Information processing in the different Basal Ganglia sub regions

The basal ganglia (BG) are a group of sub-cortical nuclei, which together with the cortex, work to execute habitual and goal directed behaviors requiring motor, cognitive and limbic structures. Experimental and theoretical studies have suggested that the BG nuclei implement a reinforcement learning algorithm in order to maximize gained reward. However, it is not clear how information is processed along this network, thus enabling it to perform its functional role. I therefore studied the information processing along the BG nuclei in response to appetitive and aversive behavioral events and in different brain states. To this end, I recorded the activity of single cells in different areas of the BG of behaving monkeys in two states: while they were engaged in a classical conditioning paradigm involving rewarding (food), aversive (air-puff) and neutral (neither) outcomes and while their eyes were open vs. when they were closed. I compared the neuronal activity within and across BG regions; namely, the input stage of the BG, the striatum and its pallidal downstream targets, the central nucleus of the BG, the external segment of the globus pallidus (GPe) and the output stage of the BG, the internal segment of the globus pallidus (GPi). In primates the striatum is divided into three territories: the putamen, caudate and ventral striatum (VS) which all converge onto the pallidum and onto the same motor pathway. This parallel organization suggests that there are multiple and competing control systems over behavior in the BG network. To explore which mechanism(s) enables the different striatal domains to encode behavioral events and control behavior, I compared the neural activity across striatal territories during the performance of the classical conditioning task. I found that although neurons in all striatal territories displayed similar temporal modulations of their discharge rates to the behavioral events, their correlation structure was profoundly different. Specifically, putamen neuron pairs displayed increased correlations compared to the closer to zero correlations in the caudate and VS. The synchronization only observed between putamen cells transiently increased following the behavioral events and displayed different correlation dynamics to rewarding vs. neutral/aversive cues. These results indicate that the increased neural correlations observed solely in the putamen enabled efficient information encoding and were suggestive of a more efficient information transfer from the putamen to its downstream targets, GPe and GPi, during the performance of a well-practiced task. Furthermore, I studied how the activity in the striatum contributed to the activity generated in its downstream target, in the GPe. I examined the response profile and dynamical behavior of striatal and GPe neurons during the performance of the classical conditioning task. Both cell populations displayed sustained average activity to cue presentation. However, the population average response of striatal cells was composed of three distinct response groups which were temporally differentiated and fired in serial episodes along the trial. The population average response of GPe cells was composed of two response groups which overlapped in their time of activation and displayed persistent activity also at the single cell level. These results therefore support a functional, and not just anatomical convergence of striatal response groups on to GPe cells. Finally, current anatomical models of the BG network predict reciprocal discharge patterns between the GPe and GPi. However, physiological studies revealing similarity in the transient responses of GPe and GPi neurons cast doubts on these predictions. I found that both pallidal populations exhibited decreased discharge rates in the "eye closed" state accompanied by elevated values of the coefficient variation (CV) of their inter-spike interval (ISI) distributions. In addition, the pallidal discharge modulations were gradual, starting prior to closing of the eyes. Thus, changes in GPe and GPi discharge properties were positively correlated, suggesting that the sub-thalamic nucleus and/or the striatum are the main common driving force for both pallidal segments. Together these results suggest that in a well-practiced behavior, the BG network is organized such that interconnected sub networks in the putamen encode behavioral events and enable efficient information transfer to downstream BG