A team of scientists has conducted a study to understand how our brains process phrases that include negation, such as “not hot.” The study, led by Arianna Zuanazzi, a postdoctoral fellow at New York University, found that negation does not invert the meaning of adjectives, but rather mitigates it. Negation serves as a mitigator of adjectives like “cold” or “hot,” reducing their intensity rather than completely changing their meaning. This research sheds light on the cognitive processes involved in interpreting subtle changes in meaning in language.
Negation is commonly used in various forms of communication, from advertising to legal filings, to intentionally obscure meaning. The study shows how humans process phrases containing negation and may provide insights into improving artificial intelligence functionality. However, large language models in AI tools often struggle to interpret passages with negation, highlighting the complexity of human language processing compared to machine language understanding. The team’s results offer a deeper understanding of how our brains interpret phrases with negation and suggest ways to enhance AI comprehension.
Human language has the remarkable ability to generate new or complex meanings through word combinations, but the exact process of how this occurs is not well understood. To investigate this, the researchers conducted experiments where participants rated the meaning of phrases with and without negation using a mouse cursor. The experiments revealed that participants took longer to interpret phrases with negation, indicating that negation slows down the processing of meaning. It was observed that negated phrases were initially interpreted as affirmative but later shifted to a mitigated meaning, suggesting that negation weakens rather than flips the original meaning of the adjective.
In addition to the behavioral experiments, the scientists used magnetoencephalography (MEG) to measure the magnetic fields generated by participants’ brain activity during phrase interpretation tasks. The neural data collected during the study supported the findings from the behavioral experiments, showing that negation makes neural representations of polar adjectives like “cold” and “hot” more similar. This suggests that the meaning of phrases like “not hot” is interpreted as “less hot,” rather than as “cold,” strengthening the conclusion that negation mitigates the intensity of adjectives along the semantic continuum.
The research team included other authors such as Pablo Ripollés, Jean-Rémi King, Wy Ming Lin, Laura Gwilliams, and David Poeppel, all from various academic institutions with expertise in psychology, neuroscience, and language processing. The study was supported by a grant from the National Science Foundation, emphasizing the importance of interdisciplinary collaboration in advancing our understanding of language comprehension and cognitive processes. The findings contribute to the ongoing exploration of language processing in the human brain and offer valuable insights for future research in the field.