, and at least a great many neuroscientists.
< Brook

And perhaps all those people were mistaken! And perhaps the Gestaltists were right all along!

Brook >
Do you really believe that the word 'red' has to *be* red to *represent* redness?
< Brook

Yes, thats a powerful argument. But if you view it from the other paradigm suddenly it looks very different. For although the word "red" does not have to *be* red to *represent* redness, the COLOR red DOES have to *be* red to *represent* redness. It sounds absurd to you that a portion of your brain should "turn red" when you perceive red things. But when you realize that everything in your experience is just a picture in your brain, suddenly it makes sense again, because certain portions of that picture are "red", and that redness expresses its own content and meaning right there by simply being red.

Everything looks completely different from the epistemologically inverted world! For one thing, many of the most profound paradoxes simply vanish.

Steve



[Original Message]
Date: Sun, 15 May 2005 16:03:48 -0400
From: Andrew Brook
Subject: Re: Analogical or Analytical?

Lehar >
And perhaps all those people were mistaken! And perhaps the Gestaltists were right all along!
< Lehar

Steve, stop being obtuse. The point is, our view is an extremely common one so there is no basis for foisting a view that we *don't* hold on us and presenting it as self-evident truth as you do over and over.

Lehar >
Yes, thats a powerful argument. But if you view it from the other paradigm suddenly it looks very different. For although the word "red" does not have to *be* red to *represent* redness, the COLOR red DOES have to *be* red to *represent* redness.
< Lehar

The colour red does not represent redness, not normally anyway. Something gets to represent something else by being selected by a cognitive system or a system of cognitive systems (a society) to have that function (plus, maybe, by satisfying other conditions). Surface reflectances are seldom thus selected. It is also true, however, that virtually anything *can* be selected for that role, including things that do not resemble what they have been selected to represent in the slightest.



[Original Message]
Date: Mon, 16 May 2005 07:11:13 -0500
From: Steven Lehar
Subject: Re: Analogical or Analytical?

Reply to Andrew Brook:

Brook >
Steve, stop being obtuse. The point is, our view is an extremely common one so there is no basis for foisting a view that we *don't* hold on us and presenting it as self-evident truth as you do over and over.
< Brook

I don't mean to be obtuse, I am merely pointing out that even though your view is indeed an extremely common one, paradigms do change occasionally, and when they do, the extremely common view suddenly becomes the rejected view. My point is that we should decide the issue on the facts and the arguments, not on the basis that it is a common view.

As for "foisting a view that you don't hold", I have done no such thing. I have merely pointed out that your concept of direct representationalism is *either* indistinguishable from regular representationalism, and thus vacuous, *or* it involves having experience of things which are not explicitly represented in your brain, which is impossible in principle.

We have already established that our former misunderstanding on this point is due to the fact that the part of the world that is experienced without being represented is its volumetric spatial extendeness, which you and others redefine to not be experience at all, but an aspect of the world "viewed directly". So my response was that the aspect of the world that you believe to be viewing directly *through* your experience is in fact part of your experience (which is why you experience it) and is not the world itself.

The onus is on you to explain how any aspect of the world, including its analogical spatial extendedness, can possibly penetrate into our experience *without* being represented in our brains. Can you explain that to us?

Steve



[Original Message]
Date: Mon, 16 May 2005 16:54:23 -0400
From: Andrew Brook
Subject: Re: Analogical or Analytical?

Sorry, I should have added this in my response to Steve:

As I have said over and over, being represented and being an object of direct perception are not alternatives to one another. When you insist on saying or implying that for me, they are, you are foisting onto me a view that I do not hold. Of course everything of which we are directly aware is represented. How else could we become aware of these things? We are aware of then *via*, *through* representations. But we are aware of the things, not merely of our representations of them. This too I have said over and over. And argued. And presented evidence for. Steve, you are welcome to reject this view. But you are not welcome to make out that I don't hold it.

On the 'detailed representation' issues, I have said -- in detail and more than once -- what is represented in detail in the brain: in vision, the things in the world that the foveal spot is representing and analogously for the other senses.

Andrew



[Original Message]
Date: Mon, 16 May 2005 18:01:24 +0100
From: Jeff Dalton
Subject: Re: Analogical or Analytical?

I think Andrew Brook is right on one (local) point.

Steven Lehar holds that everything we experience / in experience has to be represented in the brain, and that is close(?) to being a self-evident truth.

But he also holds that the representation is analogical; and that is at least far less clearly true.

The two views should not be confused, but in some messages they seem to be treated as if they must go together. (Some argument, at least, is needed.)



[Original Message]
Date: Mon, 16 May 2005 18:22:37 -0500
From: Steven Lehar
Subject: Re: Analogical or Analytical?

Reply to Jeff Dalton:

Dalton >
Steven Lehar holds that everything in experience has to be represented in the brain, and that is close(?) to being a self-evident truth. But he also holds that the representation is analogical; and that is at least far less clearly true.
< Dalton

Yes, good, lets talk about that one!

