Abstract
Stimuli previously associated with reward have been shown to slow responses when presented as irrelevant distractors in subsequent, unrewarded tasks (value driven attentional capture, or VDAC). During reward training, participants search for a target (color-defined circle) and discriminate the orientation of a line contained inside. Only correct responses made before a deadline are rewarded. To maximize earnings, participants must consider multiple reward contingencies: responding too quickly increases error rates; missing the response deadline reduces reward probability to zero, whereas responding at chance still yields reward half the time. Does facing these reward contingencies during training alter the way in which participants balance speed and accuracy when reward is removed and the response deadline lengthened (i.e., the VDAC test phase)? Using the VDAC procedure (Anderson et al. 2011, PNAS, Experiment 3, 800ms/1200ms response deadlines in training/test), we randomly assigned 46 participants to receive either performance-contingent reward during training or correct/incorrect feedback only during training and a flat reward. We focus here on distractor-absent test trials to eliminate any value driven attentional confounds. Using maximum likelihood estimation, individual reaction time (RT) distributions were fit with a three parameter, exponentially modified Gaussian function: mu and sigma describe the mean and standard deviation of the Gaussian portion; tau describes the exponential portion, which gives the distribution its characteristic positive skew. Despite the absence of previously rewarded stimuli, participants who received performance-contingent reward during training showed significantly wider RT distributions during test (greater sigma values), relative to the control group. Sigma parameter estimates were also positively correlated with error rates, reflecting a behavioral cost to this increased variability. These data provide preliminary evidence that performance-contingent reward training increases reaction time variability when reward is no longer at stake, even in the absence of previously rewarded stimuli.
Meeting abstract presented at VSS 2017