Cortical reliability amid noise and chaos

Abstract

Typical responses of cortical neurons to identical sensory stimuli are highly variable. It has thus been proposed that the cortex primarily uses a rate code. However, other studies have argued for spike-time coding under certain conditions. The potential role of spike-time coding is constrained by the intrinsic variability of cortical circuits, which remains largely unexplored. Here, we quantified this intrinsic variability using a biophysical model of rat neocortical microcircuitry with biologically realistic noise sources. We found that stochastic neurotransmitter release is a critical component of this variability, which, amplified by recurrent connectivity, causes rapid chaotic divergence with a time constant on the order of 10-20 milliseconds. Surprisingly, weak thalamocortical stimuli can transiently overcome the chaos, and induce reliable spike times with millisecond precision. We show that this effect relies on recurrent cortical connectivity, and is not a simple effect of feed-forward thalamocortical input. We conclude that recurrent cortical architecture supports millisecond spike-time reliability amid noise and chaotic network dynamics, resolving a long-standing debate.All rights reserved. No reuse allowed without permission.(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

Interstitial ions: A key regulator of state-dependent neural activity?

“…1B and Table 1 ), triggered this transition ( Markram et al, 2015 ). Comparable findings were recently obtained, showing that decreasing [Ca 2+ ] o from 1.35 to 1.2 mM caused neuronal activity to transition from a state of regular, burst-like firing to a state of irregular, tonic-like firing, and this transition was further characterized by a shift from a supercritical to a subcritical state ( Nolte et al, 2019 ). Collectively, these data seem to suggest that within the physiological range, higher [Ca 2+ ] o associate with burst firing and periods of synchrony, while lower [Ca 2+ ] o associate with tonic firing and desynchronous activity ( Markram et al, 2015 ; Newton et al, 2019 ).…”

Topology of synaptic connectivity constrains neuronal stimulus representation, predicting two complementary coding strategies

“…This means that a neuron participating in connection motifs that generate correlated input, will be less subject to noise, and constrained more tightly to the manifold, improving the value of its spike train in decoding the manifold. Directed simplices have been shown to generate such correlations before (Nolte et al, 2019).…”

“…It models morphological detail in the form of 55 morphological types of neurons and type-specific synaptic transmission between them (Ramaswamy et al, 2015). Crucially, the model of synaptic transmission includes short-term plasticity (Stevens and Wang, 1995; Markram and Tsodyks, 1996; Abbott, 1997) and both spontaneous release (Simkus and Stricker, 2002; Ling and Benardo, 1999) and synaptic failure (Ribrault et al, 2011), both important sources of noise in neural microcircuit (Nolte et al, 2019; Faisal et al, 2008).…”

“…Such redundancy is often a means to overcome the presence of noise in a system (Tkacik et al, 2010). Indeed, there are various sources of noise affecting neural activity and it has been demonstrated to be often unreliable (Nolte et al, 2019; Tolhurst et al, 1983; Stern et al, 1997; Shadlen and Newsome, 1998). Consequently, a model to study the structural implementation of a neural manifold needs to include these noise sources, to fully understand its function.…”

Computational roles of plastic probabilistic synapses

“…On the functional side several properties may be attributed to probabilistic synapses, such as encoding perceptual uncertainty (Fiser et al, 2010;Aitchison and Latham, 2015), escaping local optimum (Seung, 2003;Blundell et al, 2015;Kappel et al, 2015Kappel et al, , 2018, but also reflecting biophysical constraints (Harris et al, 2012;Costa et al, 2015Costa et al, , 2017c) in the addition to contributing to information processing (Zhang and Peskin, 2015;Nolte et al, 2018). It is conceivable that plastic probabilistic synapses enable not just one, but several of these computational functions.…”