First, generally speaking, if the representation in our brain was *not* analogical, for example if it were abstract-symbolic of some form, as spatial data is often represented in computers, then why would our experience of the world not also be an abstract-symbolic experience? How would we even *know* that the world is analogical? Why would our experience of the world not be something like our experience of numbers, or equations, or logic? Those are abstract-symbolic experiences, so we are capable of having them, and so is your experience of my statement when I say, for example, that my couch at home is in front of my TV with a coffee table in between. You *can* form a spatial mental image of the situation, but you don't know the specific shapes and distances and configurations of my living room, so you probably have a very abstract fuzzy image of my statement. But if you were here (as I am) you would have in addition this very vivid spatial experience. Why would that be if not for a spatial representation?

Second: consider the *information content* of a spatial experience. It is a "filled-in" or "reified" data structure, that is, some process has gone to the trouble of painting in every point on every visible surface, and we can even perceive every point in the empty space around those solid objects, and every point on every surface, and in empty space, is experienced simultaneously and in parallel, where they are experienced as continuous surfaces and volumes. The experience is a "manifold" in the mathematical sense, so you simply could not store or register that information without having the full information content of a volumetric scene.

This point is elaborated in detail in this cartoon on the Neurophysiological Implications of Spatial Perception.

http://cns-alumni.bu.edu/~slehar/spatial/spatial.html

(Follow that one through frame by frame to the end)

Steve



[Original Message]
Date: Tue, 17 May 2005 11:44:55 +0100
From: Cathy Reason
Subject: Re: Analogical or Analytical?

Brook >
Steve, you don't get it. We *deny* that experience has a structure analogous to the structure of what it represents -- as do philosophers of all realist and irrealist stripes, both connectionist and classical AI researchers, most cognitivists, and at least a great many neuroscientists. Do you really believe that the word 'red' has to *be* red to *represent* redness?
< Brook

Now, there is clearly something wrong here. It seems that "experience" in this statement has been used to mean the neural or cognitive correlates of experience. What is said here may well be true of representations in the brain (assuming they exist) but it isn't true of subjective experience itself, which does have spattially extended structure, as we can determine by introspection?

Cathy



[Original Message]
Date: Tue, 17 May 2005 12:18:19 -0400
From: Andrew Brook
Subject: Re: Analogical or Analytical?

Reason >
What is said here may well be true of representations in the brain (assuming they exist) but it isn't true of subjective experience itself, which does have spattially extended structure, as we can determine by introspection?
< Reason

No we can't. All we can determine by introspection is that experience *appears* to have spatial, etc., structure. For all we know, that appearance could fail to correspond to how things are. This point has been a commonplace of cognitive research since Kant first made it, though many people have stoutly resisted it, usually by appeal to some principle that experience is different. Here, how things seem is a guarantee of how they are. Not so, as the hundreds of confabulation experiments and a host of other work has shown.

Andrew



[Original Message]
Date: Tue, 17 May 2005 20:55:10 +0100
From: Cathy Reason
Subject: Re: Analogical or Analytical?

Brook >
No we can't. All we can determine by introspection is that experience *appears* to have spatial, etc., structure. For all we know, that appearance could fail to correspond to how things are.
< Brook

Once again, you seem to be confusing subjective experience with objective perception. What you say would be perfectly true if we were talking about introspection on to our brain-states. Objective perception involves correlating an observing state with some observed state, and so for the observing state itself to be observed, we would have to correlate some new state with it. (This is the basis of the little paradox I mentioned in my message to Anthony Sebastian.)

But I'm not talking about introspecting on to our brain-states. I'm talking about introspection on to our subjective states. And our subjective experience of something is, by definition, how that thing appears to us. So there cannot be some feature of our subjective experience which is other than how something appears to us, because they are the same thing. If something appears to us to have spatial extension, then our subjective experience of that thing also has spatial extension.

Cathy



[Original Message]
Date: Wed, 18 May 2005 10:09:27 -0400
From: Andrew Brook
Subject: Re: Analogical or Analytical?

Cathy, repeating a point over and over does not make it any closer to true. You get it exactly right when you say,

Reason >
But I'm not talking about introspecting on to our brain-states. I'm talking about introspection on to our subjective states. And our subjective experience of something is, by definition, how that thing appears to us.
< Reason

As Dennett put it, on how things appear to you, you are the final, ultimate authority. But on the properties that an act of appearing has, you are no authority at all. Who knows what an act of something appearing to you as X is like? It is most unlikely that it has the properties of X. That something appears to you to be a certain way tells you nothing whatsoever about what an appearance of that sort is like. End of story. If I *interpret* a representation as being about Y, does that interpretation automatically have the properties that Y has?

Reason >
So there cannot be some feature of our subjective experience which is other than how something appears to us, because they are the same thing. If something appears to us to have spatial extension, then our subjective experience of that thing also has spatial extension.
< Reason

Not so, no matter how many times you say it. For the reason I just gave.



[Original Message]
Date: Wed, 18 May 2005 17:35:08 +0100
From: Cathy Reason
Subject: Re: Analogical or Analytical?

Brook >
As Dennett put it, on how things appear to you, you are the final, ultimate authority. But on the properties that an act of appearing has, you are no authority at all. Who knows what an act of something appearing to you as X is like? It is most unlikely that it has the properties of X. That something appears to you to be a certain way tells you nothing whatsoever about what an appearance of that sort is like. End of story.
< Brook

Well, Dennett is wrong about so many things that I'm not in the least surprised he's wrong about this as well. ;-)

Andrew, all you are doing is constructing elaborate forms of words which, when deconstructed, amount to claiming that subjective experience has properties which subjective experience does not have. Our subjective experience of something is how it appears to us. They are the same thing. Your entire argument is predicated on trying to find some way of saying that subjective experience of something, and how something appears to us, are different things.

