Abstract
Identifying a single item presented within a sequence of items requires segregating its signal from that of the preceding and/or succeeding task-irrelevant items. In contrast, representing a global aspect of the sequence, in terms of its average orientation for example, requires information pooling across time. Together, these two opposing processes (segregation and pooling) enable us to balance the need to mind individual events while maintaining a general sense of our surroundings. Thus, understanding the relations between the two and how they unfold across time has important theoretical and practical implications for theories of human visual perception. Typically, pooling and segregation were studied separately. Here, we used the same stimuli and orientation estimation procedure to examine both processes and to determine whether and how they vary across different temporal scales. In five experiments, the participants viewed a sequence of three oriented items. Depending on the experiment, they had to either reproduce the orientation of the second item in the sequence (segregation) or report the average orientation of all three items (pooling). The SOAs were either short (≤130ms; short temporal scale) or long (150-475ms; long temporal scale). Mixture modeling analyses of the error distribution revealed a dissociation between pooling and segregation and their temporal scales. With the long temporal scale, the SOA affected encoding precision in both segregation and pooling tasks, but in a very different manner. With the short temporal scale, the SOA also had distinctive effects on the two tasks. While the SOA affected the guessing rate without affecting precision in the segregation task, it substantially affected precision without affecting the guessing rate in the pooling task. These results suggest that temporal segregation and pooling are mediated by different processes and that each of these processes further reflects different mechanisms when considered over a short vs. long temporal scale.