The effect of noise correlations in populations of diversely tuned neurons

Abstract

The amount of information encoded by networks of neurons critically depends on the correlation structure of their activity. Neurons with similar stimulus preferences tend to have higher noise correlations than others. In homogeneous populations of neurons this limited range correlation structure is highly detrimental to the accuracy of a population code. Therefore, reduced spike count correlations under attention, after adaptation or after learning have been interpreted as evidence for a more efficient population code. Here we analyze the role of limited range correlations in more realistic, heterogeneous population models. We use Fisher information and maximum likelihood decoding to show that reduced correlations do not necessarily improve encoding accuracy. In fact, in populations with more than a few hundred neurons, increasing the level of limited range correlations can substantially improve encoding accuracy. We found that this improvement results from a decrease in noise entropy that is associated with increasing correlations if the marginal distributions are unchanged. Surprisingly, for constant noise entropy and in the limit of large populations the encoding accuracy is independent of both structure and magnitude of noise correlations.

Tuning and spontaneous spike time synchrony share a common structure in macaque inferior temporal cortex

“…This has been postulated by several reports (Averbeck et al 2006;DiCarlo et al 2012;Ecker et al 2011;Fujita et al 1992;Tanaka 2003), and several studies have investigated population coding by IT neurons at coarse spatial scales (Hung et al 2005;Rolls and Tovee 1995;Sato et al 2013). However, to our knowledge, our study is the first to explicitly test this for correlated responses within and across neighboring columns, from which most nearby neurons Tuning responses were grouped as k-means clusters (black), random clusters (red), within-penetration averages (green), and within-depth averages (blue).…”

“…This link has been explored as pairwise interactions between neurons in macaque inferior temporal (IT) cortex, in terms of their tuning correlations (similarity in their stimulus preferences) and spike time correlations. Investigating this link is important because the specific correlational structure is postulated to affect redundancy and representational capacity, depending on the relationship between tuning and spike time correlation and on the size of the neuronal population (Averbeck et al 2006;Cohen and Kohn 2011;Ecker et al 2011;Gawne and Richmond 1993). Spike correlation within a narrow time window (spike time synchrony) is also postulated to enable downstream neurons to integrate information via coincidence detection and specific timing delays (Schwarzlose et al 2005) and may be important for learning, e.g., via spike timing-dependent plasticity (Dan and Poo 2006;Masquelier and Thorpe 2007;Yao et al 2004).…”

On the Cramér–Rao bound applicability and the role of Fisher information in computational neuroscience

“…An assessment of the eect of heterogeneity on the coding accuracy of a correlated population can be found in Chelaru and Dragoi (2008), Shamir and Sompolinsky (2006), Ecker et al (2011) and Wilke and Eurich (2001). These studies concluded that a high variability of the tuning curves among the population of the neurons is benecial to the coding accuracy of the ensemble.…”

Correlated discharges in the primate prefrontal cortex before and after working memory training

“…, 1994), although this may also depend on how signals are pooled (Abbott & Dayan, 1999; Romo et al. , 2003) and the heterogeneity of the neuronal population (Ecker et al. , 2011).…”