When Abstract Becomes Concrete: Naturalistic Encoding of Concepts in the Brain DOI Open Access
Viktor Kewenig, Gabriella Vigliocco, Jeremy I Skipper

et al.

Published: Sept. 17, 2024

Language is acquired and processed in complex dynamic naturalistic contexts, involving simultaneous processing of connected speech, faces, bodies, objects, etc.. How words their associated concepts are encoded the brain during real-world still unknown. Here, representational structure concrete abstract was investigated movie watching to address extent which responses dynamically change depending on visual context. First, across shown encode different experience-based information separable sets regions. However, these differences reduced when multimodal context considered. Specifically, response profile becomes more concrete-like scenes highly related meaning. Conversely, unrelated a given word, activation pattern resembles that conceptual processing. These results suggest while generally habitual experiences, underlying neurobiological organisation not fixed but depends available contextual information.

Language: Английский

Computational Language Modeling and the Promise of In Silico Experimentation DOI Creative Commons
Shailee Jain, Vy A. Vo, Leila Wehbe

et al.

Neurobiology of Language, Journal Year: 2023, Volume and Issue: 5(1), P. 80 - 106

Published: Jan. 24, 2023

Abstract Language neuroscience currently relies on two major experimental paradigms: controlled experiments using carefully hand-designed stimuli, and natural stimulus experiments. These approaches have complementary advantages which allow them to address distinct aspects of the neurobiology language, but each approach also comes with drawbacks. Here we discuss a third paradigm—in silico experimentation deep learning-based encoding models—that has been enabled by recent advances in cognitive computational neuroscience. This paradigm promises combine interpretability generalizability broad scope We show four examples simulating language then both caveats this approach.

Language: Английский

Citations

27
Brain-model neural similarity reveals abstractive summarization performance DOI Creative Commons
Zhejun Zhang, S.-L. Guo,

W H Zhou

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: Jan. 2, 2025

Language: Английский

Citations

1
Representational maps in the brain: concepts, approaches, and applications DOI Creative Commons

Takahiro Noda,

Dominik F. Aschauer, Anna R. Chambers

et al.

Frontiers in Cellular Neuroscience, Journal Year: 2024, Volume and Issue: 18

Published: March 22, 2024

Neural systems have evolved to process sensory stimuli in a way that allows for efficient and adaptive behavior complex environment. Recent technological advances enable us investigate processing animal models by simultaneously recording the activity of large populations neurons with single-cell resolution, yielding high-dimensional datasets. In this review, we discuss concepts approaches assessing population-level representation form representational map. such map, not only are identities distinctly represented, but their relational similarity is also mapped onto space neuronal activity. We highlight example studies which structure maps brain estimated from recordings humans as well animals compare methodological approaches. Finally, integrate these aspects provide an outlook how concept could be applied various fields basic clinical neuroscience.

Language: Английский

Citations

3
Narrative 'twist' shifts within-individual neural representations of dissociable story features DOI Creative Commons
Clara Sava‐Segal, Clare Grall, Emily S. Finn

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 13, 2025

Abstract Given the same external input, one’s understanding of that input can differ based on internal contextual knowledge. Where and how does brain represent latent belief frameworks interact with incoming sensory information to shape subjective interpretations? In this study, participants listened auditory narrative twice, a plot twist in middle dramatically shifted their interpretations story. Using robust within-subject whole-brain approach, we leveraged shifts neural activity between two listens identify where are represented brain. We considered terms its hierarchical structure, examining global situation models subcomponents–namely, episodes characters–are represented, finding they rely partially distinct sets regions. Results suggest our brains narratives hierarchically, individual elements being dynamically updated as part changing information.

Language: Английский

Citations

0
The Voxelwise Encoding Model framework: a tutorial introduction to fitting encoding models to fMRI data DOI Creative Commons
Tom Dupré la Tour, Matteo Visconti di Oleggio Castello, Jack L. Gallant

et al.

Imaging Neuroscience, Journal Year: 2025, Volume and Issue: 3

Published: Jan. 1, 2025

Abstract The Voxelwise Encoding Model framework (VEM) is a powerful approach for functional brain mapping. In the VEM framework, features are extracted from stimulus (or task) and used in an encoding model to predict activity. If able activity some part of brain, then one may conclude that information represented also encoded brain. VEM, separate fitted on each spatial sample (i.e., voxel). has many benefits compared other methods analyzing modeling neuroimaging data. Most importantly, can use large numbers simultaneously, which enables analysis complex naturalistic stimuli tasks. Therefore, produce high-dimensional maps reflect selectivity voxel features. Moreover, because performance estimated test dataset not during fitting, minimizes overfitting inflated Type I error confounds plague approaches, results generalize new subjects stimuli. Despite these benefits, still widely neuroimaging, partly no tutorials this method available currently. To demystify ease its dissemination, paper presents series hands-on accessible novice practitioners. based free open-source tools public datasets, reproduce presented previously published work.

