A new study by neuroscientists at MIT has uncovered what kind of sentences are most likely to stimulate the brain’s main language processing centers. Utilizing an artificial language network, the researchers discovered that sentences with unusual grammar or unexpected meanings produce stronger responses in these areas; while straightforward sentences or nonsensical word sequences hardly engage these regions at all. The study is published today in Nature Human Behavior.
The research team focused their work on language-processing centers located in the left hemisphere of the brain, which encompasses Broca’s area and other parts of the left frontal and temporal lobes. They compiled a set of 1,000 sentences from diverse sources and had five human participants read each sentence while their brain activity was measured using functional magnetic resonance imaging (fMRI). The same sentences were then analyzed with a large language model similar to ChatGPT.
Following this, the team trained an “encoding model” that matched the observed activation patterns in the human brain with those from the artificial language model. This allowed them to predict how the human language network would react to any new sentence based on the artificial language network’s response to these original sentences.
Using the encoding model, the researchers identified 500 new sentences designed to stimulate maximum brain activity, and others to elicit minimal activity. They then tested these sentences on a new group of three human participants and found the sentences effectively drove and suppressed brain activity as predicted. This novel method of modulating brain activity during language processing was hailed as a significant first in areas implicated in higher-level cognition.
To understand more about why certain sentences stimulated more brain activity than others, the researchers examined the sentences based on 11 different linguistic factors, including grammar, emotional valence, how easily the sentence could be visually constructed, and the sentence’s ‘surprisal’ — or comparative rarity among other sentences. The team discovered that sentences with higher surprisal and complexity produce increased brain responses. Meanwhile, exceptionally simple sentences or those so complex they made no sense evoked little response from the language network.
Senior author Evelina Fedorenko noted: “The sentences that elicit the highest brain response have a weird grammatical thing and/or a weird meaning. There’s something slightly unusual about these sentences”. The team now plans on testing whether these findings remain consistent in speakers of other languages and exploring what might trigger language processing in the brain’s right hemisphere.
