The rationale for this work is based upon the observation that in humans, organization, planning and flexibility of behavior depend critically on the ability to encode sequences of events in episodic memory. Recollecting episodes then allows the likely outcome of situations that follow a familiar pattern to be anticipated. How memories are coded in the brain and used for purposeful behavior remains unclear. In particular, the link between various behavioral time scales needs to be explained. Our goal here is to help better understand and replicate how this brain system memorizes and organizes information at several different timescales and how multiple brain areas are coordinated for this. The regimes focussed on here include:
- Focus 1: Linking immediately previous events with imminent behavioural choices to form a representation of a brief series of events (on the order of 1-10 s).
- Focus 2: Linking series of events to form episodes (on the order of tens of seconds to minutes).
- Focus 3: Engaging experience from multiple episodes (on the order of hours or more) for decision making (i.e. action selection and strategy selection).
- Neural bases of working memory and application for imminent action (leader: LPPA).
LPPA will investigate the prospective and retrospective firing in Str, Hpc and Pfc. The experiments record neurons predicting simple decision making is performed (turn left or turn right). Hippocampal models will be examined for whether Pfc and striatum modules are necessary for this property to emerge. LNC will determine how neuronal activity in Pfc is linked to executive processing and working memory. Pfc and Hpc may also cooperate for encoding those aspects. To investigate these interactions, we will perform simultaneous unit recordings in Hpc and the Pfc. - Neural bases of coding series of events within episodes (leader: LNC).
Neurophysiological recordings in rats performing behavioral task requiring sequences of visits to arms on a radial maze (simple decision making) are performed. As a tool to determine what neural activity is related to these memory functions, recordings are compared between task versions requiring memory of previous visits or not. - Neural bases of decision on multiple episodes (leader: GNT).
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. - Building global models from local circuits (leader: ETIS).
The ETIS hippocampus-prefrontal cortex model (where future choices are coded as place-to-place transitions) will be integrated with the GNT spiking model, in consultation with LPPA and LNC for neurophysiologically based algorithms. Robotic implementation of these models in challenges based upon the behavioral neuroscience experiments will permit comparison of the relative benefits and drawbacks of the respective modeling approaches in approaching (or surpassing) the performance of the biological systems.
The scientific program involves four main elements:
- execution and analysis of neurophysiological and behavioral recordings of prospective and working memory related activity in the Hpc-Pfc-Str system,
- extending and integrating existing models based on these and earlier related data,
- refining these models, making simulations and implementing them on an autonomous robot platform, and
- dissemination of these results at scientific congresses and in publications submitted to international refereed journals.