Cathy



[Original Message]
Date: Tue, 17 May 2005 13:04:56 -0400
From: william seager
Subject: Re: Analogical or Analytical?

I would disagree that introspection reveals that experiences have spatial properties. How big is your experience? Ten cubic feet? When I look at the night sky I can see (with naked eye) the Andromeda Galaxy some 2 million light years away - so is the "spatially extended structure" of the experience some 8 million cubic light years in size.

What we can tell by introspection is that we are *seeing* a world that has spatial structure. (To deal with the possibility of hallucination, this can be amended to "introspection reveals that we are having visual experiences as of a world that has spatial structure.) It is not logically valid to infer from this that the experiences themselves have any spatial structure. Of course, if experiences are identical to certain brain processes then they probably have spatial structure of their own in the brain - but introspection does not reveal this. And such neural spatial structure is completely distinct from the spatial structure of the world which visual experience reveals.

William Seager



[Original Message]
Date: Wed, 18 May 2005 04:18:46 -0500
From: Steven Lehar
Subject: Re: Analogical or Analytical?

Seager >
How big is your experience? Ten cubic feet? When I look at the night sky I can see (with naked eye) the Andromeda Galaxy some 2 million light years away - so is the "spatially extended structure" of the experience some 8 million cubic light years in size?
< Seager

Of course not! *Phenomenally* speaking the Andromeda galexy is experienced to be *no farther* than the sun, moon, and stars, all of which appear as if pasted on the dome of the sky, and appear barely farther than the farthest mountains on the horizon. Without scientific knowledge we would have no way of knowing that one celestial object is closer or farther than the others.

Obviously all sizes in experience are relative, although there are a couple of benchmark distances available. One is the perceived size of our own body. If you have a hallucination in which your own body is invisible (I have had such a thing) there is no way to know the size of the things in that experience. (Protagoras: "Man is the measure of all things") The other benchmark is the perceived size of the dome of the sky, that remains fixed no matter how big a space we are perceiving.

But the issue is complicated by the phenomenon of perspective

http://cns-alumni.bu.edu/~slehar/epist/epist6.html

that makes the mapping from phenomenal to objective size nonlinear, and maps an *infinite* objective distance within the finite sphere of experience.

In fact, the phenomenon of perspective is itself one of the most convincing proofs that experience is *not* direct, because it has a prominent curvature and distortion that are clearly not a property of the objective world "viewed directly", but an essential property of the phenomenal world that allows the brain to encode an essentially infinite space in a finite spatial model.

http://cns-alumni.bu.edu/~slehar/cartoonepist/cartoonepist2.html

Steve



[Original Message]
Date: Thu, 19 May 2005 15:50:32 -0400
From: Arnold Trehub
Subject: Re: Direct perception and terminological disputes

Steve, would you be willing to agree that we do not *perceive* the representations in our brain, and accept the claim that these internal representations *constitute* our phenomenal experience of the external world? The television example is a poor analogy because it requires an extra observer (sort of like a homunculus, perhaps?). I think Alex Green would agree. If he doesn't agree, I'm sure he will correct me.

Steve, if you would agree to the suggested rewording, one major point of dispute between you and Andrew might be resolved.

Arnold Trehub



[Original Message]
Date: Fri, 20 May 2005 06:54:03 -0400
From: Steven Lehar
Subject: Re: Direct perception and terminological disputes

Trehub >
Steve, would you be willing to agree that we do not *perceive* the representations in our brain, and accept the claim that these internal representations *constitute* our phenomenal experience of the external world? ...
< Trehub

Yes yes yes, representations *constitute* our phenomenal experience. I sometimes use loose and casual wording such as "perceiving" or "seeing" our representations. But every representationalist knows exactly what I mean whichever way I phrase it, and every direct perceptionist *misunderstands* what I mean whichever way I phrase it.

Trehub >
if you would agree to the suggested rewording, one major point of dispute between you and Andrew might be resolved.
< Trehub

Not a chance! Because it really does *not* come down to a difference in terminology, it is a profound paradigmatic difference in conceptualization that manifests itself in different terminology. But no amount of re-wording will resolve this question that easily!

The foundational difference at the core of the debate is that direct perceptionists really and truly believe that the spatial structure that they (constitute, perceive, experience, see, have aquaintance with, you name it!) actually *IS* the world itself, outside their head, whereas representationalists recognize that that vivid spatial structure *IS* an image of the world in their head.

The reason why the debate goes round and round in futile circles is that this difference is not a conclusion arrived at by logical reasoning, but it is an *initial assumption* that we/they take as self-evidently true, and we/they are willing to warp and morph the rest of understanding to fit in with that "obviously true" initial assumption.

The only reasonable way to address paradigmatic differences of this sort (I said "address", not "resolve", that would be impossible!) is to consider both alternatives as if they could actually be true, and compare the two alternative world views in their totality, to see which one leads to a more coherent picture of perception with the least number of paradoxical anomalies or "incredible" components.

The UN-reasonable way to not-address paradigmatic differences is to make the choice based on the perceived credibility (or otherwise) of the *initial hypotheses* themselves. It is *not* a fruitful argument to begin with "It is *self evident* that perception is (direct/indirect)", because that is exactly the issue in question!

