Trial-specific Dynamics in Task Responses
We are developing machine learning approaches (e.g. using adapted Hidden Markov Models) that can estimate trial-specific temporal dynamics of cognitive processing with high temporal precision in electrophysiological data (e.g. M/EEG). These can reveal dynamics at much higher temporal resolution than is possible with traditional, trial-averaging methods.
We are using this to make possible:
- the analysis of naturalistic stimuli, where the exact timing and duration of cognition varies wildly on a trial-to-trial basis.
- temporal unconstrained encoding/decoding, which reveals trial-specific sequences of task-related activity.
- the inference of task-related large-scale dynamic brain network activity.
- the identification of transient spectral (e.g. beta) bursts.
Tools for doing this kind of analysis (HMMMAR) are available to download and use at the OHBA Analysis Group Software Page.
Quinn AJ, Vidaurre D, Abeysuriya R, Becker R, Nobre AC, Woolrich MW. Task-Evoked Dynamic Network Analysis Through Hidden Markov Modeling. Front Neurosci. 2018
Vidaurre D, Myers N, Stokes M, Nobre AC, Woolrich M. Temporally unconstrained decoding reveals consistent but time-varying stages of stimulus processing. Cerebral Cortex. 2018.
van Ede F, Quinn AJ, Woolrich MW, Nobre AC. Neural Oscillations: Sustained Rhythms or Transient Burst-Events?
Trends Neurosci. 2018