Abstract
The role of attention in perceptual learning (PL) has been a topic of controversy. A number of studies have reported that a task-irrelevant feature is not learned (Ahissar & Hochstein 1993 PNAS; Shiu & Pashler, 1992 P&P). It was concluded that focused attention to a feature is necessary for the feature to be learned. In contrast, another line of studies have shown that PL occurs even on task-irrelevant features that are subthreshold, and, therefore, concluded that attention on a feature is not required to learn that feature (Watanabe et al., 2001 Nature; Seitz and Watanabe 2003 Nature; Dinse et al, 2003 Science). Here we attempt to reconcile these divergent results. Subjects performed 10-days of training with a foveal letter identification task while a dynamic random-dot display (DRD) was presented in the background. Previously we have reported directional specific learning for an exposed DRD direction that was below thresholds. Here, we systematically explore the relation between motion coherence during training and the resultant magnitude of PL. We used a within-subject design in which each subject was exposed to two different directions of motion, each at a different, but consistent, coherence level. Before and after the exposure, subjects were tested on motion-direction discrimination for six directions including the two exposed directions. We find performance improvements only for the direction trained at a low, perithreshold, coherence level. The present results are in accord with the results of the aforementioned different lines of studies that appeared to contradict to each other. At the same time, the present results more readily support the hypothesis that only irrelevant feature signals that are suprathreshold are “noticed”, inhibited (Tsushima et a., 2006, Science) and fail to be learned than the hypothesis that only an attended feature is learned.
Funding for this research is provided by NSF (BCS-0345746, BCS-0549036, BCS-PR04-137 Center of Excellence for Learning in Education, Science, and Technology), NIH (R01 EY015980 and R21 EY017737), and the Human Frontier Science Program Organization (RGP18/2004) to T.W.