Abstract
Purpose. Spatiotemporal interpolation (STI) refers to perception of complete objects from fragmentary information across gaps in both space and time. Palmer, Kellman & Shipley (2006) found that STI for both illusory and occluded objects produced performance advantages in a discrimination paradigm. Here we tested accuracy of metric properties in representations obtained through STI. Method: The PSE for the length of the base of a moving triangle was obtained using a staircase method for 15 subjects. A white triangle oscillated horizontally for 4 seconds on white background. The triangle was specified in the control condition as three dots located at the vertices and in the STI condition as a white illusory triangle specified by sequential partial occlusions of black background elements in its path. Spatiotemporal (vs. spatial) interpolation was ensured by allowing no more than one vertex of the triangle to be visible at any moment. Two edge lengths (6 and 4.5 degrees), three triangle shapes, and two conditions were tested using interleaved staircases in a within subjects design. Although the dots provided the same spatial and timing information as in the STI condition, they did not induce perception of interpolated contours or a coherent object. If STI does not produce metric representations, we expected similar performance for control and STI conditions. Results . Neither shape nor edge extent affected accuracy of perceived edge length (p [[gt]] .05). STI produced more accurate perception of extent (p [[lt]].001). STI can produce object representations with accurate metric properties. An object constructed through STI does not exist at any moment in the stimulus, yet its size appears to be accurately encoded. Cognitive inferences about length based on position and timing (control condition) were much less accurate, suggesting a special role of perceptual object formation from STI in producing representations of functionally important object properties.