Abstract
In order to understand the physiological correlates of form-cue-invariance, we investigated the visual response properties of single neurones in area MT of marmosets anaesthetised with sufentanil (6mg.kg-1.hr-1) and nitrous oxide/oxygen (7:3). The visual stimuli were bars of different lengths, directions of motion and speeds, presented against a background consisting of dynamic visual noise. Specifically, we compared the responses of MT cells to luminance-defined (“solid”) bars, which were uniformly darker than the background, and kinetic (“camouflaged”) bars, which were filled with the same noise pattern as the background, hence becoming visible only due to the coherent motion of its component elements. We found that the vast majority of MT cells (63/ 82, 77%) had cue-invariant responses, whereby the tuning curves for direction of motion were similar irrespective of the type of stimulus (solid or camouflaged). This contrasts with the smaller proportions (approximately 1/3) of cue-invariant units observed in caudal visual areas (V1 and V2). The responses of MT cells to camouflaged bars were typically as strong as those to solid bars of similar length, direction and speed. Moreover, while in V1 and V2 the responses to camouflaged bars were significantly delayed by relative to the responses to solid bars (by 30–40 ms), in MT we found a large population of cells that responded to the two classes of stimulus with similar time-courses. These results are compatible with the notion that cue-invariant responses involving luminance-defined and motion-defined figures emerge in “early” visual cortex as result of feedback interactions from MT, and perhaps other dorsal stream areas (JA Bourne et al., 2002, Cereb Cortex 12: 1132–1145).