Rosanne Rademaker, UC San Diego
Mechanisms of robust working memory representations
Many everyday tasks require snippets of information to be momentarily kept in mind. In this talk I will explore some of the mechanisms by which the brain achieves robust representations of such information in working memory. First, systematic biases provide a unique window into the underpinnings of working memory. When people recall a visual orientation after a short delay, two dissociable biases emerge: One depends on the wiring of our sensory system, determined during visual development, while the other is tied to our physical environment. Second, presenting irrelevant visual inputs during a working memory task reveals region-wide multiplexing abilities in classic sensory areas, with population-level response patterns in early visual cortex representing the contents of working memory concurrently with passively viewed sensory inputs. Third, having people pay attention to sensory inputs during working memory tests the limits of this simultaneity. In visual cortex, the fidelity of actively viewed representations increases as the fidelity of memory representations decreases. Thus, trade-offs between memory and sensory representations are due to changes in attentional priority as opposed to just the presence or absence of concurrent input. Collectively, as behavioral requirements change, memory representations may change along with it. Ultimately, the working memory system achieves remarkable stability, and recall is only moderately affected by systematic biases and passive or active interference.