Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams
Predicting negative features of words in context: The role of negative bias in younger adults
Poster E37 in Poster Session E, Saturday, October 8, 3:15 - 5:00 pm EDT, Millennium Hall
Li-Chuan Ku1, Vicky Tzuyin Lai1; 1University of Arizona
Language prediction studies typically examined the activation of concrete semantic features, e.g., animacy, rather than abstract features, e.g., emotional valence. Do people pre-activate affective features of a word in sentences? One recent study found that emotionally unexpected yet plausible (vs. emotionally expected) target words in sentence-final positions require more neural resources to override anticipated affective representations (Chou et al., 2020). However, they combined positively and negatively valenced stimuli in the analysis, and did not consider readers’ cognitive/affective tendency. Younger readers often showed a negativity bias, i.e., the tendency to attend to negative information (Ku et al., 2020). Here we investigated if and how readers pre-activate positive or negative features of a word in emotionally ambiguous sentences. We hypothesized that undergraduates will show a negativity bias proactively, i.e., in pre-activating more negative features of an upcoming word in sentences. Twenty-seven undergraduates participated (Mage=18.9, 15 females). We assessed participants’ depression, anxiety, and stress levels (DASS), as these variables may bias emotional expectation, and verbal cognitive control (Letter and Category Verbal Fluency Test), as it was associated with language prediction ability. Stimuli included 120 sentence primes with emotionally ambiguous interpretations (e.g., Joan was stunned by her final exam result.), each paired with positive and negative target words (success/distress). The sentence-word pairs were equally semantically related between conditions based on norming (N=46. On a 0-3 scale, M(positive words)=1.99 and M(negative words)=1.93). Target words were matched on length, frequency, part-of-speech, concreteness, and emotional intensity. During EEG recording, in each trial, participants first read an emotionally ambiguous sentence that has either a positive or negative continuation, while actively predicting a continuation. They then saw a positive or negative target word, and judged whether their predicted continuation was similar to the target word on the screen on a 0 (not similar at all) to 3 (very similar) scale. Behaviorally, negative target words (M=1.44, SD=0.26) on average received higher similarity ratings than positive ones (M=1.31, SD=0.3; p=.044), indicating a better match between participants’ predictions and the presented words in the negative condition. For ERPs, positive target words elicited a larger N400 (350-550 ms) than negative words (p=.006), suggesting retrieval difficulty due to incongruency between the predicted negative information and the presented positive word. This N400 was driven by 18 (out of 27) participants who showed negativity bias, evidenced by their higher similarity ratings for negative target words than positive ones. Additionally, positive target words elicited a reduced late positivity (or a sustained negativity; 550-800 ms) than negative ones (p=.012), indicating less meaning elaboration (or continued feature retrieval) due to fewer activated features in positive words. Moreover, participants’ verbal cognitive control predicted both the anterior N400 (β=.626, p=.021) and late positivity (β=.85, p=.002) effects, whereas DASS predicted neither. This suggests that cognitive abilities, rather than speaker affect, modulates affective feature pre-activation. In conclusion, younger adults pre-activate negative features more than positive features during language processing, consistent with “negativity bias”. Such negativity bias is driven by cognitive control, which highlights the importance of individual differences, e.g., age, in affective prediction.
Topic Areas: Meaning: Lexical Semantics, Meaning: Discourse and Pragmatics