“Competition between active and latent items during working-memory-guided behaviour”
Abstract: Increasing evidence suggests that contents of working memory (WM) are encoded in qualitatively different states depending on their momentary task-relevance. Items that are used for current behavior are thought to be in an “active” state that biases information processing to facilitate decision-making, whereas currently irrelevant items can be held in a “latent” state without affecting on-going cognition. It remains unknown, however, how latent working memories are transformed and consolidated into active decision circuits when behavioural priorities change.
Here, we used time-resolved decoding of Electroencephalography (EEG) data to compare the representation of active and latent items in a task that required flexible WM-based behaviour. 30 participants judged the orientation of target stimuli (gabor patches) relative to changing decision boundaries held in WM. The task consisted of short blocks (16 trials) each starting with the presentation of two items that served as boundaries for the remainder of the block. On each trial, a high or low-pitch tone signalled which boundary should be used for decision-making. This design allowed us to obtain independent neural markers of active and latent WM items, so we could examine their representation as a function of dynamically changing task demands.
We observed that switching between boundaries created a substantial performance cost that rapidly recovered after only a single application of the new boundary. Thus, although adjusting priorities within WM created temporary interference, participants were able to memorise both items while performing the task. Along the same lines, the EEG signal encoded both items immediately after a boundary switch, but was dominated by the active item alone after the first application. These results suggest that shifting behavioural priorities entails transient competition between active and latent WM items, likely reflecting lingering activation of the most recently used contents. Additional analyses of perceptual and decision-related signals will be presented.