Post-Task Responses Following Working Memory and Movement Are Driven by Transient Spectral Bursts with Similar Characteristics

Neurology Optics Physics Psychiatry Biophysics Medical Imaging Life Sciences Medicine
spectral bursts hidden Markov model magnetoencephalography neural oscillations post-movement beta rebound

Background and Research Questions

Working memory and post-task responses (PTRs) in the brain have been a focal point in neuroscience research. Previous studies have shown that post-movement beta rebound (PMBR) is a reliable and stable phenomenon in the cerebral cortex that can be studied and measured using magnetoencephalography (MEG). Recent research further reveals that PTRs are not limited to post-movement beta rebound but are widespread across various frequency bands (such as θ, α, and β bands) and brain regions. However, it remains unclear whether these post-working memory PTRs are driven by transient high-amplitude spectral bursts similar to PMBR. The main objective of this paper is to explore whether the driving mechanisms of PTRs are similar across working memory and visual motion tasks by comparing datasets from these two types of tasks.

Source of the Paper and Author Information

This research paper was collaboratively written by scholars from several institutions, including the University of Nottingham, Young Epilepsy, Oxford Centre for Human Brain Activity, Wellcome Centre for Integrative Neuroimaging, and University of Birmingham. The paper was published in the journal “Human Brain Mapping,” accepted on April 14, 2024. The primary authors include Sebastian C. Coleman, Zelekha A. Seedat, Daisie O. Pakenham, Andrew J. Quinn, Matthew J. Brookes, Mark W. Woolrich, and Karen J. Mullinger.

Research Methods and Procedures

The study used two different task paradigms to collect data: the n-back working memory task and the handgrip visual motion task. Below is a detailed process of data collection, preprocessing, HMMs analysis, and results comparison for each task.

n-back Task

  1. Task Paradigm: The task comprises three conditions: 0-back, 1-back, and 2-back, representing different working memory loads. Each task period lasts for 30 seconds, followed by a 1-second pause where a target letter is displayed, requiring participants to press a button upon seeing the target letter.
  2. Data Collection: 20 healthy volunteers participated in the test, and data were collected using a 275-channel CTF MEG system configured in third-order gradiometer mode, with a data sampling rate of 600Hz.
  3. Preprocessing: Data were analyzed using FieldTrip toolbox for ICA analysis to remove eye blink and cardiac artifacts. After removing bad data segments, 2960 data segments were retained for further analysis.

Handgrip Visual Motion Task

  1. Task Paradigm: The task included grip operations of 2 seconds, 5 seconds, and 10 seconds with a 30-second rest period after each operation. Volunteers used a grip bar to maintain 30% of their maximum voluntary force.
  2. Data Collection: 15 healthy volunteers participated in the test, with a data sampling rate identical to the n-back task, at 600Hz.
  3. Preprocessing: Similar to the n-back task, ICA analysis was used for artifact removal, resulting in 1050 data segments.

The study also utilized multivariate Hidden Markov Models (HMMs) to identify bursting states driving PTRs, providing a way to identify burst events without prior knowledge of frequency content or response time.

Main Research Results

Comparison of PTRs in Working Memory and Visual Motion Tasks

  1. Temporal Comparison of Burst Features: Comparison of time-frequency responses from single-trial MEG data showed that PTR states correspond to transient high-power activity, primarily in the α band, but also appearing in θ and β bands across tasks.
  2. Burst Duration: Burst duration exhibited similar variations within different brain regions for both tasks, showing significant correlation (r² = 0.56). Longer bursts appeared in the motor cortex, while shorter bursts appeared in the prefrontal cortex.
  3. Spectral Content Comparison: Power spectral density (PSD) distributions for each state were extracted using the multitaper method. Both tasks showed strong correlations in α and β band power across brain regions, but relative power in the θ band displayed greater variability.

Application of HMMs in MEG Data

Extracting PTR states through HMMs not only enhanced signal isolation but also significantly improved correlation with behavioral measures such as reaction time. This result supports the hypothesis that PTRs are related to behavioral indices of task difficulty.

Conclusion and Significance

The study indicates that PTRs following working memory and motor-related tasks are driven by transient bursting events with similar characteristics. These findings suggest that the frequency and duration of bursts reflect the emission regions and the recruited network architecture rather than functional changes in PTRs across different tasks. The results also demonstrate the effectiveness of HMMs in isolating signals of interest from MEG data and enhancing behavioral correlations.

This research provides crucial insights into the fundamental neural mechanisms of PTRs and proposes new research directions for neural processing based on transient bursts, with potential applications in brain disease diagnosis and neurorehabilitation.

Research Highlights

  • This study is the first to systematically compare the spectral characteristics and durations of PTRs following working memory and motor tasks, confirming the similar transient bursting driving mechanism of these PTRs.
  • The usage of HMMs analysis method not only improved signal accuracy but also enhanced correlations with behavioral indices, highlighting its potential application value in neuroscience research.