As I see it, the paradoxical anomalies in representationalism are the following:

1: It is truly incredible that the brain, as we know it, is capable of constructing vivid spatial structures in real time of the degree of detail and complexity as the world we see around us.

2: Even if it could, it is truly incredible that those internal pictures should somehow become conscious of their own spatial structure.

To accept those two points would require that we *totally and fundamentally revise* our notions of how the brain represents and processes information, and our notions of what consciousness is at a foundational level.

The paradoxical anomalies in direct perception are the following:

1: It is impossible in principle to become aware of, and thus gain information about, objects outside of our self *except* by means of sensory input and internal representations.

2: The information content of our experience cannot be greater than the information explicitly represented in the brain.

To accept those points would require that we *totally and fundamentally revise* our notions of information, representation, and processing, and (Rickert & Sizemore) accept that those functions can somehow occur outside of our heads, independent of the computational machinery of our brain, or (Andrew Brook) to accept that our experience can somehow be an experience of the world directly, unmediated by the representations of it in our brain.

The paradigmatic partizans on both sides can be heard endlessly repeating again and again ... and AGAIN! the *initial assumptions* of their paradigm, as if the other side has not heard or grasped them yet.

I do however detect an asymmetry in this debate. For my part I fully acknowledge and recognize the profound paradoxes of the representationalist view, and have pointed them out many times, although in the balance I consider them less of a problem than the profound paradoxes of the direct perception view. But I have *never once* heard the direct perceptionists acknowledge the profound paradoxes of their position, they argue as if they were blissfully unaware of *any* kind of problem there, and they repeatedly use familiar terminology in unfamiliar senses as if they can somehow redefine themselves out of their problems. To me that sounds like rationalizations to justify their initial paradigmatic hypothesis, without ever seriously questioning that hypothesis itself. This is why the debate goes uselessly round and round.

I would have a lot more respect for the direct perceptionist view if I heard them acknowledge that both positions contain paradoxical components, but that in their judgment they consider their paradoxical problems less problematic than those of the representationalist view. We would still have to ultimately agree to disagree, but it would put an end to the endless and futile debate, and place focus where it belongs, on the *relative* incredibilities of the two opposing views. Then we could have a rational exchange.

Steve



[Original Message]
Date: Fri, 20 May 2005 17:56:29 +0100
From: Jeff Dalton
Subject: Re: Direct perception and terminological disputes

Quoting Steven Lehar :

Lehar >
The reason why the debate goes round and round in futile circles is that this difference is not a conclusion arrived at by logical reasoning, but it is an *initial assumption* that we/they take as self-evidently true, and we/they are willing to warp and morph the rest of understanding to fit in with that "obviously true" initial assumption.
< Lehar

I usually agree with what Steven Lehar says, but I can't agree with that.

I do agree that representationalism "is not a conclusion arrived at by logical reasoning" -- because there's an empirical component, and because the reasoning isn't strictly deductive or absolutely conclusive.

But it's not an initial assumption either.

Where I started was, if anywhere, with a naive form of direct perception. I moved away from that via an exposure to the relevant science, by considering the arguments for the various philosophical views, and by considering my own conscious experience.

I also suspect that that's a fairly typical path to follow.

-- Jeff



[Original Message]
Date: Fri, 20 May 2005 17:04:57 -0400
From: Steven Lehar
Subject: Re: Direct perception and terminological disputes

I don't mean "initial" as necessarily precedent in time, but rather as "foundational", an assumption upon which all the rest of one's reasoning stands. You are right in temporal sequence, one usually begins with naive realism.

Steve



[Original Message]
Date: Sat, 21 May 2005 08:48:26 +0100
From: Jeff Dalton
Subject: Re: Direct perception and terminological disputes

I don't think it's an assumption at all. The "assumption" phrasing came in when you wrote:

The reason why the debate goes round and round in futile circles is that this difference is not a conclusion arrived at by logical reasoning, but it is an *initial assumption* that we/they take as self-evidently true, and we/they are willing to warp and morph the rest of understanding to fit in with that "obviously true" initial assumption.

Just to be clear: I don't think it's self-evidently true either; nor am I willing to warp or morph anything to fit with it. Those ways of putting it are a bit too strong.

-- Jeff



[Original Message]
Date: Mon, 23 May 2005 11:42:37 -0400
From: Steven Lehar
Subject: Re: Direct perception and terminological disputes

Reply to Jeff Dalton:

Paradigmatic assumptions are assumptions in the sense that they are not arrived at by logical reasoning, but they are simply accepted as self-evident. All science necessarily has foundational assumptions, for example the assumption that there is an external world with independent objective existence. Without that assumption, all science would be useless. But it is not something that can be proven beyond a doubt, it is just an assumption that all scientists share.

Andrew Brook and the other direct perceptionists start from the assumption that we experience the world itself directly, where it lies out in the world. They say so again and again, as if the truth of this were *self-evident*, and thus beyond question. That is a paradigmatic assumption. They are not willing to negotiate on that question. To them, it is inconceivable that it should be otherwise. And that in turn forces them to warp their definition of the meaning of words like information and representation and processing, (and Andrew Brook to deny that visual experience is spatially structured) in order to fit with their initial assumptions.

