Representations of shape in modal and amodal completion conditions tested with shape frequency adaptation

Abstract

While we do not see the occluded part in amodal completion, we still have an unambiguous perception that the occluded surface continues behind the occluder. In modal completion, on the other hand, we actually see illusory contours. How does the visual system establish such a perception and distinguish modally completed and amodally completed surfaces? A higher level representation of surface shapes can influence the figure-ground and the border-ownership computations at a lower level. The shape representation at the higher level may have a different effect on modal and amodal completion. Specifically, only in modal completion, the lower level neurons may be activated corresponding to illusory contours. Conventionally, the completion phenomenon is investigated with simple shapes in which smooth interpolation is plausible. However, what if an image suggests an illusory surface with a complex shape? Can the modally completed contours reflect the higher level shape representation instead of a smooth interpolation of the inducing edges? To investigate the representation of illusory surfaces, we applied the shape frequency adaptation paradigm (Gheorghiu & Kingdom, 2007, Vision Research, 47(6), 834–844). They showed that after presenting a sine wave for a sufficient time, the perception of the frequency in a sine wave presented next is affected due to frequency adaptation. If a variation of the Kanizsa square is constructed with sine wave contours, it is possible that the sine wave illusory contours are perceived, which may cause frequency adaptation. We tested this adaptation effect in the Kanizsa square ("modal"), a variation of Kanizsa square where amodal completion is perceived ("amodal"), and occluded surfaces ("occlusion"), all accompanied with sine wave contours. We observed configuration-dependent effects: the modal figure created a stronger effect than the other figures. We report a detailed analysis of the data and discuss their implications for models of modal and amodal completion. Meeting abstract presented at VSS 2013