From perceiving words to reading: Neural multivariate representations of sublexical vs. lexico-semantic processing...

Area of Science:

• Neurobiology of Language
• Cognitive Neuroscience
• Neuroimaging

Background:

• Neural basis of semantic cognition is well-studied, but extracting meaning from visual language input remains unclear.
• Previous research used psycholinguistic properties and Natural Language Processing (NLP) models, lacking direct comparison of sublexical and lexico-semantic processing.
• Understanding the interaction between sublexical (phonological, orthographic) and lexico-semantic systems is crucial for the neurobiology of language.

Purpose of the Study:

• To investigate the distinct neural representations of sublexical and lexico-semantic information during visual word processing.
• To compare brain activity patterns associated with psycholinguistic properties, NLP models, and their relationship to semantic and sublexical features.
• To elucidate how sublexical and lexico-semantic systems interact within the brain.

Main Methods:

• Functional Magnetic Resonance Imaging (fMRI) study with 30 participants performing a lexical decision task.
• Careful control of sublexical (phonological, orthographic) and lexico-semantic properties (word frequency, familiarity, concreteness).
• Representational Similarity Analysis (RSA) comparing multivariate brain patterns with models of sublexical, semantic, and NLP (word vector) relations.

Main Results:

• Sublexical and lexico-semantic representations were found to recruit different brain areas within the left inferior frontal gyrus (IFG) and ventral occipitotemporal cortex (vOTC).
• Anterior IFG and vOTC showed representations of semantic models, whereas posterior IFG (e.g., SMA) and posterior vOTC (e.g., V3-V4) represented sublexical models.
• Both semantic and NLP models converged in key semantic hubs, including the inferior anterior temporal lobe (ATL), parahippocampal gyrus, and anterior IFG.

Conclusions:

• The findings delineate distinct neural pathways for processing word form versus word meaning in the human brain.
• This research provides critical insights into the neurobiological mechanisms underlying visual language comprehension.
• The convergence of semantic and NLP models in specific brain regions highlights their role as semantic hubs.