Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams

Behavioral and neural dynamics of negation

Poster C74 in Poster Session C, Friday, October 7, 10:15 am - 12:00 pm EDT, Millennium Hall

Arianna Zuanazzi¹, Pablo Ripollés¹, Wy Ming Lin², Laura Gwilliams³, Jean-Rémi King*⁴, David Poeppel*¹; ¹New York University, ²University of Tuebingen, ³UC San Francisco, ⁴PSL University, CNRS

Negation (e.g., ‘no’, ‘not’) is a communication device that is essential and unique to human language. Broadly speaking, negation reverses or modifies the meaning of a word, phrase, or sentence. The function of negation in natural language has been a matter of debate among philosophers, psychologists, and linguists. Conversely, relatively little cognitive neuroscience research has investigated how the brain represents negated concepts. In particular, two fundamental questions remain to be addressed: (1) How does negation operate at the neural level? Psycholinguistic studies suggest that negation reduces the availability of information within its scope; this hypothesis is supported by neuroimaging data showing that negation operates through general-purpose inhibitory mechanisms. These data, however, are mostly limited to the effect of negation on action representation. (2) When does negation operate? Previous psycholinguistic and neuroimaging studies offer contrasting evidence supporting either the hypothesis that negation operates in two steps, whereby the affirmative information is first represented and then suppressed/revised, or the hypothesis that negation operates in a single step, whereby negation incrementally fuses with the representation within its scope. To investigate how and when negation operates on the representation of its scope, we conducted an online behavioral experiment (continuous mouse tracking, 78 participants) and a magnetoencephalography experiment (MEG, 26 participants), in which participants read antonymous adjectives (e.g., 'good'/'bad', 'loud'/'quiet' etc.) within affirmative (e.g., ‘really really good’) or negated phrases (e.g., ‘not really bad’). Each affirmative and negated adjective phrase was rated for its overall meaning on a continuous scale (e.g., from ‘really really bad’ to ‘really really good’). Behavioral results show that participants were slower in deriving the composed meaning of negated than affirmative phrases, suggesting that negation increases processing difficulty. Importantly, mouse-tracking trajectories show that participants initially interpreted negated phrases as affirmative phrases (e.g., ‘not good’ was interpreted as ‘good’) and then modified their interpretation towards the opposite meaning (e.g., 'bad'), however never reaching the opposite side of the scale (e.g., ‘not good’ < ‘bad’). These behavioral results are further illuminated by our MEG decoding results: First, we found that the representation of the adjective (e.g., ‘good’) is similarly encoded in affirmative (e.g., ‘really really good’) and negated (e.g., ‘not really good’) phrases, within the time window associated with lexical-semantic processing (i.e., between 140 and 560 ms from the onset of the adjective); however, the decoding accuracy of the antonyms is reduced in negated compared to affirmative phrases; finally, we observed higher beta power in left-lateralized sensorimotor brain regions for negated than affirmative phrases. Together, our behavioral and MEG results show that adjectives are similarly represented in affirmative and negated phrases. This demonstrates that negation does not readily invert the representation of a scalar adjective to that of its antonym. Moreover, our results show that negation decreases the strength of the representational difference between antonyms and elicits greater beta power, supporting the hypothesis that negation reduces the availability of information within its scope, possibly through general-purpose inhibitory systems.

Topic Areas: Meaning: Lexical Semantics, Meaning: Combinatorial Semantics