The major difference is that the first version began with a detailed description of the Gestalt principles of Emergence, Reification, Multistability, and Invariance, before presenting the Harmonic Resonance theory, and how it accounts for those perceptual properties.
As a consequence of my exchange with Geissler in the first review, I added a detailed description of the Neuron Doctrine, and the fundamental problems with that paradigm of neurocomputation. In order to keep the paper to managable length, I also reduced the description of the Gestalt properties of perception, since that was already contained in the Bubble World paper, which was under review by BBS at the time.