Hebbian Learning in a Random Network Captures Selectivity Properties of Prefrontal Cortex

摘要: Complex cognitive behaviors, such as context-switching and rule-following, are thought to be supported by prefrontal cortex (PFC). Neural activity in PFC must thus specialized specific tasks while retaining flexibility. Nonlinear 'mixed' selectivity is an important neurophysiological trait for enabling complex context-dependent behaviors. Here we investigate (1) the extent which exhibits computationally-relevant properties mixed (2) how could arise via circuit mechanisms. We show that cells recorded during a task moderate level of specialization structure not replicated model wherein receive random feedforward inputs. While connectivity can effective at generating selectivity, data shows significantly more than predicted with otherwise matched parameters. A simple Hebbian learning rule applied connectivity, however, increases allows match accurately. To explain achieves this, provide analysis along clear geometric interpretation impact on selectivity. After learning, also matches measures noise, response density, clustering, distribution selectivities. Of two styles tested, simpler biologically plausible option better data. These modeling results give intuition about neural cognition make experimental predictions regarding various would evolve animal training.