Abstract
A complex pattern of empirical results on the role of feedback in perceptual learning has emerged: Whereas most perceptual learning studies employed trial-by-trial feedback, several studies documented significant perceptual learning with block, partial, or even no feedback, and no perceptual learning with false, random, manipulated block, and reversed feedback (Herzog & Fahle, 1997). Shibata et al (2009) showed that arbitrary block-feedback facilitated perceptual learning if it is more positive than the observer's actual performance. At high training accuracies, feedback is not necessary (Liu, Lu & Dosher, 2008), and significant learning was found in low training accuracy trials when they were mixed with high accuracy trials (Petrov, Dosher, & Lu, 2006; Liu, Lu & Dosher, 2009). We conducted a computational analysis of the complex pattern of empirical results on the role of feedback with the Augmented Hebbian Reweighting Model (AHRM; Petrov, Dosher & Lu, 2005), in which learning occurs exclusively through incremental Hebbian modification of the weights between representation units and the decision unit, by simulating existing feedback studies in the literature. The Hebbian learning algorithm incorporates external feedback, when present, simply as another input to the decision unit. Without feedback, the algorithm uses observer's internal response to update the weights. Block feedback was used to modify the weights of the bias unit in the model. The simulation results are both qualitatively and quantitatively consistent with the data reported in the literature. Augmented Hebbian Reweighting accounts for the complex pattern of results on the role of feedback in perceptual learning.