Shape representation by econd-order somorphism, and the horus model:

Proximal mirroring of distal similarities is, at present, the only solution to the Problem of Representation that is both theoretically sound (for reasons discussed in the target article) and practically feasible (as attested by the performance of the Chorus model). Augmenting the latter by a capability for referring selectively to retinotopically defined object fragments should lead to a comprehensive theory of shape processing. 1 An overview of the commentaries The relationships among the stances taken by the commentators on the various issues having to do with representation and similarity can be visualized with the help of Figure 1. This figure depicts a two-dimensional embedding of a textually defined “commentary space” in which each commentary is represented by a point labeled by its author’s initial. The center of the plot is occupied by commentaries that touch upon relatively few of the 11 issues used to define conceptual similarity in this visualization exercise. Whereas the units along the two dimensions are, of course, arbitrary, the locations and the proximities in the plot can be given an interpretation. For example, the upper right corner contains the minders of computational issues, and, in particular, of top-down influences; the lower right is occupied by the champions of nonlinear dynamics, and the lower left is where the proponents of combined metric and structural representations are. All these issues, along with some of the specific concerns raised by the commentators, are discussed next.