Distributed representation of tone frequency in highly decodable spatio-temporal activity in the auditory cortex

Area of Science:

• Neuroscience
• Auditory System Research
• Computational Neuroscience

Background:

• The tonotopic map, representing tone frequency by place in the auditory cortex, is widely accepted.
• However, tone-evoked activation loses frequency specificity at higher sound pressure levels.
• This suggests frequency representation may involve more than a simple place code.

Purpose of the Study:

• To investigate how tone frequency is represented in spatio-temporal neural patterns within the auditory cortex.
• To elucidate the role of neural activity timing and firing rates in frequency encoding.
• To test the hypothesis of a distributed code for auditory frequency representation.

Main Methods:

• Utilized decoding-based analysis on multi-unit activities in anesthetized rat auditory cortices.
• Employed sequential dimensionality reduction (SDR), a recursive feature elimination (RFE) variant with support vector machine (SVM).
• Identified optimal spatio-temporal windows for decoding test frequencies.

Main Results:

• SDR identified optimal spatio-temporal window patterns, significantly outperforming spatial-only or high-spike-rate patterns for decoding frequency.
• Tone frequency is encoded in both temporal and spatial structures, including low-spike-rate neural windows.
• While high-spike-rate windows were more recruited, low-spike-rate windows proved crucial for fine frequency differentiation.

Conclusions:

• The auditory cortex employs a distributed code for tone frequency representation, integrating spatial and temporal neural activity.
• Both high- and low-spike-rate neural activities play complementary roles in frequency encoding and discrimination.
• This distributed coding strategy enhances the auditory system's ability to discriminate frequencies across varying sound levels.