Research Project 4 - Internal state dynamics of navigation and memory

Research Project 4 – Internal state dynamics of navigation and memory Lead: Lisa Giocomo PhD Project Summary In RP4 (Navigation and memory) we leverage the navigation system to experimentally investigate theoretical and computational principles for how external sensory inputs and internal network dynamics, across

different brain states, interact to generate the neural computations necessary for navigation. We focus on four brain regions that provide complementary computations for visually guided navigation in mice: primary visual cortex (V1), medial entorhinal cortex (MEC), hippocampus (HPC) and retrosplenial cortex (RSC). In our first

aim, we consider how internal dynamics interact with external sensory inputs to generate a unified percept of position. We leverage virtual reality and the high density recording capabilities of Neuropixel silicon probes to explicitly test predictions of a Bayesian cue integration framework developed in RP3 (Theory and

computation of internal state dynamics). Here, internal dynamics reflect intrinsic path integration calculations, while external inputs include visual landmark and optic flow inputs. In our second aim, we consider how a change in behavioral state impacts the stability of neural maps of space and test theoretical principles,

developed in RP3 and applied to V1 in RP1 (Motivation and perception) and RP2 (Primate decision making), for how spontaneous activity influences the detection or amplification of weak sensory inputs in the navigation circuit. As in Aim 1, we leverage virtual reality and the high density recording capabilities of

Neuropixel silicon probes. Here, we consider a change in spontaneous activity as analogous to a change in internal behavioral state (satiety or arousal) and consider how this impacts the stability of internal position estimates, as measured by the spatial firing patterns of neurons, across environments with parametrically

differing external landmark strength (cue rich or cue poor conditions). In our third aim, we consider whether driving the activity of single neurons can establish causality between neural representations in RSC and visually- guided navigation. By applying a MultiSLM 2-photon Ca2+ imaging method developed in RP1, which enables

wide-field optical access for visualization and control of cellular ensembles in real time, we will test theories developed in RP2 regarding attractor states and whether critically excitable regimes capable of driving behavior, which have been observed in V1, exist in non-sensory cortical regions. Here, external input is manipulated with

single-cell resolution optogenetically, and intrinsic network dynamics for encoding internal position estimates measured using 2P Ca2+ imaging. Together, across all of our aims, our approach of investigating multiple navigationally relevant brain regions alongside V1 will allow us to rigorously consider the degree to which

foundational theories for classes of neural computation – developed in RP3 – follow universal principles across cortical regions or show divergence based on how disparate brain circuits weight internal dynamics versus external inputs.