Plato's Cave: Alternative Views
Alternative Views
What are the alternatives to the idea that the world you see around
you is an internal percept?
The Realist View
The most straightforward alternative is to say that the world you see
around you is the real world, and that the head that you know
as your own, is your real head. This is the most familiar
and obvious interpretion, known in philosophy as Realism, i.e. you
believe what you see.
The problem with this view is captured by Plato's parable. If the
room out there is the real room, then it should not be directly
visible to you, but only through the mediation of your senses. This
objection is based on the assumption that the brain is a physical
mechanism that receives input from the senses, and that perception is
a manifestation of the activity of the brain, i.e. that a percept is a
pattern of neural activity within the brain representing external
objects and events. When you look at the room around you, where is
that pattern of neural activity? Is it within the head that you know
as your own? Or is it "out there" in the world around you? If it is
inside your head, then how does the percept get back out of your
head?
Furthermore, if the world around you were the real world, why does it
reflect the distortions seen in visual illusions, which appear not as
an error or distortion in your head, but a distortion in the world
around you?
If a percept is a pattern of neural activity in the brain, and you
perceive the world around you, then that world must be either the
pattern of neural activity itself, or at least the information
represented by that pattern of activity. This precludes the
possibility that the world you see around you is the real world
itself.
Projection Hypothesis
Another possibility is that perception occurs within your brain, but
that the percept gets "projected out" and superimposed on the external
world. If so, what is the mechanism of this projection? How does it
relate to known neurophysiology?
Direct Perception
Gibson
[Gibson 1904]
was addressing the same question posed by this argument: if the
retinal input is so impoverished, how does your percept of the world
become so rich? Gibson however chose the opposite conclusion: the
world appears so rich in detail because your perception of that world
is "direct", i.e. that it occurs directly in the world around you,
rather than inside your head. This argument is certainly appealing on
subjective grounds, because the world does indeed appear to be outside
your head, and contains far more detail than whatever it is that we
experience inside the head that we know as our own. The problem with
this view is that it suggests that perception somehow bypasses the
bottleneck of the senses, and is thus inconsistent with the idea that
perception is a computation performed by the brain on the information
received from the senses.
The World is a Sufficient Model of Itself
A variation of the direct perception idea is a hypothesis supported by
Rodney Brooks
[Brooks 1991]
and Kevin O'Regan
[O'Regan 1992]
(among others), that the world is a sufficient model of itself, and
can be sampled immediately by visual saccades as if the external world
were an internal memory, i.e. a visual saccade is analogous to a
memory access from an internal memory, except that that information is
stored externally as the world itself. The problem with this view is
that the information sampling of the saccade cannot read the
three-dimensional information directly from the world, but only
through the retinal image which is a much impoverished two-dimensional
projection of the external world, and the three-dimensional
interpretation must be reconstructed from that two-dimensional
projection. This reconstruction operation is not a trivial matter, as
it depends on a good deal of contextual information.
The idea that the external world can be accessed as if it were an
internal memory only seems intuitively plausible
because of the illusion of Plato's Cave, i.e. when we
think we are looking at the external world, we really
are looking at the internal copy of that world,
which is indeed accessing an internal rather than an external memory.
If however we could see how the raw input appeared before the
processing performed by the visual system, this idea would be patently
absurd. Indeed, patients with the condition of visual
agnosia do see the world without pre-processing, and cannot make
any sense out of that visual input. Visual agnosia is not a problem
in the accessing of the external information, for the retinal input
comes in just fine. Rather it is a problem of assembling and
interpreting that information as a three-dimensional percept which is
deficient.
Abstraction
The most common view represented in computational models of vision is
that perception is a process of abstraction from the visual input,
i.e. that the visual system extracts features such as edges and
surfaces from the retinal image, and passes them on to higher level
processing stages which extract ever more abstract and high level
features from the lower level representations, resulting in an
abstract compressed representation of the external world in some
internal symbolic code. The problem with this view is that it
completely ignores the most salient manifestation of perception, i.e.
the very solid and real-looking world that appears to surround you.
This world is clearly not an abstraction, for unlike the retinal image
it is fully spatial, and rendered at the highest possible resolution.
Indeed it contains far more information than the retinal input upon
which it is based. Where in the hierarchy of abstraction is this
percept represented?
Any Other Ideas?
If I have missed any alternatives which resolve this dilemma, please
send me email
and let me know, and I will include it here.
Return to argument
Return to Steve Lehar