Separating Spike Count Correlation from Firing Rate Correlation

Abstract

Populations of cortical neurons exhibit shared fluctuations in spiking activity over time. When measured for a pair of neurons over multiple repetitions of an identical stimulus, this phenomenon emerges as correlated trial-to-trial response variability via spike count correlation (SCC). However, spike counts can be viewed as noisy versions of firing rates, which can vary from trial to trial. From this perspective, the SCC for a pair of neurons becomes a noisy version of the corresponding firing-rate correlation (FRC). Furthermore, the magnitude of the SCC is generally smaller than that of the FRC, and is likely to be less sensitive to experimental manipulation. We provide statistical methods for disambiguating time-averaged drive from within-trial noise, thereby separating FRC from SCC. We study these methods to document their reliability, and we apply them to neurons recorded in vivo from area V4, in an alert animal. We show how the various effects we describe are reflected in the data: within-trial effects are largely negligible, while attenuation due to trial-to-trial variation dominates, and frequently produces comparisons in SCC that, because of noise, do not accurately reflect those based on the underlying FRC.

Condition-Dependent Neural Dimensions Progressively Shift during Reach to Grasp

“…A third part of firing rate variance in M1 neurons, like that of many other CNS neurons, is noise. Some of this noise may reflect variation in neural signals related to features not controlled in the current experiment, which becomes evident in the shared noise correlations among a population of M1 neurons ( Lee et al, 1998 ; Goris et al, 2014 ; Vinci et al, 2016 ). Noise also may include a stochastic component that reflects individual neuron variability in the cellular processes underlying spike generation (see, however, Mainen and Sejnowski, 1995 ).…”

Discrimination of Motion Direction in a Robot Using a Phenomenological Model of Synaptic Plasticity

“…In general, these shared fluctuations are measured between pairs of neurons over multiple presentations of an identical stimulus. To examine these coordinated fluctuations in spiking activity, we used a measure of spike count correlation (SCC) between pairs of neurons in V1 during motion discrimination [47]. In doing so, we used a nonoverlapping time window of 1 millisecond to compute the total number of spikes emitted from each neuron at a given time step.…”

Adjusted regularization in latent graphical models: Application to multiple-neuron spike count data

“…However, two major complications arise. First, the correlation across repeated trials between the spike counts of two neurons is corrupted by Poisson-like variation within the trials of each neuron, and this attenuates the estimates of across-trial correlations (Behseta et al, 2009; Vinci et al, 2016). Second, methods based on λ 1 regularization are better suited to estimate sparse networks (Rothman et al, 2008; Ravikumar et al, 2011) but in the context of spike count data, networks are not typically very sparse.…”

“…Second, methods based on λ 1 regularization are better suited to estimate sparse networks (Rothman et al, 2008; Ravikumar et al, 2011) but in the context of spike count data, networks are not typically very sparse. In previous work we handled the first problem for pairs of neurons using bivariate hierarchical models (Vinci et al, 2016), where pairs of spike counts were assumed to be Poisson with bivariate log-normally distributed latent means; correlation between the latent means can then be thought of as a Poisson de-noised version of spike count correlation. We also handled the second problem, while ignoring the first, by allowing the LASSO penalty to vary with the pair of neurons based on informative covariates (Vinci et al, 2018a), which improved the detection of correct edges and the accuracy of partial correlation estimates.…”

“…The key to our approach is the availability of informative covariates, which rests on well-documented neurophysiology concerning correlation among pairs of cortical neurons. In several studies, Pearson correlation has been shown to depend strongly on both inter-neuron distance (physical distance between the neurons) and tuning curve correlation (similarity of the two neurons’s average activities over different stimuli): the dependence is typically weaker at larger inter-neuron distance, and stronger for larger tuning-curve correlation (Smith and Sommer, 2013; Goris, Movshon, and Simoncelli, 2014; Yatsenko et al, 2015; Vinci et al, 2016). These covariates are available, but their effects may vary across brain areas, neuron types, and recording techniques.…”