The human primary somatosensory cortex response contains components related to stimulus frequency and perception in a frequency discrimination task

L.c Liu, P. B. C. Fenwick, Nikolaos Laskaris, Marc Schellens, Vahe Poghosyan, Tadahiko Shibata, Andreas A. IoannidesNeuroscience 2003 · DOI

TL;DR

This study uses magnetoencephalography (MEG) to investigate the role of the human primary somatosensory cortex (SI) in discriminating stimulus frequencies. They found that SI activity not only reflects the physical properties of the stimulus but also plays a role in perceptual decisions, with distinct temporal patterns emerging for stimulus analysis versus perception and decision-making.

What Problem Does This Paper Address?

Understanding how the brain processes sensory information and makes decisions is a fundamental challenge. While animal studies have shown that the primary somatosensory cortex (SI) is involved in processing tactile stimuli and even in discrimination tasks, it's been difficult to determine if human SI passively reflects input or actively participates in comparison and decision-making. This is partly because direct experimental manipulation of SI in humans, as done in animal models, is not feasible. This paper aims to bridge this gap by using non-invasive MEG to investigate the temporal dynamics of human SI activity during a frequency discrimination task, correlating brain responses with subjective perception.

Methods

The researchers employed magnetoencephalography (MEG) to record brain activity from four participants performing a frequency discrimination task. The core of their methodology involved:

  • Single-Trial Analysis: Instead of averaging MEG signals, they analyzed each trial individually. This is crucial for understanding how brain activity relates to a specific perception or decision on a given trial, especially when subjects might use different strategies.
  • Magnetic Field Tomography (MFT): This technique was used to estimate the 3D distribution of electrical activity in the brain from the MEG signals. It allowed them to pinpoint the location and temporal evolution of neural activity.
  • Region of Interest (ROI) Analysis: A specific region in the left SI (hand area) was identified as the primary focus for analysis.
  • Signal-to-Noise Ratio (SNR) Measurement: This was used to quantify the consistency of the neural response across trials within specific frequency bands (particularly 18-30 Hz), indicating how well the brain activity reflected the stimulus.
  • Rayleigh Quotient (RQ) Analysis: This statistical method was used to quantify the separability between different groups of single-trial activation curves (acv-curves). High RQ values indicate clear differences in neural responses between groups, suggesting distinct processing.
  • Multidimensional Scaling (MDS): This dimensionality reduction technique was used to visualize the structure of the acv-curves and make the discriminative analysis more robust. It helps to understand how different trials cluster together based on their neural patterns.

Datasets

The study used data collected from four healthy, right-handed male subjects. No external datasets were used; the data was generated specifically for this experiment. The experiment involved approximately 600 pairs of stimuli per subject, with 180 trials recorded for MEG analysis.

Key Insights and Ideas

  • SI is more than a passive receiver: The study provides strong evidence that human SI is not just passively reflecting sensory input but actively participates in processing and decision-making related to tactile frequency discrimination.
  • Temporal Segregation of Functions: The SI response exhibits distinct temporal phases. An early phase (within the first 600 ms of Stim2) appears to be primarily involved in analyzing the stimulus frequency itself, while a later phase (beyond 650 ms) correlates more strongly with the subject's perception and decision.
  • Frequency-Specific Signatures: The 18-30 Hz frequency band of the MEG signal was found to carry the most information about the stimulus frequency, highlighting its importance in tactile processing.
  • Perception-Driven Differences Emerge Later: When comparing correct responses, the neural differences between perceived frequencies become more pronounced and organized later in the Stim2 epoch, suggesting that perception and decision processes unfold over time.
  • Individual Variability, Common Principles: While subjects showed some variability in their strategies and performance, the underlying temporal patterns of SI involvement in frequency analysis and perception/decision were consistent across participants.

Key Results

  • Early SI Activation: The left SI showed its first activation peak between 35-48 ms after Stim1 onset, with sustained activity throughout both stimulus periods.
  • Frequency Discrimination Begins Early: Differences in SI activation between Stim2 frequencies (21 Hz vs. 26-29 Hz) emerged within 100 ms of Stim2 onset and were sustained.
  • Perception-Related Differences Enhance Later: When analyzing only correct responses, the differences in SI activity became even more pronounced at later latencies (around 650 ms).
  • Distinct Temporal Patterns for Perception vs. Analysis: For one subject, the analysis of trials where the perceived frequency differed from the input frequency showed sustained differences only before 650 ms, suggesting a shift in SI's role from stimulus analysis to perception/decision.
  • High SNR in 18-30 Hz Band: This frequency band consistently showed high SNR values during both Stim1 and Stim2 periods, indicating its crucial role in processing tactile frequency information.

Limitations and Future Work

  • Limited Number of Subjects: The study involved only four subjects, which limits the generalizability of findings.
  • Subject Fatigue: The 4-hour experimental duration may have led to subject fatigue, potentially affecting performance, especially in later blocks.
  • Individual Strategy Variability: Subjects employed different strategies for discrimination, which, while accounted for by single-trial analysis, adds complexity.
  • Future Work: The authors suggest exploring other brain areas that might be involved in the decision-making process alongside SI, such as the secondary somatosensory cortex (SII), prefrontal cortex, and medial premotor cortex. They also propose that their single-trial SI findings provide a foundation for investigating these interconnected networks.

Summary updated

BibTeX
@inproceedings{liu2003human,
  author = {L.C Liu and P. B. C. Fenwick and Nikolaos Laskaris and Marc Schellens and Vahe Poghosyan and Tadahiko Shibata and Andreas A. Ioannides},
  title = {{The human primary somatosensory cortex response contains components related to stimulus frequency and perception in a frequency discrimination task}},
  booktitle = {Neuroscience},
  year = {2003},
  doi = {10.1016/s0306-4522(03)00353-1},
}
Figure