Researchers from MIT, led by neuroscience associate professor Evelina Fedorenko, have used an artificial language network to identify which types of sentences most effectively engage the brain’s language processing centers. The study showed that sentences of complex structure or unexpected meaning created strong responses, while straightforward or nonsensical sentences did little to engage these areas. The study results appear in Nature Human Behaviour.
The team focused on language processing regions in the brain’s left hemisphere. They started with a dataset of 1,000 sentences from various sources, read by human participants. Functional MRI was used to measure language network activity. The same sentences were then fed into a large language model similar to ChatGPT, with activation patterns measured in response to each phrase.
The team then trained an ‘encoding model’ to correlate activation patterns seen in the human brain with those observed in the artificial language model. This model was used to predict human language network response to any new sentence based on the artificial network’s response. Using this model, the team identified 500 new sentences predicted to generate maximal activity in the human brain and sentences predicted to generate minimal activity.
The research team analyzed sentences based on 11 different linguistic properties to determine what features led to more brain activity. Sentences that were novel or unusual generated higher responses in the brain, as did sentences of higher linguistic complexity. Both extremely simple sentences and overly complex, nonsensical phrases evoked little brain activation. Sentences requiring some effort to interpret were most effective at engaging language processing regions in the brain.
The study was partly funded by the MIT-IBM Watson AI Lab, the National Institutes of Health, and the McGovern Institute. The team now plans to extend the study to speakers of languages other than English, exploring the potential activation of language processing regions in the right hemisphere of the brain.
