Abstract
Aim: Curved features are regarded as critical for representing global shapes but are not the only local features available. Here we employed an adaptation paradigm to investigate whether inflection points are also units of information for representing shape. Method: We measured the perceived amplitude of ‘S’-shaped curves in the upper and lower visual field following a period of adaptation. Observers were adapted to a high amplitude ‘whole-S’ curve in one visual hemi-field, while simultaneously adapting to two temporally interleaved ‘half-S’ curves, formed by splitting an S curve into two at the inflection point, in the other. Results: A bigger shift in perceived amplitude occurred after adapting to a whole-S curve (i.e. containing an inflection) compared to two half-S curves (i.e. with no inflection). Control experiments showed that the advantage for the whole-S curve could not be explained on the basis of local curvature or orientation adaptation. The whole-S curve advantage was however abolished when the test and adapting stimuli were opposite in sign. Conclusions: Our data suggest that there are shape encoding mechanisms that selectively respond to curves containing inflections. However, unlike the encoding of curvature, the coding of an inflection is selective for its sign. These findings have important implication for psychophysical and physiological models of shape processing.
NSERC grant #OGP01217130 given to F.K.