Abstract
To the extent that temporal synchrony plays a role in the binding of local image features into global object representations, one might expect the visual system to be sensitive to synchronized changes, abstracted from the nature of the changing stimulus attribute. In a series of experiments, we investigated whether different “messengers” of temporal structure group into a coherent spatial form. Observers viewed an array of Gabor patches that repeatedly underwent orientation, spatial frequency, phase, or contrast changes (the “messengers”). Elements within a figure region changed at times specified by one stochastic point process, whereas ground elements changed at times specified by a different point process; in some conditions, extracting the shape of the figure region required grouping across synchronized changes within different stimulus attributes. Our results indicate that observers can effectively extract spatial form from temporal structure both when all information resides within a single messenger and when information must be combined across different messengers. Further, mixing messengers of temporal structure is cost-free: Task performance when grouping across different messengers approximates the level achieved when all information resides within one messenger. Finally, our results suggest that different types of synchronous change cannot be distinguished within a figure-ground segregation task when the size of the changes are equated perceptually. These findings cannot be explained by contrast summation generated by low-pass filtering. Thus, the visual system appears to abstract temporal structure from the messenger of the dynamic change; a coherent spatial structure emerges from this abstracted temporal structure. The effectiveness of abstract temporal structure for grouping is consistent with the idea that synchronous neural changes — at least when stimulus-driven — play a role in the binding of local features into global objects.
Supported by NEI EY07760 to RB.