Abstract
There is a great deal of debate over the number and quality of items that can be represented in visual working memory (VWM). One theoretical model of VWM predicts that working memory has a discrete capacity limit of three to four items; while other models predict that VWM is a limited pool of resources that can be distributed fluidly. However, the question remains: What is the mechanism determining VWM performance? One such mechanism could be the allocation of attention. To test this prediction participants performed a working memory task for colour information by reporting the colour of a probed sample item on a colour wheel. In Experiment 1, six sample items were presented with probabilistic spatial cues directing attention toward likely target locations; in Experiment 2, four sample items were presented. Probabilistic cues designated the locations of the sample items that were likely to be probed. Responses were examined with a three-component mixture model as well as overall error and compared on the basis of the number of cues and predictive value per cue. The number of cues presented were between one and the maximum number designated by the load; predictive values varied from 100.0% to 33.33%, with uncued items probed between 0% and 22.22% of trials. Results from both studies indicate that the predictive value of cues is a better predictor of working memory performance than the number of cues across all measures for both cued and uncued items. These results suggest that subjects were able to selectively allocate working memory resources to sample items based on the predictive values conveyed by the attention cues. Consequently, consistent with continuous resource models of VWM, attention may by a limiting mechanism in VWM performance.
Meeting abstract presented at VSS 2015