Abstract
Adapting to moving patterns changes the perceived direction of motion (DoM) of subsequently viewed stimuli. Specifically, the perceived direction of test stimuli is biased away from that of the adapting stimulus. This basic phenomenon can be explained by gain reduction in direction-selective units that are responsive to the adaptor (Levinson and Sekuler 1976). But what happens if the adaptor is perceived as a transparent combination of two patterns with distinct motions? The simplest hypothesis is that each of the perceived directions arises from the activity of a distinct subpopulation of direction-selective neurons. If each of these subpopulations adapts, we would predict the direction of subsequently viewed test stimuli to be repulsed away from both directions. We tested this prediction experimentally by asking subjects to report the direction of a moving test stimulus using the method of adjustment, following a prolonged exposure to a stimulus composed of the superposition of a pair of drifting square-wave gratings (a plaid). We found that adaptation to plaids that are seen to be moving in a single coherent direction could not be explained as the superposition of the adaptation effects of the components. The perceived DoM was shifted away from the coherently perceived direction of the plaid. More surprisingly, we found that even when subjects were adapted to plaids that were perceptually transparent the effect was similar: perceived DoM was shifted away from the unique single direction corresponding to a physical translation of the plaid pattern. We infer from these results that even when an adapting plaid stimulus is perceived as transparent, the system retains a representation of the coherent motion direction, and the subpopulation of neurons underlying this representation form the primary locus of adaptation.