Abstract
Visual Working Memory (VWM) is a highly limited resource that we use to temporarily store information on items that we encounter. How different items are encoded into VWM remains unclear. Some think individuals can encode multiple items in parallel. An alternative idea is that encoding is serial, with only one item being processed at a time. We used a whole report task in which participants were asked to remember a single or two coloured patches simultaneously. The stimuli were shown for various durations (ranging from 0 to 580ms), and were masked directly after. Participants sequentially indicated the colours they remembered on two continuous colour wheels. We modelled the distributions of errors (as a function of exposure duration) with a mixture model of VWM components, quantifying how much VWM resource was devoted to each stimulus. This demonstrated that the colours' VWM representations improved with exposure duration, and were better for one than for two stimuli. Crucially, we analysed response error as a function of response order. This yielded no difference between error for a single-item response and for the first of the two-item responses. In addition, we modelled the probabilities of individuals remembering none, only one, or both of the simultaneously presented items. Initially, participants were highly likely to remember only one stimulus (at 70ms), but with higher exposure durations (from 200ms) participants were more likely to remember both. Our results are in line with serial encoding into VWM, with one item (or location occupied by an item) being encoded first, followed by the second.
Meeting abstract presented at VSS 2016