Proceedings of the Annual Meeting of the Cognitive Science Society

A striking limitation of human cognition is our inability to ex-ecute some tasks simultaneously. Recent work suggests thatsuch limitations can arise from a fundamental trade-off in net-work architectures that is driven by the sharing of representa-tions between tasks: sharing promotes quicker learning, at theexpense of interference while multitasking. From this perspec-tive, multitasking failures might reflect a preference for learn-ing efficiency over parallel processing capability. We explorethis hypothesis by formulating an ideal Bayesian agent thatmaximizes expected reward by learning either shared or sep-arate representations for a task set. We investigate the agent’sbehavior and show that over a large space of parameters theagent sacrifices long-run optimality (higher multitasking ca-pacity) for short-term reward (faster learning). Furthermore,we construct a general mathematical framework in which ratio-nal choices between learning speed and processing efficiencycan be examined for a variety of different task environments.