Fig. 18

Concept (A–E) and simulation results (F–H). A Four activities without any clear meaning. b Activities in A are V1 response to oriented lines. C Comparison of brain’s view of spikes (left; apparently intractable) and scientist’s view of spikes (right; decoding possible). D Visuomotor agent set up. E Invariance principle. F Ideal state(s)-action(a) mapping R(s, a) (a), learned R(s, a) (b: synthetic input), learned R(s, a) (c: natural input). G Input (a), initial gaze trajectory (b), and learned gaze trajectory (c). H Learned state-action mapping (a: unordered; b: reordered rows), and learned receptive fields (c: unordered; d: reordered as b) [1]