My own paradigmatic assumption is that mind is a physical process taking place in the physical mechanism of the brain. I am not willing to negotiate on that point, because there is no way to prove that assumption. And that assumption in turn forces me to warp my understanding to the point where I propose that spatial structures in the brain can become conscious of their own spatial structure. It is not a comfortable conclusion, but one to which I fell compelled by the facts, given my initial assumptions.

My point is that there is a difference between such foundational assumptions and the more ordinary conclusions arrived at by logical reasoning, such as that the retina has a certain resolution, or that the optic nerve contains so many fibers. These are issues that people are willing to modify and revise when faced with new data. But people are not willing to revise their paradigmatic assumptions, so it is useful to identify them as such, so as to alert us to the fact that debating them further is futile.

Steve



[Original Message]
Date: Mon, 23 May 2005 14:46:53 -0400
From: Andrew Brook
Subject: Re: Direct perception and terminological disputes

We really are getting nowhere and it is time for me to bow out. But in doing so, let me try one last time to set the record straight.

Lehar >
Andrew Brook and the other direct perceptionists start from the assumption that we experience the world itself directly, where it lies out in the world. They say so again and again, as if the truth of this were *self-evident*, and thus beyond question.
< Lehar

Absolutely not. I have *argued* not assumed that direct perception is the case.

Lehar >
That is a paradigmatic assumption. They are not willing to negotiate on that question. To them, it is inconceivable that it should be otherwise. And that in turn forces them to warp their definition of the meaning of words like information and representation and processing, (and Andrew Brook to deny that visual experience is spatially structured) in order to fit with their initial assumptions.
< Lehar

Most of this is simply wrong. In particular, far from denying that experience is spatially structured, I have insisted that it is -- over and over. What I claim are two things: that the spatial structure of an experience is usually nothing like the spatial structure it represents objects to have, and the spatial structures that it represents are no guide whatsoever to the spatial structure it has.

Lehar >
My own paradigmatic assumption is that mind is a physical process taking place in the physical mechanism of the brain. I am not willing to negotiate on that point, because there is no way to prove that assumption.
< Lehar

As you know, I entirely agree with you about the conclusion, though I of course think that the point can be argued. Materialism is not a matter of faith, it is a well-justified theory of the relationship of mind and brain, one that shows (Cartesian) dualism to be false.

Lehar >
And that assumption in turn forces me to warp my understanding to the point where I propose that spatial structures in the brain can become conscious of their own spatial structure. It is not a comfortable conclusion, but one to which I fell compelled by the facts, given my initial assumptions.
< Lehar

Why does being a materialist about mind force you into any such position? What a curious thing to say!

Andrew



[Original Message]
Date: Thu, 2 Jun 2005 13:42:51 -0400
From: Andrew Brook
Subject: Re: vision science or common sense

[In reply to a posting by Dennis Lomas]

Dennis,

In haste -- more later.

I couldn't be more hostile to sense data theory if I tried. Sense data theory is usually contrasted with direct perception theory. Steve in fact is fairly close to traditional sense data theory.

Andrew



[Original Message]
Date: Fri, 3 Jun 2005 17:13:52 -0400
From: Steven Lehar
Subject: Re: vision science or common sense

Brook>
I couldn't be more hostile to sense data theory if I tried.
< Brook

How can one possibly object to the concept of sense data?

Sense data is not really a theory, nor is it contrasted with direct perception. It is merely a factoring, or separation, of two entities which are easily confused: the object of perception (e.g. a chair), and the sensory *experience* of that object (your perceptual experience of the chair). One might claim, as a direct realist, that those two items are one and the same thing. But that does not deny the distinction between the two. For example you experience only the exposed surfaces of a chair, whereas the chair itself exists throughout its volume, all the way to its hidden rear surfaces. The experience is composed of color and surface, and is limited to a finite resolution, and it changes in different lighting, and ceases to exist in darkness, whereas the chair itself is composed of wood or metal or plastic, and exists to a very much finer resolution to the scale of atoms and molecules, and it does not change in different lighting, nor cease to exist in the dark. Direct perceptionists claim that your experience of a chair is a *subset* of the chair, whereas representationalists claim that it is a distinct entity that is not even spatially superimposed on the real chair that it represents. But whether they are the same or distinct entities, it is always valid to factor things conceptually so that we can discuss their components separately, even if we ultimately conclude that the factoring is illusory, and that the two are one and the same.

What Andrew Brook objects to in sense data theory is that it highlights the *indisputable fact* that our visual experience is spatially structured.

Steve



[Original Message]
Date: Sun, 5 Jun 2005 12:58:20 -0400
From: Andrew Brook
Subject: Re: vision science or common sense

Steve, I reject the idea of sense-data because the representational experiences in my theory are nothing like sense-data. Sense-data are images, something like pictures, at least in traditional sense-data theory (Mill, Russell). My representations present the world itself, not images of any kind (well, some tiny number of representations present images -- when I am looking at a picture, for example). I know you find my view utterly weird, even though it accords with commonsense and is compatible with vision science, but it is what I believe and I have tried to explain and justify it a number of times, indeed show that it is the only coherent point of view. As have others on the list.

