Abstract
Objects viewed in the visual periphery lack the fine detail available in foveal vision. This is partly due to a reduction in visual resolution and stronger susceptibility to clutter (i.e., ‘crowding’) with increasing eccentricity. However, what exactly is lost and what is retained in peripheral vision remains elusive. For example, in forced choice paradigms, which restrict responses into coarse categories such as letters, target appearance can strongly diverge from the provided categorical alternatives and still yield correct responses. To address this shortcoming, we introduce Geometrically Restricted Image Descriptors (GRIDs) to capture the appearance of shape. In this paradigm, stimuli are constructed solely from segments on a 3x3 grid. We created letters and letter-like shapes (matched in perimetric complexity) to investigate the influence of familiarity on peripheral shape perception. Stimuli were presented at 10° eccentricity in the right visual field, using a gaze-contingent display to prevent foveation. The task was to capture target appearance by connecting points on printed grids placed in front of the participants. Observers viewed the display as long as necessary and were allowed to look back and forth between the screen and the printed grid. GRIDs treat each segment of the presented shapes as distinct targets. Responses were analyzed in terms of character and segment accuracy. Performance depended strongly on familiarity: 94% of the letters and only 53% of the non-letters were depicted correctly. The segment errors in the non-letters, including omissions and truncations, revealed the exact origins of the observed character degradations. We propose that GRIDs are an effective tool to investigate object and shape perception, permitting the capture of fundamental characteristics of visual appearance.
Acknowledgement: This research was supported by the Swiss National Science Foundation (SNF PP00P1_163723 awarded to Bilge Sayim).