Abstract
In many perceptual learning studies, detection and discrimination thresholds are assessed using adaptive (e.g. staircase) methods. The measured decrease in threshold is interpreted purely as an improvement in perceptual ability, assuming either an unbiased observer or temporally-invariant levels of bias. Unfortunately, this method does not allow one to empirically determine whether observers truly remain unbiased or change their decisional criteria during learning. To investigate these possibilities, we have carried out contrast discrimination experiments with the method of constant stimuli, allowing us to systematically determine sensitivity (d′) and bias (c), along with an estimation of discrimination threshold (JND). We have previously reported about such experiments, in which we documented reliable liberal shifts in decisional criteria as a function of experience (Copeland & Wenger, VSS 2003). The present work represents an elaboration and refinement. Observers were presented with pairs of Gabor patches that could vary only in contrast. One of the patches was designated as the standard for the entire experiment, while the other could be either (a) a patch at one of 12 levels of different contrast (requiring a ‘different’ response), or (b) a patch at the same level of contrast as the standard (requiring a ‘same’ response). These two classes of stimuli were presented with equal frequency. Observers were required to make same/different responses and were not provided with feedback regarding the accuracy of their responses. Observers were exposed to a total of ∼600 trials for each stimulus contrast over the course of ∼20 hourly sessions. Reliable decreases in threshold, increases in sensitivity, and liberal shifts in bias were obtained for three observers. Thus, it seems that perceptual learning in contrast discrimination reflects much more than simple changes in perceptual ability and we suggest that this may be also the case for other perceptual learning tasks.