Information theoretic approaches to deciphering the neural code with functional fluorescence imaging

Abstract

6 Information theoretic metrics have proven highly useful to quantify the relationship between behaviorally 7 relevant parameters and neuronal activity with relatively few assumptions. However, such metrics are 8 typically applied to action potential recordings and were not designed for the slow timescales and variable 9 amplitudes typical of functional fluorescence recordings (e.g. calcium imaging). Therefore, the power of 10 information theoretic metrics has yet to be fully exploited by the neuroscience community due to a lack 11 of understanding for how to apply and interpret the metrics with such fluorescence traces. Here, we used 12 computational methods to create mock action potential traces with known amounts of information and 13 from them generated fluorescence traces to examine the ability of different information metrics to recover 14 the known information values. We provide guidelines for the use of information metrics when applied to 15 functional fluorescence and demonstrate their appropriate application to GCaMP6f population recordings 16 from hippocampal neurons imaged during virtual navigation. 17

Choice of method of place cell classification determines the population of cells identified

“…To investigate the information about location carried in place or non-place cells identified by the different methods, we calculated the mutual information between the cell’s activity and the animal’s location, providing an approximation of the spatial information carried in each cell’s calcium signals [26–28]. More spatial information was carried in place cells identified by the Peak method compared to non-place cells, but there was no difference in mutual information held by the place cell and non-place cell populations for the Stability method (Fig 5D).…”

Choice of method of place cell classification determines the population of cells identified

“…The original method relied on the detection of calcium events, but to reduce the effect of such preprocessing we adapted the method to be applied directly to the relative fluorescence. As using the fluorescence directly has been shown to accurately reflect the spatial information [34], we do not expect this to negatively impact the method's ability to detect place cells. We first calculated the fluorescence maps (the average fluorescence in each location bin) for each cell from which we determined the spatial information using Eq 3, where N is the number of bins, f i is the fluorescence in bin i of the fluorescence map, and f is the average fluorescence across the map.…”

“…To investigate the information about location carried in place or non-place cells identified by the different methods, we calculated the mutual information between the cell's activity and the animal's location, providing an approximation of the spatial information carried in each cell's calcium signals [32][33][34]. More spatial information was carried in place cells identified by the Peak and Information method compared to non-place cells, but there was no difference in mutual information held by the place cell and non-place cell populations for the Stability or Combination methods (Fig 6D).…”

Behavior determines the hippocampal spatial mapping of a multisensory environment

“…We attempted to sample from different fields of view in each session, though it is possible that some populations were overlapping. We used a spatial information metric (Climer and Dombeck, 2020;Gothard et al, 1996b;Skaggs et al, 1993) to classify each neuron according to its preferred coordinate, if any. This revealed start-spatial (i.e., distance from start), visuo-spatial (i.e., distance from visual target), and olfacto-spatial (i.e., distance from odor target) neurons that fired in sequences tiling their preferred coordinate (Figures 2A-2C and S2A).…”

“…The population of CA1 neurons preferentially maps the behaviorally relevant coordinate Our findings thus far were deduced from the proportion of neurons that contained the most information about the different coordinates, but even sparsely active or low-information neurons can also contribute to the hippocampal cognitive map (Climer and Dombeck, 2020;Meshulam et al, 2017). We therefore performed Bayesian reconstruction, a population-level analysis method, for each coordinate during each type of navigation, to determine the robustness of our findings when including the full population of active neurons.…”