Purchase this article with an account.
Seha Kim, Manish Singh, Jacob Feldman; Negative parts and local reversal of figure and ground. Journal of Vision 2012;12(9):310. doi: https://doi.org/10.1167/12.9.310.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Deep indentations in shapes are often perceived as negative parts, i.e., as regions "subtracted" from a base shape (such as a bite taken out of an apple; Hoffman & Richards, 1984). Previous work (Kim & Feldman, 2009) has shown that figure/ground (f/g) assignment can reverse in the vicinity of a negative part. Here we study the conditions that favor such a reversal and provide a mathematical framework for understanding it. We constructed shapes by cutting a narrow strip out of one side of an elongated elliptical base shape. In principle, the resulting contour can be perceived as a single object with a complex shape or as a very simple object (the elongated shape) from which a second shape (the strip) has been subtracted, in which case f/g tends to reverse within the strip so that the indentation is perceived as a figure. We used the motion-probe task, which assesses local border ownership, placing probes both within and outside the indentation. We manipulated the smoothness of the corner between the base shape and the indentation, and found a systematic effect of smoothing: within the indentation, but not outside it, there was a tendency towards local f/g reversal which was maximal when the junction formed a sharp corner and diminished monotonically as the corner was increasingly smoothed. We interpret this smoothing effect within a probabilistic framework in which the two interpretations, one complex shape vs. a simple shape with a negative part (i.e., a combination of a two shapes with a f/g reversal), compete via a principled Bayesian comparison. The model can accommodate a broad range of shape manipulations, in all cases showing how local variations in f/g assignment result from the selection of the most probable (maximum posterior) or, equivalently, simplest (minimum description length) shape interpretation.
Meeting abstract presented at VSS 2012
This PDF is available to Subscribers Only