Abstract
The role of feedback is an issue of much uncertainty in research of perceptual learning. While it is commonly acknowledged that feedback can aid in perceptual learning (Herzog and Fahle, 1997), there are many examples in which it is not required. For instance, learning without feedback can occur for stimuli that are irrelevant to the subject's task (Seitz and Watanabe 2003). This result together with others suggest that internal reward signals can serve the same role as external reward signals (Herzog & Fahle, 1998; Seitz & Watanabe, 2005). Does this indicate that external feedback is not necessary in any perceptual learning? Here we report results from two different studies in which feedback turned out to be necessary for learning.
In the first study subjects were trained for 10 days to report the orientation of bars (of two orthogonal orientations) that were masked by spatial noise. Trials of 8 different SN levels (yielding a psychometric function that ranged from chance to ceiling) were randomly interleaved.
In the second study subjects were trained for 1 day to report the direction of motion (of four oblique directions) of a 10 degree patch of 100% coherent, but low-contrast dynamically moving dots. Trials of 10 different contrast levels (yielding a psychometric function that ranged from chance to ceiling) were randomly interleaved.
In both studies, subjects who received no feedback failed to show any learning, whereas subjects who were given trial-by-trial feedback showed significant learning across all SN or contrast levels.
This research was supported by NSF grant (BCS-9905914), NIH grant (R01EY015980-01), and Human Frontier Research grant (RGP18/2004) and NSF (CELEST).