Studies involving hippocampal neural ensemble recordings in rodents performing a delayed-nonmatch-to-sample (DNMS) task have demonstrated a distinct correlation between a sequence of external stimuli (i.e., behavioral events), patterns of firing across hippocampal neurons in response to those stimuli, and the animal's decision-making process which determines the response on a given trial. Behavioral errors have been shown to correlate to neural activity patterns (i.e. ensemble """"""""states"""""""") that were at variance with the actual stimuli or behavioral responses executed by the animal. As a model of decision making, hippocampal activity during performance of the DNMS task can be viewed as the result of a neural circuit that integrates information regarding task-relevant stimuli, task demands, spatial position and trial sequence across trials, and predicts the appropriate signal strength for encoding task-relevant information on the next trial. Thus, ensemble firing pattern of weak to moderate strength (as measured via information theoretic and canonical discriminant analyses) may sufficiently encode information sufficient for correct responses on trials with short delays, but not on long delays. In contrast, strongly encoded information would bridge a long delay, but could interfere with the encoding of novel information on the subsequent trial. On a given trial, the role of a prediction circuit including the hippocampus is to determine the appropriate response based on the prior task-relevant information. In fact, two decisions are being performed by this neural circuit, the appropriate response, and the appropriate signal strength for the encoding. This project is designed around the following goals with respect to studying this hippocampal decision circuit: (1) In depth characterization of patterns of spatial and temporal correlation between hippocampal neurons during the decision-making stages of the spatial and nonspatial variants of the DNMS task. (2) Identification of the temporal nature of neural firing patterns that correlate to the decisions required for successful behavioral response, combined with known placement of recorded neurons with respect to hippocampal anatomy. (3) Construction of a theoretical model of the information processing performed by the hippocampus. (4) Implementation of an Artificial Neural Network model of the processing performed by the hippocampus. (5) Evaluation of the """"""""biological computing"""""""" capabilities of the in vivo mammalian hippocampus by presenting specific computational problems in the form of trial-specific stimulus sequences, and reading the solutions to those computations as a combination of hippocampal neural activity and behavioral outcome.
