Complex, unfamiliar sentences make the brain’s language network work harder, study finds

With the help of an artificial language network, MIT neuroscientists have discovered which types of sentences are most likely to trigger the brain’s key language-processing centers.

The new study finds that more complex sentences, either due to unusual grammar or unexpected meaning, generate stronger responses in these language processing centers. Very simple sentences barely engage these regions, and nonsense word sequences don’t do much for them either.

For example, researchers found that this brain network was more active when reading unusual sentences such as “Buy and sell signals remain peculiar,” taken from a publicly available linguistic dataset called C4. However, it would get quiet when playing something very simple, like “We were sitting on the couch.”

“The input has to be language-like enough to engage the system,” says Evelina Fedorenko, an associate professor of neuroscience at MIT and a member of MIT’s McGovern Institute for Brain Research. “And then, in that space, if things are really easy to process, then you don’t have much of an answer. But if things get difficult or surprising, if there’s an unusual construction or an unusual set of words what you are doing.” maybe not very familiar, then the network has to work harder.

Fedorenko is the lead author of the study, which appears today in Human behavior. Greta Tuckute, an MIT graduate student, is the lead author of the paper.

Processing language

In this study, researchers focused on language processing regions found in the left hemisphere of the brain, which includes Broca’s area as well as other parts of the left frontal and temporal lobes of the brain.

“This language network is very selective based on language, but it’s more difficult to understand what’s happening in these language regions,” says Tuckute. “We wanted to find out what types of sentences, what types of linguistic inputs drive the language network in the left hemisphere.”

The researchers began by compiling a set of 1,000 sentences from a wide variety of sources: fiction, spoken word transcriptions, web texts, and scientific articles, among others.

Five human participants read each of the sentences while the researchers measured the activity of their language network using functional magnetic resonance imaging (fMRI). The researchers then fed those same 1,000 sentences into a large language model – one similar to ChatGPT, which learns to generate and understand language by predicting the next word in huge amounts of text – and measured the models activation of the model in response to each sentence.

Once they had all this data, the researchers trained a mapping model, called a “coding model,” that connects the activation patterns observed in the human brain with those observed in the artificial language model. Once trained, the model could predict how the human language network would respond to any new sentences based on how the artificial language network would respond to those 1,000 sentences.

The researchers then used the coding model to identify 500 new sentences that would generate maximum activity in the human brain (the “drive” sentences), as well as sentences that would elicit minimal activity in the brain’s linguistic network (the “drive” sentences). DELETE “) .

In a group of three new human participants, the researchers found that these new sentences actually stimulated and suppressed brain activity as expected.

“This closed-loop modulation of brain activity during language processing is novel,” says Tuckute. “Our study shows that the model we use (which maps language pattern activations and brain responses) is precise enough to do this. This is the first demonstration of this approach in brain areas involved in cognition higher level, such as the language network.

Linguistic complexity

To understand what makes certain sentences stimulate activity more than others, the researchers analyzed the sentences based on 11 different linguistic properties, including grammaticality, plausibility, emotional valence (positive or negative), and ease. to visualize the content of the sentence.

For each of these properties, the researchers asked participants on crowdsourcing platforms to rate the sentences. They also used a computer technique to quantify the “surprise” of each phrase, or how rare it is compared to other phrases.

This analysis found that sentences with higher surprise generate higher responses in the brain. This is consistent with previous studies showing that people have more difficulty processing sentences with greater surprise, the researchers say.

Another linguistic property that correlated with the language network responses was linguistic complexity, which is measured by the extent to which a sentence adheres to the rules of English grammar and by its plausibility, that is, the meaning that the content has, apart from grammar.

Sentences at either end of the spectrum – either extremely simple or so complex that they make no sense – evoked very little activation in the language network. The most responses came from phrases that make some sense but require work to understand, like “Jiffy Lube of—of therapies, yes,” which comes from the Corpus of Contemporary American English dataset.

“We found that the sentences that elicit the greatest brain response have strange grammar and/or strange meaning,” says Fedorenko. “There is something a little unusual about these sentences.”

The researchers now plan to see if they can extend these findings to speakers of languages ​​other than English. They also hope to explore what types of stimuli can activate language processing regions in the right hemisphere of the brain.