Canonical template tracking: Measuring the activation state of specific neural representations

Abstract

Multivariate analyses of neural data have become increasingly influential in cognitive neuroscience since they allow to address questions about the representational signatures of neurocognitive phenomena. Here, we describe Canonical Template Tracking: a multivariate approach that employs independent localizer tasks to assess the activation state of specific representations during the execution of cognitive paradigms. We illustrate the benefits of this methodology in characterizing the particular content and format of task-induced representations, comparing it with standard (cross-)decoding and representational similarity analyses. Then, we discuss relevant design decisions for experiments using this analysis approach, focusing on the nature of the localizer tasks from which the canonical templates are derived. We further provide a step-by-step tutorial of this method, stressing the relevant analysis choices for functional magnetic resonance imaging and magneto/electroencephalography data. Importantly, we point out the potential pitfalls linked to canonical template tracking implementation and interpretation of the results, together with recommendations to mitigate them. To conclude, we provide some examples from previous literature that highlight the potential of this analysis to address relevant theoretical questions in cognitive neuroscience.

Proactive selective attention across competition contexts

“…Analysis of the overlap between activity patterns from the perceptual localizer and preparation interval of the main task allowed to examine whether the nature of the category-specific anticipation was similar to perceptually driven patterns (Kaplan et al, 2015; Palenciano et al, 2023). Results showed similarities between perceptual and preparatory category patterns in high competition.…”

Concurrent response and action effect representations across the somatomotor cortices during novel task preparation

“…Having said that, the consistent results linked to the irrelevant templates suggest that this information was somehow anticipated. Although it could be tempting to interpret this finding as the suppression of irrelevant category representations during preparation, the similarity measurements computed (i.e., non-cross-validated; Walther et al, 2016) require that relative instead of absolute activation estimates are considered (Palenciano et al, 2022; Walther et al, 2016). Hence, no inference should be performed based only on the irrelevant categories’ below-zero representational strength.…”

“…The main goal of this study was to assess whether the motoric and visual representations guiding overt performance and perception were covertly engaged during novel mapping preparation. To do so, we employed a canonical template tracking procedure, a pattern analysis technique that enables the comparison of neural data between a targeted cognitive paradigm and theoretically driven localizer tasks (Palenciano et al, 2023; for other examples, see González-García et al, 2021; Wimber et al, 2015). Specifically, we estimated a series of canonical neural representations (or templates) from each localizer data.…”

“…The experimental design sought to eliminate such confound by balancing each localizer’s set of templates (e.g., the four response effectors’ templates) between the relevant and irrelevant conditions. Nonetheless, we computed two reliability measurements per template (González-García et al, 2021; Palenciano et al, 2023). First, to estimate the overall univariate noise, we computed the Signal-to-Noise Ratio (SNR), dividing the absolute mean t value from a given template by its standard deviation.…”

“…The copyright holder for this preprint this version posted November 2, 2023. ; https://doi.org/10.1101/2023.10.30.564708 doi: bioRxiv preprint paradigm while keeping the remaining task features as independent as possible. This was done to avoid confounding specific template activation with more general motoric or perceptual processes (Palenciano et al, 2023). Each localizer block lasted approximately 5 minutes, and the order in which they were presented was counterbalanced across participants.…”