Divide and conquer: How avian 'prefrontal' and hippocampal neurons process extinction learning in complementary ways

Abstract

Context-dependent extinction learning enables organisms to acquire an inhibition of responding to cues that no longer signal reward in specific environmental settings. Both the hippocampus and the prefrontal cortex play key but complementary roles in encoding extinction. To understand what drives the differential contributions of these two structures, we recorded single-unit responses from the pigeon hippocampus (HPC) and the prefrontal nidopallium caudolaterale (NCL), while the animals were engaged in a repeated appetitive ABA extinction learning paradigm. HPC carried more information about experimental phases (acquisition (context A), extinction (context B), and renewal (context A)) than NCL. In addition, hippocampal neurons with mixed selectivity integrated all contextual stimuli and conditioned cues, thereby possibly enabling context integration into phase-dependent extinction events. The mixed selectivity pattern of NCL neurons differed: They encoded all relevant task parameters separately, thereby possibly enabling the weighing and deciding between response alternatives. Repeatedly testing the same animals over many months enabled us to reveal that the activity of NCL neurons increasingly became predictive of the number of responses that animals made during renewal. Thus, NCL neurons possibly assumed a meta-learning ability that enabled a behavioral adaption to the deeper task structure that always followed the sequence of acquisition[->]extinction[->]renewal. Our results make it likely that different processing strategies of hippocampus and prefrontal NCL enable a differential encoding of experimental phase, context, decision-making, and meta-learning during extinction learning.