Hippocampo-cortical contributions to world building in freely behaving macaques

点击翻译按钮获取中文摘要

PROJECT SUMMARY/ABSTRACT When learning in complex, realistic, or even real worlds, we have the benefit of using different strategies adaptively. For most primate brains, adaptive means adjusting as a function of where we are, who we are with, and what things of use are in view or in reach. Learning theories like Complementary Learning Systems (CLS) originally suggested that the hippocampus and neocortical structures contributed distinct computations to represent different kinds of memory. This theory relied heavily on assumptions about the finer structure of neurons in these areas, built largely from knowledge of these structures in rats and to some extent mice. Methodological limitations have prevented measuring in primates (human or monkey) the fine circuit computations predicted by these models. This has led to assumptions that the computations are similar to those in rodents, yet rodents have very different real-world behaviors from primates. We propose to check these assumptions and extend and/or revise the theory, by recording wirelessly in macaques who learn rules about objects in an immersive, real-world enclosure. We will use high- density, multi-site recordings in and around the hippocampus to test two major aspects of memory theory in need of resolution. First, we ask if there are differences in the two main hippocampal-CA1 inputs in supporting episodic and category learning. This question derives from an untested prediction of our expanded CLS model. We will record wirelessly as macaques make decisions about the assignment of object exemplars (‘FauXna’) on displays set up in their environment. The model predicts that CA3-CA1 inputs are particularly supportive of the arbitrary mappings required for episodic memory, whereas layer III entorhinal cortex ‘direct’ inputs are more involved in integrating information across trials, affording object category learning. Using high-definition linear arrays, we can resolve CA1 dendritic field currents as well as multi-site ensemble unit activity, allowing us to test our prediction for the first time. Second, we ask if the hippocampal and connected neocortical dynamics play a role in memory retrieval as a function of either memory age or of the episodic/semantic nature of the task. We will use closed-loop stimulation to interrogate the necessity of each region during recall, and the role of coordinated activity between hippocampus and neocortex for recall, across memory age and type. From these experiments we will (1) disambiguate several competing theories of the division of labor across nodes in the memory network, (2) create the first conceptual microcircuit model of these memory systems in the primate brain, and (3) contrast with our expanded computational model.