Andrew



[Original Message]
Date: Mon, 6 Jun 2005 07:01:37 -0400
From: Steven Lehar
Subject: Re: vision science or common sense

Brook >
Steve, I reject the idea of sense-data because the representational experiences in my theory are nothing like sense-data. ... My representations present the world itself, not images of any kind
< Brook

Yes, I understand that you deny explicit spatial images in the brain, and that your experience is an experience of the world. But even so, the world you experience "directly" manifests itself to you in two separable aspects, the directly perceived modal colored surfaces exposed to your view, and the indirectly perceived amodal solid volumes that you infer from those modal experiences. Whether or not these experiences are in your head or out in the world, the former, directly visible modal surfaces are *conceptually separable* from the invisible solid volumes that they delimit, and "sense data" simply refer to those conceptually separable modal surfaces.

How could one possibly object to making that distinction, whatever one's theoretical conclusions might be?

Steve



[Original Message]
Date: Mon, 6 Jun 2005 10:14:58 -0400
From: Andrew Brook
Subject: Re: vision science or common sense

Of course one can make some kind of experience/world distinction; but to call the experience side in my approach 'sense-data' would just sow conceptual confusion. Experiences in my theory are nothing remotely like sense-data as that term has been understood in the tradition. The first step to doing good science is constructing clear, precise concepts.

Andrew



[Original Message]
Date: Mon, 6 Jun 2005 13:50:41 -0400
From: Steven Lehar
Subject: Re: vision science or common sense

Brook >
Experiences in my theory are nothing remotely like sense-data as that term has been understood in the tradition.
< Brook

Yes, and that is exactly what is lacking in your theory! There is no accounting in it for the existence of sense data anywhere, whereas we know sense-data to exist because we experience them to exist. That is the most direct first-person verification possible in principle! We know this with more certainty than we can possibly know anything else! Its an epistemological fact.

Far from sowing conceptual confusion, the concept of sense-data highlights a specific aspect of experience, that is, its spatially extended nature, and thereby demands an explanation for how our expericence can become spatially extended except by means of an analogical spatially extended representation. Direct perception solves this problem by denial, flat-out denying that most certain of all truths, that my experience is spatially extended!

Steve



[Original Message]
Date: Thu, 2 Jun 2005 15:58:56 -0400
From: Steven Lehar
Subject: Re: vision science or common sense

[In response to a posting by Dennis Lomas]

...

I know for a fact that I have an experience, and I know for a fact that it is spatially structured.

I infer from that experience that I live in a spatially structured world, and it is that world that I am seeing in my experience. But I cannot be certain that that inference is right. It may be that solipsism is true, and that nothing exists except me and my structured experience. More plausibly, it may be that there is in fact an external world, but I am not seeing it directly, but rather I am directly experiencing a spatial replica of that spatial world, which is thus perceived indirectly. Andrew Brook objects vehemently to this very possibility, as if it were so self-evidently false that it does not even have to be disproven. But epistemologically speaking, there is no way for Andrew to prove this to himself or anyone else, it is just an assumption that he is making, and making it dogmatically as if you would have to be crazy not to see this as obviously true.

But in the process, Andrew denies that his *experience* is spatially structured, a denial that *I* consider to be so self-evidently false that it does not even have to be disproven.

Epistemologically I stand on firmer ground insisting that my *experience* is spatially structured, than Andrew with his insistance that perception is obviously and necessarily direct, a fact that is neither self-evident nor proven.

Because whether or not perception is direct, it is in fact undeniable that visual experience is spatially structured, even if we cannot tell whether that experience is a structure in the world or in our brain.

Steve



[Original Message]
Date: Sun, 5 Jun 2005 12:52:50 -0400
From: Andrew Brook
Subject: Re: vision science or common sense

Steve, you are certain that certain events in you *appear* to have these properties but, as I have suggested a number of times, you have no evidence whatsoever that these appearances, sense-data, call them what you will, *have* such properties. That is an article of faith. Introspection tells us only how things appear to be, not how they are.

Andrew



[Original Message]
Date: Mon, 6 Jun 2005 07:13:55 -0400
From: Steven Lehar
Subject: Re: vision science or common sense

When I see a chair, and I experience it as a volumetric structure, it may be an illusion, and the chair is not volumetric at all, there could even be no chair there at all. But I *can* say with absolute certainty that my *experience* is volumetric. It is impossible for my experience to mis-represent itself. An experience is *by definition* how it appears to me, not how it is represented in my brain, or what it represents in the external world.

If the theory of direct perception (or direct representationalism) requires that we deny that our experiences have the properties that we experience them to have, then that is itself a fatal blow to the theory, because it is a blatant violation of epistemology.

Steve



[Original Message]
Date: Mon, 6 Jun 2005 10:11:49 -0400
From: Andrew Brook
Subject: Re: vision science or common sense

Lehar >
An experience is *by definition* how it appears to me, not how it is represented in my brain, or what it represents in the external world.
< Lehar

What makes experience of experience so different. An experience can misrepresent itself, or not represent most or even all of its properties, same as it can misrepresent everything else. Now, you may deny this as an article of faith but you have and could have no evidence that what I say here is false.

