Evolutionary constraints on visual cortex architecture from the dynamics of hallucinations

Abstract

In the cat or primate primary visual cortex (V1), normal vision corresponds to a state where neural excitation patterns are driven by external visual stimuli. A spectacular failure mode of V1 occurs when such patterns are overwhelmed by spontaneously generated spatially self-organized patterns of neural excitation. These are experienced as geometric visual hallucinations. The problem of identifying the mechanisms by which V1 avoids this failure is made acute by recent advances in the statistical mechanics of pattern formation, which suggest that the hallucinatory state should be very robust. Here, we report how incorporating physiologically realistic long-range connections between inhibitory neurons changes the behavior of a model of V1. We find that the sparsity of long-range inhibition in V1 plays a previously unrecognized but key functional role in preserving the normal vision state. Surprisingly, it also contributes to the observed regularity of geometric visual hallucinations. Our results provide an explanation for the observed sparsity of long-range inhibition in V1--this generic architectural feature is an evolutionary adaptation that tunes V1 to the normal vision state. In addition, it has been shown that exactly the same long-range connections play a key role in the development of orientation preference maps. Thus V1's most striking long-range features--patchy excitatory connections and sparse inhibitory connections--are strongly constrained by two requirements: the need for the visual state to be robust and the developmental requirements of the orientational preference map.

Fig. 1.

(A) Simplified Douglas–Martin microcircuit. The blue circle corresponds to inhibitory neurons, and the red circle to excitatory neurons. Similarly, excitatory connections are shown as red arrows from their source and inhibitory connections as blue arrows. Inputs to the microcircuit vary and are not shown. (B) Simplified representation of patchy connections between hypercolumns in V1. Each hypercolumn is represented by a circle, with both E–E and E–I connections between hypercolumns indicated by red arrows.

Fig. 2.

(A) Turing pattern of neural excitation in visual cortex coordinates. (B) The same pattern represented in visual field coordinates (i.e., in the coordinates that a patient undergoing geometric visual hallucinations would see). The logarithmic map between visual field and V1 is responsible for the dramatic logarithmic spiral structure of the hallucination. Although the image shown here is from computation, its qualitative features are very similar to those reported by patients (7). Figure generated with the parameters indicated in the text plus , with all other .

Fig. 3.

Schematic representation of how long-range inhibition leads to instability in strips of visual cortex. An excess of excitatory over inhibitory activity is indicated in red and the converse in blue. A illustrates dynamics with normal visual cortex architecture, where long-range inhibition is forbidden. These dynamics are analogous to normal diffusion. The upper row shows an initial distribution of activity and the lower row shows the evolution of this distribution at a later time. The spatial distribution of activity is smoothed. B illustrates dynamics with added long-range inhibition, whose dynamics are analogous to backward diffusion. The upper row shows the same initial distribution of activity, and the lower row shows the evolution of this distribution at a later time. Under reverse diffusion, the spatial distribution of activity becomes less smooth, leading to short length scale spatial structures.