Neural bases of decision based on multiple episodes

Task leader: GNT.

Over multiple experienced episodes, patterns of rewarding and/or efficient behaviour can be extracted, stabilizing choices of overall strategies and of individual actions within a strategy. The emergence of consistent strategy and action selection correlates with the emergence and structuring of activity in the Pfc-hippocampal-striatal circuit. Different behavioural strategies are known to require different zones of the striatum, suggesting that competition at the scale of whole cortico-striatal loops leads, over time, to the choice of correct strategy. For individual actions, the emergence of hippocampal-Pfc coherence precedes and predicts correct decisions at maze choice points. The convergence of the hippocampal and Pfc output at the striatum suggests coherent activity in these structures is a prerequisite for striatal computation of action choice. The purpose of this task is to pull together these threads, developing an understanding of the neural bases for the emergence of consistent action and strategy selection over experience.


Here, rats perform binary choice tasks (action selection), but also must switch between different strategies requiring different striatal zones (of which some do not receive hippocampal inputs). The rats are recorded during learning, and thus their acquisition of new rules is based on many trials, or even several sessions of training. The simultaneous recording data will inform the development of the spiking neuron models which permit development of local field potentials, coherent oscillations and cell assembly formation and replay.

Subtask 1 - How does Pfc-hippocampal coherence arise at decision points on the basis of experience?

To show how the projection from hippocampal formation to Pfc, linking two complex sets of microcircuits, could underpin prospective encoding in Pfc for goals or future actions, based on past experience. We will test two hypotheses: that coherence between the oscillatory activities in both structures is necessary for selection of correct goals and/or actions (Benchenane et al, 2007); and that this coherence emerges across multiple episodes through experience-dependent plasticity of the hippocampal-Pfc pathway.

Subtask 2 - What computations does the striatum perform on convergent hippocampal and Pfc input for the selection of action and strategies?

To understand how the striatum engages these limbic inputs for behavioral choices. Pfc and hippocampal formation input converges in a restricted region of ventral striatum and its dorso-medial edge. These inputs are integrated by individual striatal neurons, but the computations performed by the striatum are still an open question. We will explore separate hypotheses for the core and shell regions of the ventral striatum (Humphries & Prescott, 2009). First, convergence in the core leads to selection of locomotion direction (and hence construction of paths). Second, convergence in the shell underpins evaluation of the action outcome. We will thus study in detail striatal network processing of inputs.

Subtask 3 - Neural bases of action and strategy selection based on hippocampal inputs

To determine brain algorithms for learning these functions and compare with model predictions. Recordings of coherence of LFPs and emergence of cell assemblies in the Hpc, Pfc and striatum of behaving rats as they learn, perform and switch between tasks that are dependent or independent of hippocampal function.