Decision Processes in Continuous-Outcome Retrieval from Visual Working Memory

Decision-making models of memory retrieval hold that across-trial variability in memory representations and within-trial variability in matching the contents of memory to the retrieval cue both contribute to variability in the speed and accuracy of retrieval decisions. In contrast, most recent studies of visual working memory (VWM) using continuous-outcome decision tasks, in which responses are made on continuous scales, do not distinguish between memory variability and decision-process variability in characterizing retrieval errors and are silent on decision times. We combined a successful model of continuous-outcome decisions, the circular diffusion model (CDM), with a peaked, heavy-tailed distribution of memory representations, the Jones-Pewsey distribution, to characterize the speed and accuracy of continuous-outcome decisions about the hues of saturated and desaturated color patches in four experiments that varied the difficulty of the decision task and the memory load of the VWM task. The CDM provided a good account of the marginal and joint distributions of response times and response errors and the anisotropies of stimuli in different parts of the color space across all experiments. VWM precision followed either an inverse square-root law or power law in set size, depending on saturation, as predicted by sample-size and attention-weighted sample-size models, respectively. The results support a model in which the capacity limits of VWM reflect the statistical properties of a population of noisy neurons that are divided among items in memory under attentional control and determine the precision of recalled items.