Characterizing the dynamics of mental representations: the temporal generalization method

By Jean-Remi King, Stanislas Dehaene. Trends in Cognitive Sciences, 2014

855 citations recorded by the source corpus.

Parsing a cognitive task into a sequence of operations is a central problem in cognitive neuroscience. We argue that a major advance is now possible owing to the application of pattern classifiers to time-resolved recordings of brain activity [electroencephalography (EEG), magnetoencephalography (MEG), or intracranial recordings]. By testing at which moment a specific mental content becomes decodable in brain activity, we can characterize the time course of cognitive codes. Most importantly, the manner in which the trained classifiers generalize across time, and from one experimental condition to another, sheds light on the temporal organization of information-processing stages. A repertoire of canonical dynamical patterns is observed across various experiments and brain regions. This method thus provides a novel way to understand how mental representations are manipulated and transformed.

Subfields: Artificial Intelligence, Cognitive Neuroscience

View Paper: https://www.semanticscholar.org/paper/ecc80f66bbd6b0df5194450c70e7cae4cd93dc70

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Cognitive Neuroscience

Citations in the corpus

Under the Hood: Using Diagnostic Classifiers to Investigate and Improve how Language Models Track Agreement Information (2018, background): “With a total of T timesteps, this gives us T×T DC-accuracies that together constitute a Temporal Generalization Matrix (King and Dehaene, 2014; Fyshe et al., 2016).”
Decoding Dynamic Brain Patterns from Evoked Responses: A Tutorial on Multivariate Pattern Analysis Applied to Time Series Neuroimaging Data (2016, methodology): “…of “ brain decoding ” methods to the analysis of fMRI data has been highly influential over the past 15 years in the field of cognitive neuroscience (Kamitani & Tong, 2005; Carlson, Schrater, & He, 2003; Cox & Savoy, 2003; Haxby et al., 2001; Edelman, Grill-Spector, Kushnir, & Malach, 1998).”
A statistical framework for neuroimaging data analysis based on mutual information estimated via a gaussian copula (2016, methodology): “The temporal generalization method [King and Dehaene, 2014], in which a classiﬁcation algorithm is trained with neural responses from one time point and then tested on another time point, can reveal similar representations between the two time points.”
Pleasure systems in the brain (2015, background): “It remains possible that more fine-grained spatial and temporal multivariate pattern analysis techniques (Haynes and Rees, 2006; King and Dehaene, 2014) will identify subsets of limbic neural circuitry particular to just one type of reward (Chikazoe et al., 2014).”
The emergence of number and syntax units in LSTM language models (2019, positive): “matical number of the subject from network activity in response to the presentation of the subject, and tested its prediction on test sets from all time points (King and Dehaene, 2014), in incongruent conditions only of the nounPP task.”
Deconstructing multivariate decoding for the study of brain function (2017, background): “On neurophysiological data, it has become common to train a classifier at one point in time and test it at another to see whether it can generalize across time (King and Dehaene, 2014).”
Open multi-center intracranial electroencephalography dataset with task probing conscious visual perception (2025, unknown): “Cross-temporal generalization 49,50 for decoding faces vs. objects employs support vector machines and 5-fold cross-validation.”
Distractor anticipation during working memory is associated with theta and beta oscillations across spatial scales (2024, unknown): “To test similarity in oscillatory activity patterns across time bins, we employed cross-temporal generalization (King and Dehaene, 2014; Stokes, 2015; Grootswagers et al., 2017) (Figure 3).”
Prioritizing flexible working memory representations through retrospective attentional strengthening (2023, unknown): “We calculated the temporal generation of color and space through decoding analysis (King and Dehaene, 2014).”
Tracking neural correlates of prioritizing working memory representations through retrospective attentional strengthening (2022, methodology): “We calculated the 1 temporal generation of color and space through decoding analysis (65).”