MIT researchers led by Evelina Fedorenko tested 376 participants with an eight-hour battery of 20 nonliteral language tasks and replicated the results in 400 more. They identified three clusters of pragmatic inference — social conventions, intonation, and world-knowledge causal reasoning — which appear to reflect distinct pragmatic skills rather than general intelligence or auditory processing. The study, published in the Proceedings of the National Academy of Sciences, has implications for clinical assessment and future brain-imaging studies of communication.
How We Read Between the Lines: 3 Types of Pragmatic Inference Your Brain Uses
Continuous line drawing of a man and a woman having a conversation. Credit: OneLineStock / Shutterstock.
A brief chat with a neighbor that delays your commute by a few minutes relies on surprisingly complex mental work. To understand what someone really means, listeners routinely infer unstated meanings behind ordinary words — decoding metaphors, spotting irony or polite deceptions, catching jokes, or tracking subtle shifts in tone. In longer exchanges, a listener may need to perform many of these inferences in the same conversation.
Scientists refer to these skills as pragmatic language abilities, but only recently have researchers begun to map how those abilities are organized in the mind and brain. New work led by MIT cognitive neuroscientist Evelina Fedorenko suggests that pragmatic skills cluster into three distinct types of inference and that these groups may rely on related neural mechanisms. The team published their results in the Proceedings of the National Academy of Sciences.
“Pragmatics is trying to reason about why somebody might say something, and what is the message they're trying to convey given that they put it in this particular way,”
— Evelina Fedorenko, McGovern Institute for Brain Research, MIT
To probe these abilities, the researchers tested 376 participants using an eight-hour battery of 20 nonliteral comprehension tasks. The tasks covered a wide range of phenomena — including irony, indirect requests, polite deceits, metaphor, and humor — so the team could measure individual differences across many forms of nonliteral language. They then replicated the patterns in an independent sample of 400 participants.
Three Clusters Of Pragmatic Inference
Analysis showed that participants' pragmatic strengths tended to cluster into three groups:
- Social Conventions — reasoning about politeness, social norms, and others' mental states to preserve rapport or "save face" (for example, interpreting ironic remarks used to soften criticism).
- Intonation — using tone, pitch, and prosody to infer emotion, emphasis, or subtle shifts in meaning.
- World-Knowledge Causal Inference — applying general background knowledge to draw causal connections and make sense of remarks in context.
The team found that these groupings were not explained by general intelligence or by basic auditory processing abilities, suggesting the clusters reflect distinct pragmatic capacities rather than broad cognitive or perceptual differences.
Why It Matters
The findings offer a clearer framework for studying nonliteral communication. Clinicians could use this breakdown to better understand why some individuals — for example, people on the autism spectrum — struggle with specific components of pragmatic language. Researchers also plan to use brain imaging to test whether the three pragmatic categories correspond to distinct patterns of neural activity.
By teasing apart the different ways we "read between the lines," this work brings more of the unspoken language that shapes everyday conversation into scientific view.
Published in: Proceedings of the National Academy of Sciences
