A disinhibition-based model of CA1 place field sequences

Abstract

Because of the paucity of pattern-generating recurrent excitation in the CA1 hippocampal region, place field sequences are assumed to be inherited from upstream regions. Alternatively, disinhibition may play a role in shaping pyramidal cell activity. We propose a model of CA1 activity using one layer of excitatory cells and one layer of inhibitory cells, with local excitatory-to-inhibitory connectivity, broad inhibitory-to-excitatory connectivity without recurrent excitatory connections, and broad inhibitory-inhibitory connections. The layers need not follow a topographic organization; instead they can represent sequences of place cells structured by local inhibition. Our mathematical model of CA1 neurons with data-inspired connectivity demonstrates place fields sharpened by disinhibition, mimicking a recently discovered decrease in inhibitory input within place fields and matching in vivo patterns of disinhibition. Mathematical analysis of our model suggests the importance of inhibitory-to-excitatory connections in control of place cell firing rate and sparsity. We also demonstrate that a disinhibition-based mechanism of pattern generation is more flexible without recurrent excitation than the recurrent excitation-rich CA3-like model. Finally, we demonstrate disinhibition as an efficient mechanism for internally-generated sequential activity, where the speed of sequential change depends on both input strength and inhibitory-to-excitatory connection strength. Overall, our model suggests that theta time-scale disinhibition is a foundational mechanism of self-organized sequential activity in the CA1 region, and, possibly, in other networks.