Representational similarity analysis reveals commonalities and differences in the semantic processing of words and...

Area of Science:

• Neuroscience
• Cognitive Science
• Psycholinguistics

Background:

• Understanding word and object meanings relies on conceptual representations.
• It's debated whether these semantic representations are modality-independent (amodal) or modality-specific.
• This study investigates the neural basis of semantic processing across different input types.

Purpose of the Study:

• To determine if semantic representations are processed in a modality-independent manner.
• To identify brain regions involved in processing concepts from written words and pictures.
• To explore differences in representational content across brain regions and modalities.

Main Methods:

• Human fMRI study examining neural processing of written words and pictures.
• Searchlight-based multivoxel pattern analysis (MVPA) on activation maps.
• Representational similarity analysis (RSA) and novel data-driven analyses (k-means clustering, seeded correlation).

Main Results:

• Common semantic category effects for words and pictures found in a left-lateralized network (LpMTG, angular gyrus, LIPS).
• Object-specific processing in ventral temporal cortex; word-specific processing in anterior MTG.
• Representations in LIPS were modality-invariant, while LpMTG representations differed between words and pictures.

Conclusions:

• Both LpMTG and LIPS are crucial for semantic processing.
• LIPS exhibits modality-independent semantic processing.
• LpMTG's functional role in semantic processing varies depending on the input modality (word vs. object).