Neural basis of heading discrimination in macaque MSTd   

Hong Xu, Pascal Wallisch and David Bradley

Psychology Dept. University of Chicago   

Heading perception has been studied since the early 1950s. Gibson (1950) noted that for a fixed gaze angle, the focus of expansion (FOE) of the retinal optic flow corresponds to the direction of heading. Humans detect heading differences as small as 0.5 degree. Regarding the neural mechanisms of heading computation, MSTd neurons respond to optic flow (e.g. expansion, rotation) and are sensitive to the FOE position. One study (Heuer and Britten, 2004) compared MSTd responses to judgments about the structure of optic flow patterns (e.g. expansion vs contraction), but no choice/response correlation was found. Another study (DeAngelis et al, 2005) showed correlation between heading judgments and MSTd responses, but it is difficult to judge the magnitude of correlation (choice probability) from this study because a trial-by-trial reference stimulus was not established. We trained a monkey in a two-interval 2AFC heading discrimination task in which each trial contained both a reference and a test stimulus. We found that judgments were linked to neural responses with a mean choice probability of 0.523 (95% confidence interval [0.504, 0.542]). This is significantly greater than chance (0.50, p<0.05, n=121), indicating that MSTd activities influenced the decision process. In comparing neural and observer sensitivity, we found that the monkey performed much better than individual neurons, with an average threshold ratio (neuronal/psychometric) of 165 (95% C.I. [6, 3028]). We conclude that heading perception is diffusely encoded in MSTd.

Keywords: MSTd, heading, population coding, perception, visual motion, electrophysiology