Dynamic Structure of Neural Variability in the Cortical Representation of Speech Sounds

Abstract

Accurate sensory discrimination is commonly believed to require precise representations in the nervous system; however, neural stimulus responses can be highly variable, even to identical stimuli. Recent studies suggest that cortical response variability decreases during stimulus processing, but the implications of such effects on stimulus discrimination are unclear. To address this, we examined electrocorticographic cortical field potential recordings from the human nonprimary auditory cortex (superior temporal gyrus) while subjects listened to speech syllables. Compared with a prestimulus baseline, activation variability decreased upon stimulus onset, similar to findings from microelectrode recordings in animal studies. We found that this decrease was simultaneous with encoding and spatially specific for those electrodes that most strongly discriminated speech sounds. We also found that variability was predominantly reduced in a correlated subspace across electrodes. We then compared signal and variability (noise) correlations and found that noise correlations reduce more for electrodes with strong signal correlations. Furthermore, we found that this decrease in variability is strongest in the high gamma band, which correlates with firing rate response. Together, these findings indicate that the structure of single-trial response variability is shaped to enhance discriminability despite non-stimulus-related noise.

Not optimal, just noisy: the geometry of correlated variability leads to highly suboptimal sensory coding

“…An abundance of work has aimed to assess whether observed correlated variability is beneficial or detrimental for neural coding[6, 9–12, 16, 19, 24–33]. These studies often quantify the discriminability or fidelity of a neural code with the linear Fisher information (LFI, see Section) [36], which is a measure of how well the neural activity could be used to discriminate between different stimuli.…”

“…Many studies have found pairwise correlations in the trial-to-trial variability of the firing rates of simultaneously recorded neurons, often called correlated variability or noise correlations [3–9]. The correlated variability observed in experimental studies typically depends on the tuning and stimuli [10–12].…”

“…The RGCs’ correlated variability for a single stimulus is shown in Figure 1e . Although correlated variability is typically considered in simultaneous single neuron electrophysiology measurements, it has been observed in calcium imaging recordings [13] and larger scale measurements such as electrocorticography recordings [9]. Correlated variability has many possible biological sources in neural populations (see Supplementary Fig.…”

Semi-automated Anatomical Labeling and Inter-subject Warping of High-Density Intracranial Recording Electrodes in Electrocorticography

“…We take advantage of tools provided in the nipy software package ( https://github.com/nipy/nipy/ ), dural surface reconstruction from ielu (LaPlante et al, 2016 ), 3D plotting in mayavi ( http://mayavi.sourceforge.net/ ; Ramachandran, 2001 ), and extend on functions available in the MATLAB-based CTMR package (Hermes et al, 2010 ). This protocol has been used to localize and label electrodes in our previously published work (Dichter et al, 2016 ; Hamilton et al, 2016 ; Leonard et al, 2016 ; Moses et al, 2016 ; Muller et al, 2016b ; Tang et al, 2017 ). In an effort to promote open and affordable access to these tools, all requirements to run the pipeline (aside from physical hardware) are freely available for download at no cost to the user.…”

Tracking the dynamic representation of consonants from auditory periphery to cortex

“…How these perceptual and articulatory distinctions are reflected in cortex is a question of significant theoretical and translational interest. Prior studies have used intracranial high-density electrode arrays, or electrocorticography (ECoG), to measure cortical activity during speech listening (Dichter et al, 2016;Mesgarani et al, 2014;Chang et al, 2010; see Leonard and Chang, 2014 for a review). Mesgarani et al (2014) showed that different electrode sites across superior temporal gyrus responded to different subsets of phonemes in a manner consistent with their articulatory properties (particularly their manner of articulation).…”