Transformation of a temporal speech cue to a spatial neural code in human auditory cortex.

Area of Science:

• Neuroscience
• Auditory Perception
• Speech Processing

Background:

• Listeners perceive speech by extracting spectrotemporal cues from acoustic signals.
• Spectral speech cues are encoded by the amplitude of neural responses in the auditory cortex.
• The neural encoding of temporal speech cues, such as voice-onset time (VOT), remains unclear.

Purpose of the Study:

• To investigate the neurophysiological mechanisms underlying the cortical representation of temporal speech cues.
• To determine if a spatial/amplitude neural code, similar to that for spectral cues, also encodes VOT.

Main Methods:

• Direct human brain recordings were used to analyze neural responses to a voice-onset time (VOT) continuum (/ba/ to /pa/).
• A simple neural network model was employed to simulate neural population responses to temporal and spectral cues.

Main Results:

• Distinct neural populations showed preferential responses to specific VOTs within phonetic categories.
• These neural populations were also sensitive to sub-phonetic VOT variations.
• The neural network model successfully replicated the observed encoding patterns, demonstrating a spatial/amplitude code.

Conclusions:

• A spatial/amplitude neural code underlies the cortical representation of both spectral and temporal speech cues.
• This unified coding mechanism explains the brain's ability to distinguish phonetic categories based on temporal information like VOT.