Decoding the meaning of unconsciously processed words using fMRI-based MVPA

Area of Science:

• Neuroscience
• Cognitive Science
• Psycholinguistics

Background:

• The human brain's capacity for non-conscious semantic processing remains an active area of research.
• Understanding how semantic representations are formed and generalized, particularly across languages, is crucial for cognitive neuroscience.

Purpose of the Study:

• To investigate whether the human brain exhibits activity patterns related to word meaning in the absence of conscious awareness.
• To determine if these non-conscious semantic representations generalize across different languages using fMRI.

Main Methods:

• Employed fMRI-based multivariate pattern analysis (MVPA) with a masked word paradigm.
• Presented visual stimuli of animal and non-animal words in Spanish and Basque, briefly and masked.
• Utilized a support vector machine (SVM) to decode word category from BOLD responses in predefined semantic regions.

Main Results:

• Successfully decoded the semantic category of non-conscious words from BOLD response patterns.
• Identified specific brain regions (e.g., inferior parietal lobe, prefrontal cortex) showing discriminative patterns for Spanish and Basque words.
• Observed no evidence of across-language generalization for non-conscious semantic representations.

Conclusions:

• Demonstrates that brain areas associated with semantic processing contain information about word meaning, even without conscious awareness or behavioral detection.
• Suggests that conscious semantic access is necessary for the generalization of semantic representations across languages.