Purchase this article with an account.
Nicholas Baker, Philip Kellman; Temporal Properties of Abstract Shape Representation. Journal of Vision 2016;16(12):789. doi: 10.1167/16.12.789.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Background: Shape is an abstract notion, as evidenced when we can see a closed contour figure from dots around a virtual contour, when a cloud appears to resemble a fish, or when we can match shape across transformations of scale and orientation. Abstract shape is not likely represented in initial visual feature registration; it must be computed. We investigated the time course of the emergence of abstract shape representations. Design: Novel shapes defined by black and white dots along the contour were generated and displayed on a gray background for between 30 and 400 msec, followed by a dot mask for 100 msec, which blocked further encoding of the first shape. A second shape was then shown, and subjects performed a forced choice same-different task. "Same" shapes were defined as having the same dot-defined shape outline, whether the same in location, orientation and size, or across a 2D transformation of translation, rotation or scaling. "Different" shapes were constructed by deformation in the outline of the original shape. Results: Subjects were not significantly above chance with 30 msec of presentation time. Performance improved monotonically up to 110 msec of presentation time, after which performance was stable. All transformation types showed similar patterns. Intermediate levels of performance occurred for presentations greater than 30 and less than 110 msec. Control experiments showed that early visual feature registration, such as dot locations, are available during the first 30 msec after stimulus onset. Conclusions: These results suggest that abstract shape representations require about 110 msec to form. This result is not explainable by the time course of early visual feature registration, but appears to involve more time-consuming processes that extract shape.
Meeting abstract presented at VSS 2016
This PDF is available to Subscribers Only