Spatiotemporally resolved multivariate pattern analysis for M/EEG. Issue 10 (18th March 2022)

• Record Type: Journal Article
• Authors: Higgins, Cameron; Vidaurre, Diego; Kolling, Nils; Liu, Yunzhe; Behrens, Tim; Woolrich, Mark
• Is Part Of: Human brain mapping. Volume 43:Issue 10(2022)
• Page Start: 3062
• Page End: 3085
• Publication Date: 2022-03-18
• Subjects: decoding -- EEG -- encoding -- MEG -- single trial task dynamics; Brain mapping -- Periodicals; 611.81
• DOI: 10.1002/hbm.25835
• Languages: English
• ISSNs: 1065-9471

Abstract

Abstract: An emerging goal in neuroscience is tracking what information is represented in brain activity over time as a participant completes some task. While electroencephalography (EEG) and magnetoencephalography (MEG) offer millisecond temporal resolution of how activity patterns emerge and evolve, standard decoding methods present significant barriers to interpretability as they obscure the underlying spatial and temporal activity patterns. We instead propose the use of a generative encoding model framework that simultaneously infers the multivariate spatial patterns of activity and the variable timing at which these patterns emerge on individual trials. An encoding model inversion maps from these parameters to the equivalent decoding model, allowing predictions to be made about unseen test data in the same way as in standard decoding methodology. These SpatioTemporally Resolved MVPA (STRM) models can be flexibly applied to a wide variety of experimental paradigms, including classification and regression tasks. We show that these models provide insightful maps of the activity driving predictive accuracy metrics; demonstrate behaviourally meaningful variation in the timing of pattern emergence on individual trials; and achieve predictive accuracies that are either equivalent or surpass those achieved by more widely used methods. This provides a new avenue for investigating the brain's representational dynamics and could ultimately support more flexible experimental designs in the future. Abstract : Neuroscientists use the high temporal resolution of recording modalities like MEG and EEG to ask questions about when the brain processes different features of a stimulus. For example, a simple visual stimulus contains information about colour, shape, and form—but also about its intrinsic value, familiarity, even danger to the observer. Understanding the timecourse over which these different traits are expressed in neural data tells us about how the brain breaks down information in order to best serve our behavioural goals. In recent years, multivariate pattern analysis methods have become a popular means of characterising these timecourses, however they normally obscure the spatial basis of the decoded signal while requiring the information processing to be perfectly synchronised in time over all trials. We instead propose a new approach, whereby we explicitly model the multivariate distribution associated with experimental stimuli simultaneously over space and time. This allows us to characterise where in the brain a signal originates from, alongside when it emerges in each individual trial. This opens up a whole new way of analysing brain activity, exploring how the timecourse of information processing is modulated by behaviour and physiology across distinct parts of the brain.