A manifold neural population code for space in hippocampal coactivity dynamics independent of place fields

Abstract

Hippocampus CA1 place cells express a spatial neural code by discharging action potentials in cell-specific locations (′place fields′), but their discharge timing is also coordinated by multiple mechanisms, suggesting an alternative ′ensemble cofiring′ neural code, potentially distinct from place fields. We compare the importance of these distinct information representation schemes for encoding environments. Using miniature microscopes, we recorded the ensemble activity of mouse CA1 principal neurons expressing GCaMP6f across a multi-week experience of two distinct environments. We find that both place fields and ensemble coactivity relationships are similarly reliable within environments and distinctive between environments. Decoding the environment from cell-pair coactivity relationships is effective and improves after removing cell-specific place tuning. Ensemble decoding relies most crucially on anti-coactive cell pairs distributed across CA1 and is independent of place cell firing fields. We conclude that ensemble cofiring relationships constitute an advantageous neural code for environmental space, independent of place fields.

A sparse set of spikes corresponding to reliable correlations is highly informative of visual stimulus on single trials

“…Together these finding suggest that reliable pairwise correlations, rather than simply rate modulation or temporal pattern, are the building blocks of the coding scheme employed in layer 2/3 of cortex. A similar coding scheme termed 'ensemble cofiring' has been recently proposed in area CA1 in hippocampus 40 .…”

Do place cells dream of deceptive moves in a signaling game?

“…Furthermore, Infomax concepts are severely challenged to explain our additional observation that the cofiring relationships between simultaneously-recorded hippocampal and MEC neurons (Fig. 4C bottom) is only weakly persistent across the stable and rotating arena conditions, despite both the entorhinal (Park et al, 2019) and the hippocampal populations each exhibiting strongly persistent, attractor-like cofiring relationships (Levy et al, 2021). If MEC representations of space project to hippocampus then their discharge should be informationally coupled, and the discharge representations should be informationally coherent, perhaps even increasing in fidelity according to Infomax predictions.…”

“…That is however, not to say the anti-cofiring relationships are unimportant. On the contrary, they tend to be rare, their prevalence increases with learning, and they tend to be the most important contributors to the ability to discriminate information from population network activity (Levy et al, 2021). Once established, persistent conventions tend to further increase their persistence.…”

“…4C middle), further indicating that despite mixed tuning to different spatial variables (Park et al, 2019), as a whole, the MEC network manifests the stationary and steady temporal discharge dynamics that are characteristic of neuronal attractor dynamics (Yoon et al, 2013;Chaudhuri et al, 2019). Like MEC, the population discharge properties of hippocampus also exhibit attractor dynamic properties, in that the collective activity of MEC cells, or the collective activity of the cells of the hippocampus CA3 or CA1 subfields tend to exist as relatively stable patterns of activity that are readily described as a low-dimensional manifold in the high-dimensional activity space that is defined by the independent activities of the population of cells (Samsonovich and McNaughton, 1997;McNaughton et al, 2006;Yoon et al, 2013;Chaudhuri et al, 2019;Gardner et al, 2021;Nieh et al, Evidence of these manifolds of low-dimensional neuronal activity can be readily measured as conserved pairwise coactivity patterns across the network, which are known to represent higher-order correlations (Schneidman et al, 2006;Levy et al, 2021) (Fig. 4C).…”

Dynamic Structure Of Motor Cortical Neuron Co-Activity Carries Behaviorally Relevant Information

“…Spike time correlations have been implicated in the propagation of spiking in networks (Bojanek et al, 2020; Chambers and MacLean, 2016) making them an intriguing signal when studying the circuit mechanisms and computations that give rise to behavior. Recent work has shown that spike time correlations are an informative description of visual stimuli (Dechery and MacLean, 2018; Kotekal and MacLean, 2020; Levy et al, 2020; Ruda et al, 2020), auditory stimuli (Betzel et al, 2019; Insanally et al, 2019), and spatial location (Levy et al, 2022; Nardin et al, 2021). In motor cortex (M1) pairwise spike count correlations have been shown to provide information about motor behavior beyond what is provided by firing rates alone (Maynard et al, 1999), and have also been used to improve encoding models that predict the activity of neurons (Stevenson et al, 2012).…”