Auditory Cortex Represents Both Pitch Judgments and the Corresponding Acoustic Cues

Jennifer K Bizley,Kerry M.M Walker,Fernando R Nodal,Andrew J King,Jan W.H Schnupp

doi:10.1016/j.cub.2013.03.003

Jennifer K Bizley, Kerry M.M Walker + Show 3 more

Open Access

https://doi.org/10.1016/j.cub.2013.03.003

Copy DOI

Abstract

SummaryThe neural processing of sensory stimuli involves a transformation of physical stimulus parameters into perceptual features, and elucidating where and how this transformation occurs is one of the ultimate aims of sensory neurophysiology. Recent studies have shown that the firing of neurons in early sensory cortex can be modulated by multisensory interactions [1–5], motor behavior [1, 3, 6, 7], and reward feedback [1, 8, 9], but it remains unclear whether neural activity is more closely tied to perception, as indicated by behavioral choice, or to the physical properties of the stimulus. We investigated which of these properties are predominantly represented in auditory cortex by recording local field potentials (LFPs) and multiunit spiking activity in ferrets while they discriminated the pitch of artificial vowels. We found that auditory cortical activity is informative both about the fundamental frequency (F0) of a target sound and also about the pitch that the animals appear to perceive given their behavioral responses. Surprisingly, although the stimulus F0 was well represented at the onset of the target sound, neural activity throughout auditory cortex frequently predicted the reported pitch better than the target F0.

Highlights

The temporal structure of the local field potentials (LFPs) reflects bottom-up sensory information, which can be modulated by cognitive processes [13,14,15]
The LFP can provide additional information to that provided by single-neuron activity
We investigated whether the characteristic frequency (CF) of a recording site influenced the likelihood of the LFP signal discriminating F0 or behavioral choice. ‘‘Near-CF’’ sites were those with a CF within an octave of the highest or lowest F0 target for the relevant testing session, and ‘‘far-CF’’ sites were those with CFs outside of this range

Summary

Introduction

ROC analysis provides a criterion-free method for estimating the discriminability between two trial types (e.g., higher or lower pitch), given the observed distributions of the recorded LFP power. The proportion of significant sites varied across analysis time windows (Figure 1F) (Kruskal-Wallis test, p < 131027), and more sites showed significant choice-related activity (red line) than significant stimulus-related activity (blue line). AaRROOCCFF00ÞÞ: The resulting CI values were positive if the neural activity was more informative about the animal’s choice than the target F0 and negative if the opposite was the case (Figure 1G).

Results

Conclusion