Higher-Order Motion Inputs For Visual Figure Tracking: Control Algorithms and Neural Circuits

Visual figures are detectable based on a range of spatiotemporal characteristics that differ from surrounding background. A figure that corresponds to an ordinary moving object generates coherent two-point space-time correlations related to the first moment of the luminance distribution. Such signals are readily detectable by the standard implementation of the Hassenstein-Reichardt elementary motion detector (EMD), and as such are referred to here as EM. Prepared in collaboration with Uppsala University, Sweden and University of California, Howard Hughes Medical Institute, Los Angeles. The original document contains color images.