Abstract
Prominent models of perceptual learning (e.g., Gold, Bennett, & Sekuler, 1999; Dosher & Lu, 1999) assume an internal response criterion that is stable across time. In addition, many studies use a two-interval two-alternative forced-choice (2AFC) staircase procedure that is assumed to be immune from the effects of response bias. However, this assumption goes counter to evidence to the contrary (e.g., Green & Swets, 1966; Nisbett & Wilson, 1977). In addition, the limited discussions of bias in this procedure concern themselves with a bias for one of the two presentation intervals, rather than a general response bias. We report a simple signal detection theory model of this task, and show that a general response bias can alter threshold estimates, even when an observer shows no interval bias. We then report an experimental investigation of perceptual learning for contrast detection using a modified version of the two-interval 2AFC staircase that allows us to collect false alarm rates. Each trial on each run of the 3-down 1-up staircase consisted of a first display (75 ms), a noise mask (75 ms), a blank interval (925 ms), the second display (75 ms), and a final noise mask (75 ms). Equal numbers of target-present and target-absent trials were presented; the target-present trials were divided evenly with respect to interval containing contrast. Observers used their index fingers to indicate interval 1, their middle fingers to indicate interval 2, and their ring fingers to indicate neither interval. Feedback was given at the end of each trial. Reliable reductions in threshold were accompanied by reliable increases in false alarm rates. The data are thus consistent with our previous results (VSS 2003, 2004) documenting liberal shifts in response bias in perceptual learning, and indicate that such shifts can be obtained in experimental conditions that have been assumed to be immune from such effects. The results are discussed with respect to decisional influences in perceptual learning.