Language: Английский

Citations

0
The cortical representation of language timescales is shared between reading and listening DOI Creative Commons
Catherine Chen, Tom Dupré la Tour, Jack L. Gallant

et al.

Communications Biology, Journal Year: 2024, Volume and Issue: 7(1)

Published: March 7, 2024

Abstract Language comprehension involves integrating low-level sensory inputs into a hierarchy of increasingly high-level features. Prior work studied brain representations different levels the language hierarchy, but has not determined whether these are shared between written and spoken language. To address this issue, we analyze fMRI BOLD data that were recorded while participants read listened to same narratives in each modality. Levels operationalized as timescales, where timescale refers set spectral components stimulus. Voxelwise encoding models used determine timescales represented across cerebral cortex, for modality separately. These reveal two modalities organized similarly cortical surface. Our results suggest that, after processing, integration proceeds regardless stimulus

Language: Английский

Citations

2
Trials and tribulations when attempting to decode semantic representations from MEG responses to written text DOI Creative Commons
Գայանե Ղազարյան, Marijn van Vliet,

Aino Saranpää

et al.

Language Cognition and Neuroscience, Journal Year: 2023, Volume and Issue: 39(9), P. 1149 - 1160

Published: June 14, 2023

Several studies have been published which show that it is possible to decode semantic representations directly from brain responses. This has repeatedly successful when the stimuli used were pictures of objects. However, there a distinct scarcity decoding responses orthographic stimuli, particularly those employing time-sensitive imaging methods. We use examples our own research highlight challenges we faced attempting MEG written words. discuss differences in and determine characteristics allow for semantics. suspect limited number on this topic indicates these are not unique experience. By bringing attention issues, hope stimulate new wave discussion leading eventual progress.

Language: Английский

Citations

5
The Cortical Representation of Language Timescales is Shared between Reading and Listening DOI Creative Commons
Catherine Chen, Tom Dupré la Tour, Jack L. Gallant

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 6, 2023

Abstract Language comprehension involves integrating low-level sensory inputs into a hierarchy of increasingly high-level features. Prior work studied brain representations different levels the language hierarchy, but has not determined whether these are shared between written and spoken language. To address this issue, we analyzed fMRI BOLD data recorded while participants read listened to same narratives in each modality. Levels were operationalized as timescales , where timescale refers set spectral components stimulus. Voxelwise encoding models used determine represented across cerebral cortex, for modality separately. These reveal that two modalities organized similarly cortical surface. Our results suggest that, after processing, integration proceeds regardless stimulus

Language: Английский

Citations

4
A low-activity cortical network selectively encodes syntax DOI Creative Commons
Adam Milton Morgan, Orrin Devinsky, Werner Doyle

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: June 20, 2024

Abstract Syntax, the abstract structure of language, is a hallmark human cognition. Despite its importance, neural underpinnings remain obscured by inherent limitations non-invasive brain measures and near total focus on comprehension paradigms. Here, we address these with high-resolution neurosurgical recordings (electrocorticography) controlled sentence production experiment. We uncover three syntactic networks that are broadly distributed across traditional language regions, but focal concentrations in middle inferior frontal gyri. In contrast to previous findings from studies, process syntax mostly exclusion words meaning, supporting cognitive architecture distinct system. Most strikingly, our data reveal an unexpected property syntax: it encoded independent activity levels. propose this “low-activity coding” scheme represents novel mechanism for encoding information, reserved higher-order cognition more broadly.

Language: Английский

Citations

1
Bilingual language processing relies on shared semantic representations that are modulated by each language DOI Creative Commons
Catherine Chen, Xue L. Gong,

Christine Tseng

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: June 28, 2024

Abstract Billions of people throughout the world are bilingual and can extract meaning from multiple languages. While some evidence suggests that there is a shared system in human brain for processing semantic information different languages, other to degree distinct between We conducted study determine how representations brains bilinguals support both Functional magnetic resonance imaging (fMRI) was used record responses while participants who fluent English Chinese read several hours natural narratives each language. These data were then specifically comprehensively compare two First, we show largely Second, finer-grained differences systematically alter same represented Our results suggest bilinguals, across languages but modulated by reconcile current competing theories language processing.

Language: Английский

Citations

1