Lehar >
If the theory of direct perception (or direct representationalism) requires that we deny that our experiences have the properties that we experience them to have, then that is itself a fatal blow to the theory, because it is a blatant violation of epistemology. < Lehar

Wrong on both counts. Direct perception is silent on the relationship of experience to itself (as is indirect representationalism) and saying that experiences can misrepresent or not represent themselves is not only no violation of epistemology, it is obviously correct if you think about it. My experience of red need not be coloured and my experience of a box need not be cube-shaped. The idea that we have little patches of colour and little three-dimensional shapes in our brain (lit up or dark? -- remember, everything is pitch black in the brain but to be aware of a colour, it has to be lit) makes the mistake of thinking that properties of what is represented have to be properties of the representing, that representations have to have the properties they represent things as having. Sure, they *appear* to have certain properties -- but appearance no more locks in reality here than it does anywhere else.

Andrew



[Original Message]
Date: Mon, 6 Jun 2005 14:04:10 -0400
From: Steven Lehar
Subject: Re: vision science or common sense

Brook >
An experience can misrepresent itself, ... Now, you may deny this as an article of faith but you have and could have no evidence that what I say here is false.
< Brook

It is not a question of evidence, it is a question of definitions! Experience is that which is primary, before analysis sets to work. It interprets nothing, it simply is as it is experienced to be. To claim otherwise is to redefine a familiar term to mean something completely different.

Brook >
saying that experiences can misrepresent ... themselves is not only no violation of epistemology, it is obviously correct if you think about it. My experience of red need not be coloured...
< Brook

The neurons in your brain need not be red. The electrical voltage in your brain need not be red. But the *experience* is what we are talking about when we say the word "red"! The *experience* of red absolutely *MUST* be red, it is the very definition of redness. To deny this would be to say that a red experience can at the same time be actually green. Or an experience of black is actually white! An experience is exactly what it is experienced to be, and no different.

This is not an article of faith, but simple plain logic! To claim otherwise is a contradiction in terms!

Steve



[Original Message]
Date: Mon, 6 Jun 2005 13:41:43 -0400
From: Steven Lehar
Subject: Re: vision science or common sense

Brook >
But experiences need not have the properties that they appear to have
< Brook

They most certainly do, unless you mean to critically redefine the dictionary defined meaning of the word. Experiences *are* appearances, that is why they appear to be as they appear, necessarily and inescapably!

This is not an article of faith on my part, but a simple definition of a word. To say that appearances appear different than they appear to appear is just a flat-out contradiction in terms!

Steve



[Original Message]
Date: Mon, 6 Jun 2005 16:50:15 -0400
From: Andrew Brook
Subject: Re: vision science or common sense

A few comments below and then I really will say no more on this topic.

Lehar >
It is not a question of evidence, it is a question of definitions! ... To claim otherwise is to redefine a familiar term to mean something completely different.
< Lehar

You have a strange theory of definition. Of course an experience is and of course things appear in it. But it need not be what appears in it.

Lehar >
The *experience* of red absolutely *MUST* be red, it is the very definition of redness.
< Lehar

This is such a strange statement that I actually do not understand what you think it says.

From one of your other messages:

Lehar >
we know sense-data to exist because we experience them to exist.
< Lehar

We do not. We know experiences to exist and experience them and they are front and centre in my theory. But sense-data theory is, well, a theory, a theory of what experiences are like -- and can be wrong like any other theory. That is presumably why it is called a *theory*. Or are you redefining the word 'sense-data'?

Andrew



[Original Message]
Date: Tue, 7 Jun 2005 09:25:58 -0400
From: Steven Lehar
Subject: Re: vision science or common sense

Brook >
A few comments below and then I really will say no more on this topic.
< Brook

Likewise for me, we have taken this as far as it will go.

But it has not been a waste of time. Not at all! We have identified the kernel of our disagreement.

The theory of direct percepton begins with the assumption that perception is direct, and that initial assumption is not open to question, it is taken as self-evidently true. Then the rest of our knowledge of perception is warped and contorted beyond all recognition in order to conform to that initial assumption.

This warping includes denying the plainly manifest properties of experience, in particular its spatial extendedness. This is achieved by simply pushing part of experience out of the definition of experience, and asserting that the spatial extendedness of experience is not a property of the experience itself, it is a property of that which the experience is *of*. Even when the experience is hallucinatory! So experience has been split into two entities, that which experience *is*, and that which experience *appears* to be. But this is nothing other than a fundamental re-definition of experience as not-experience, because experience is already that which it *appears* to be, it cannot *be* something other than appearance itself while still being experience. Experiences do indeed have "magical self-certifying powers", they prove to us their own existence, and they appear to us with their apparent properties.

Brook attempts to throw us off the scent by suggesting that this relates to the distinction between the *vehicles* and the *contents* of experience, as if the vehicle (neurons etc. in the brain) is what an experience *is*, whereas the contents are what the vehicles *appear* to be. But this too is a blatant re-definition, because experience is already by definition the *content*, it has nothing whatsoever to do with the *vehicle* of experience.

Brook objects vehemently to the notion of sense-data, the factoring of the experience of an object from the object itself. But it is always justifiable to separate things into separable components, at least conceptually, even if only to later show that the separated concepts are one and the same. But the reason Andrew Brook objects to this perfectly legitimate factoring is that it draws attention to an item which Andrew wishes passionately to ignore, which is the spatial component of experience, something that is clearly distinct from that which it is an experience *of*. In fact, Brook denies the very *existence* of that thing we call the sense-data, because it is an embarassment to the theory of direct perception. Sense data is not a theory, it is a factoring. However it seems that direct perception cannot survive if sense data are allowed to exist.

