Reward history enhances working memory precision in a continuous partial report task

Abstract

Features previously associated with reward capture attention, even when task-irrelevant and physically non-salient. We investigated the effect of such reward-associated features on the precision of working memory representations. We hypothesized that reward-associated features would enhance the precision of WM representations for items presented with that feature. Participants first learned to associate two colors with different magnitudes of reward ($0.01 or $0.05). One week later participants completed a partial report task where they remembered the orientations of three or six masked gratings presented for either 232 ms (Exp.1, n=18) or 1,800 ms (Exp.2, n=11) and then estimated the target orientation on a continuous scale. WM precision for the probed item was measured as the standard deviation of error (degree offset) in the estimation of orientation (SD_error). Importantly, on some trials a task-irrelevant and physically non-salient reward associated feature surrounded either the probed item or a non-probed item. On neutral trials no reward-associated feature was present. The presence of a reward-associated feature affected SD_error only for short exposures (Exp.1), such that SD_error was greater when the high, but not the low, magnitude reward-associated feature surrounded the probed target as compared to neutral trials (Mdiff = 2.74 SD, p = .019) and trials with a low magnitude reward-associated feature (Mdiff = 2.72 SD, p = .002). This effect was only reliable for set size six. There was no significant effect of the reward-associated feature when it surrounded a non-probed item. These results support our hypothesis that irrelevant, and physically non-salient reward-associated features improve WM precision for the stimuli towards which they direct attention, especially when WM load was high, but only when data was limited by short exposure. However, we did not find evidence that the WM precision for a given item was diminished when attention was directed elsewhere by a reward-associated feature.

Meeting abstract presented at VSS 2015