Abstract
Why do working memories become less precise the longer they are maintained? Neural models of visual working memory typically assume that simple visual features are memorized through sustained spiking activity in neural populations. The decrease in recall precision observed with increasing number of memorized items can be accounted for within these models by normalization of total spiking activity, resulting in fewer spikes contributing to the representation of each individual item. Longer retention intervals likewise reduce recall precision, but it is unknown what changes in population activity produce this effect. One possible explanation is that spiking activity decays over time, such that the same mechanism (decreased spike rate for each item) accounts for both the effect of set size and retention duration. Alternatively, reduced performance may be caused by drift in the population activity over time due to random noise in recurrent connections, without decrease in overall spiking activity. To distinguish between these hypotheses, we presented subjects with memory displays containing one to four differently-colored discs arranged around a central fixation point. After a variable delay, one of the colors was shown as a cue and subjects had to make a saccade to the memorized location of that item. Based on an integrator model of decision making, we would predict a fixed relationship between response latency and recall precision across conditions if the effects of set size and retention duration are both caused by decreased spiking activity for each item. In contrast, the drift hypothesis predicts no systematic changes in latency with increasing delays. Our results show both an increase in latency with set size, and a decrease in response precision with longer delays within each set size, but no systematic increase in latency for increasing delay durations. We conclude that working memories drift rather than decay with time.
Meeting abstract presented at VSS 2017