The choice is ours. Either we accept the apparently incredible fact that the world of experience is a spatial data structure within our own brain, or we are permanently condemned to live in a world where experience is not as it appears to appear, spatial experience is not experience, representations don't represent, indirect perception is direct, the experience of red is not actually red, and experience can somehow bypass the causal chain of sensory processing to give us awareness of spatial extendedness directly out in the world without (necessarily) any spatially extended representations in the brain. Even when that world is hallucinatory!

Its a straight paradigmatic choice: pick the paradox you feel least uncomfortable with.

But it is *NOT* a pseudo-problem. This question will be settled eventually, and I believe it will be settled in our lifetime. Representationalism will be confirmed as soon as we discover the volumetric spatial imaging mechanism in the brain that accounts for our spatial experience.

And it will be discovered all the sooner if we know what it is that we are looking for in the brain!

Steve



[Original Message]
Date: Tue, 7 Jun 2005 10:46:52 -0400
From: Andrew Brook
Subject: Re: vision science or common sense

Steve, this is so obviously a point-misser that I am going to explain the problem one more time:

Lehar >
Experiences ... appear to be as they appear, necessarily and inescapably!
< Lehar

Absolutely! But that does not entail that they *are* as they appear, that they *have* the properties that they appear to have.

Andrew



[Original Message]
Date: Tue, 7 Jun 2005 12:11:04 -0400
From: Andrew Brook
Subject: Re: vision science or common sense

I said I wasn't going to say anything more on the appearance/reality of experience issue but I have to respond to Steve's 'summary' of my position because he gets it wrong on almost every point. Most of the mistakes will be obvious to anyone who has been following the debate but there is a new one. Contrary to what Steve says, I do not align the way our experiences appear to us and what they are really like with the vehicle/content distinction. Both the vehicle and the content merely appear to us and neither appearance need tell us anything about what the respective elements of experience actually are like.

Btw, hallucinations do have two components, the event of hallucinating and what is hallucinated, i.e., the content or object of the hallucination.

Andrew



[Original Message]
Date: Tue, 7 Jun 2005 11:28:44 -0500
From: Neil W Rickert
Subject: Re: vision science or common sense

Lehar >
The theory of direct percepton begins with the assumption that perception is direct, and that initial assumption is not open to question, it is taken as self-evidently true.
< Lehar

Steven keeps repeating this. I cannot speak for Andrew. It is surely false for me.

My starting point was that perception is a deep mystery. It seems impossible. Moreover, representationism does nothing to help. If anything, representationalism makes the problem harder. For it would seem that a representationalist has to solve two different perceptual problems:

1: the problem of perceiving the world, in order to gather the information needed to generate the internal representation

2: the problem of perceiving the internal representation.

Steven has yet to explain how representationalism solves anything.

Hmm, I did have another starting assumption, and that was some sort of empiricism. I took it as implausible that the DNA could carry all of the information and the assumptions that would be required by a nativist/ innatist account of perception.

So how do we perceive the world? On my empiricist assumptions, this has to require learning. We would have to learn the way that the world is. The only way I can see that we could learn how to perceive the world, is if that learning involves interactions with the world. That way, our knowledge of the three dimensional structure of the world can come from those geometric appliances we carry around with us, namely our arms and legs (and, of course, our eyes which also move).

That my perception is *direct* is due to my having learned how to perceive via *direct* interaction with the world. Whether or not there happen to be internal representations, my interaction was with the world, not with those representations. That there may be several layers of neural processing in no way affects the directness of my interaction with the world, and therefore the directness of my perception as it results from that interaction.

Lehar >
Then the rest of our knowledge of perception is warped and contorted beyond all recognition in order to conform to that initial assumption.
< Lehar

I have no idea what warping, if any, Steven is referring to here. In any case, since I don't make the initial assumptions that Steven is talking about, those assumptions I do not make would surely not be able to warp anything.

Lehar >
This warping includes denying the plainly manifest properties of experience, in particular its spatial extendedness.
< Lehar

I certainly don't deny that. I don't recall that Andrew ever denied it.

On second thoughts, after rereading your sentence above, I do deny that spatial extendedness is a property of experience. Rather, my experience is of the spatial extendedness of the world.

I guess you could say that as I travel, I carry my experience with me. And I suppose that could be said to give experience a spatial extendedness. Yet even that is doubtful, for my experience is only at one place at a time, if it can even be said to be at any place at all.

I guess we have quite different meanings for "experience". For me, the word "experience" refers to my interactive relations with the world. For Steven, "experience" seems to refer to that 3-dimension movie now playing in the Cartesian theater that he presumes to be in his head.

Lehar >
Brook objects vehemently to the notion of sense-data, the factoring of the experience of an object from the object itself.
< Lehar

Again, I cannot answer for Andrew.

According to the sense-data theories, we supposedly infer things in the world from sense-data. I cannot fathom what this sense-data is supposed to be. For there is nothing in my experience from which I infer things in the world. It has long seemed to me that sense-data is a philosopher's invention, and not anything based on evidence. Or, to put it differently, "sense-data" appears to be a theoretical term invented to serve the needs of a particular philosophical theory of perception.

-NWR