Population coding in mouse visual cortex: response reliability and dissociability of stimulus tuning and noise correlation

Abstract

The primary visual cortex is an excellent model system for investigating how neuronal populations encode information, because of well-documented relationships between stimulus characteristics and neuronal activation patterns. We used two-photon calcium imaging data to relate the performance of different methods for studying population coding (population vectors, template matching, and Bayesian decoding algorithms) to their underlying assumptions. We show that the variability of neuronal responses may hamper the decoding of population activity, and that a normalization to correct for this variability may be of critical importance for correct decoding of population activity. Second, by comparing noise correlations and stimulus tuning we find that these properties have dissociated anatomical correlates, even though noise correlations have been previously hypothesized to reflect common synaptic input. We hypothesize that noise correlations arise from large non-specific increases in spiking activity acting on many weak synapses simultaneously, while neuronal stimulus response properties are dependent on more reliable connections. Finally, this paper provides practical guidelines for further research on population coding and shows that population coding cannot be approximated by a simple summation of inputs, but is heavily influenced by factors such as input reliability and noise correlation structure.

Synergistic population encoding and precise coordinated variability across interlaminar ensembles in the early visual system

“…Mean Fano factor across all cells in LGN was 1.11 -1.99, depending on the stimulus and 0.66 -0.91 in V1 ( Figure 1E, Fig S3), slightly lower than has previously been reported for Ca 2+ imaging in awake mice (1.39, ref. 46 ).…”

Experience-dependent reorganization drives development of a binocularly unified cortical representation of orientation

“…Furthermore, at the cellular-scale, we failed to find any systematic difference between binocular and monocular orientation selectivity in either naive or experienced animals (Supplementary Figure 3B Because three separate orientation representations emerge in the developing visual cortex prior to visual experience, we investigated if one of the representations was consistently further developed in encoding orientation-specific information at the cellular-scale. Here we employed a template matching decoder to predict the stimulus orientation presented on each trial by comparing trialevoked population activity patterns to the best matching trial-averaged response pattern ( Figure 4L) (Montijn et al, 2014). In naive animals, we could decode population responses from binocular and monocular stimulation at a rate higher than chance indicating that all three representations principally contribute to stimulus decoding ( Figure 4M, Decoding performance in naive animals: Binocular 42.9±5.8%, Contralateral 49.5±5.0%, Ipsilateral 40.2±2.7% (Mean±SEM) vs chance 12.5%).…”

“…To decode the stimulus orientation from the population activity vectors recorded during each trial, we used a normalized template matching algorithm (Montijn et al, 2014). In an effort to evaluate the decoding based on the relative responsiveness of each cell, the trial evoked ΔF/F0 responses of each cell were first individually z-scored: θ, n, i θ, n, i μ where i indexes the N th neuron, θ is the stimulation orientation, n is the trial number, θ, n, i is the z-scored response, θ, n, i is the actual ΔF/F0 response, µ is the mean ΔF/F0 value over the monocular or binocular recording period, and σ is the standard deviation from the mean ΔF/F0 value.…”

Neural ensemble activity depends on stimulus type in mouse primary visual cortex

“…However, sophisticated analysis techniques are required to analyse such large amounts of neurons and their interactions. One measure that lends itself to compare neural data under different conditions is the simultaneous firing configuration 11,12 of discretised and binarised individual spike times of each channel or neuron, henceforth termed patterns. These patterns may have different probabilities of occurring depending on what process or context generated them.…”