Robust odor coding across states in piriform cortex requires recurrent circuitry: evidence for pattern completion in an associative network

Abstract

Pattern completion, or the ability to retrieve stable neural activity patterns from noisy or partial cues, is a fundamental feature of memory. Theoretical studies indicate that recurrently connected auto-associative or discrete attractor network models can perform this process. Although phenomenological evidence for pattern completion and attractor dynamics have been described in various recurrent neural circuits, the crucial role that recurrent circuitry plays in implementing these processes has not been shown. Here we show that although odor representations in mouse olfactory bulb degrade under anesthesia, responses in downstream piriform cortex remain robust. Recurrent connections are required to stabilize cortical odor representations across states. Moreover, piriform odor representations exhibit attractor dynamics, both within and across trials, and these are also abolished when recurrent circuitry is eliminated. Thus, an auto-associative cortical circuit stabilizes output in response to degraded input, and the recurrent circuitry that defines these networks is required for this stabilization.

A Model For Hierarchical Memory Storage in Piriform Cortex

“…Recent work suggests that piriform cortex accomplishes pattern separation and completion through attractor dynamics [5, 6]. The most popular model of this activity is known as the Hopfield model which uses a Hebbian learning rule to update synaptic weights [7].…”