CORRESPONDENCE NOISE AND SIGNAL POOLING IN THE DETECTION OF COHERENT VISUAL-MOTION

In the random dot kinematograms used to analyze the detection of coher ent motion in the middle temporal visual area (MT) and in psychophysic al experiments the exact way that dots are paired between successive p resentations is not known by the observer We show how to calculate the limit to coherence threshold caused by this uncertainty, which we cal l ''correspondence noise.'' We compare ideal thresholds limited only b y this noise with those of human observers when dot density, ratio of dot numbers in two fields, area of stimulus, number of fields, and met hod of generation of the coherent dots are varied. The observed thresh olds vary in the same way as the ideal thresholds over wide ranges, bu t they are much higher. We think this difference is because the ideal detector takes advantage of the high precision with which dots are pla ced in the kinematograms, whereas the neural motion system can only op erate with low precision. When kinematograms are generated with decrea sed precision of dot placement, the ideal detector no longer has this advantage, and the gap between ideal and actual performance is greatly reduced. Because the signals that result from objects moving in the r eal world are scattered over broad ranges of direction and velocity, h igh precision is not needed, and it is advantageous for the motion sys tem to pool information over broad ranges. Other mismatches between ki nematograms and the neural motion system, and internal noise, may also elevate human thresholds relative to the ideal detector. The importan ce of external noise suggests that the neurons of MT form a vast array of optimal filters, each matched to a different combination of parame ters in the multidimensional space required to define motion in patche s of the visual field.