The mechanics of state-dependent neural correlations

Abstract

Simultaneous recordings from large neural populations are becoming increasingly common. An important feature of the population activity are the trial-to-trial correlated fluctuations of the spike train outputs of recorded neuron pairs. Like the firing rate of single neurons, correlated activity can be modulated by a number of factors, from changes in arousal and attentional state to learning and task engagement. However, the network mechanisms that underlie these changes are not fully understood. We review recent theoretical results that identify three separate biophysical mechanisms that modulate spike train correlations: changes in input correlations, internal fluctuations, and the transfer function of single neurons. We first examine these mechanisms in feedforward pathways, and then show how the same approach can explain the modulation of correlations in recurrent networks. Such mechanistic constraints on the modulation of population activity will be important in statistical analyses of high dimensional neural data.

Inhibitory control of correlated intrinsic variability in cortical networks

“…By combining spike-frequency adaptation (Destexhe, 2009; Latham et al, 2000) with high excitatory connectivity, our network is able to generate intrinsic global fluctuations that are of variable duration, arise at random times, and do not necessarily phase-lock to external input, thus creating noise correlations in evoked responses. This correlated intrinsic variability distinguishes our model from previous rate or spiking network models (Parga and Abbott, 2007; Renart et al, 2010; Wolf et al, 2014; Doiron et al, 2016), as well as from phenomenological dynamical systems (Macke et al, 2011; Pachitariu et al, 2013), all of which create noise correlations by injecting common noise into all neurons, an approach which, by construction, provides little insight into the biophysical mechanisms that generate the noise (Doiron et al, 2016). …”

“…While a number of modeling studies have explored the impact of correlations on sensory coding (Shadlen et al, 1996; de la Rocha et al, 2007; Averbeck et al, 2006; Pillow et al, 2008; Ecker et al, 2011; Moreno-Bote et al, 2014), there have been few efforts to identify their biophysical origin; the standard assumption that correlations arise from common input noise (de la Rocha et al, 2007; Doiron et al, 2016; Lyamzin et al, 2015) simply pushes the correlations from spiking to the membrane voltage without providing insight into their genesis. Models that use external noise to create correlations have been used in theoretical investigations of how network dynamics can transform correlations (Doiron et al, 2016), but no physiological source for the external noise used in these models has yet been identified.…”

“…Models that use external noise to create correlations have been used in theoretical investigations of how network dynamics can transform correlations (Doiron et al, 2016), but no physiological source for the external noise used in these models has yet been identified. However, no external noise is needed to generate the correlated activity that is observed in vivo; in vitro experimental studies have shown that cortical networks are capable of generating large-scale fluctuations intrinsically (Sanchez-Vives et al, 2010; Sanchez-Vives and McCormick, 2000), and in vivo results suggest that the majority of cortical fluctuations arise locally (Cohen-Kashi Malina et al, 2016; Shapcott et al, 2016).…”

Practical approximation method for firing-rate models of coupled neural networks with correlated inputs

“…4(a) shows correlations as low as −0.3 and as high as 0.3. Thus, the viability of our approximation is not limited to small correlation values, but can robustly capture the full range of correlation values observed in cortical neurons [5,32]. …”

From the statistics of connectivity to the statistics of spike times in neuronal networks

“…Collective spiking arises from two mechanisms: connections among neurons within a population, and external inputs or modulations affecting the entire population [11–13]. Experiments suggest that both are important.…”