Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Why does predictable emotional language deactivate the amygdala?
There is a Poster PDF for this presentation, but you must be a current member or registered to attend SNL 2023 to view it. Please go to your Account Home page to register.
Poster A79 in Poster Session A, Tuesday, October 24, 10:15 am - 12:00 pm CEST, Espace Vieux-Port
Nati Beltrán1, Winnie Yeh1, Jeremy Skipper1; 1UCL
Introduction: What role does the amygdala play in speech perception and language comprehension? It is activated by emotionally valent and/or arousing words. In contrast, when participants are asked to perform linguistic tasks like ‘affect labelling’ or ‘reappraisal’, the amygdala is typically deactivated. This suggests that engaging speech production systems deactivates the amygdala. Indeed, we discovered that the intersection between ‘language’ (N=1101), ‘speech’ (N=642), and ‘motor’ (N=2565) term-based meta-analyses demonstrates that the amygdala is deactivated when regions associated with speech production are engaged. We and others have shown that speech perception regularly engages speech production systems to predict acoustic patterns arriving in auditory cortex. This led us to test the hypothesis that emotional words that are highly predictable from context will activate speech production systems and, thus, deactivate the amygdala. Methods: Participants watched full length movies during functional magnetic resonance imaging (Naturalistic Neuroimaging Database; Aliko et al., 2020). We first located peaks and troughs in the left and right amygdala, defined as 1.5 standard deviations above or below mean activation. Only those with positively or negatively valenced words above/below 1.5 standard deviations from mean valence in the preceding 4-second windows were kept. To quantify high/low context predictability, we used BERT to calculate the semantic distance between the 4 seconds before troughs/peaks and the preceding 4 seconds, keeping those whose distances were in the top and bottom third. Finally, we averaged brain data in the 4-second window before each remaining peak/trough by hemisphere and condition. These images were entered into two linear mixed effects group analyses (left/right amygdala), each with factors peak (peak/trough), valence (positive/negative), and context (high/low). Results: Preliminary results suggest that, in highly predictable contexts, there was a strikingly lower frequency of amygdala peaks and troughs. Nonetheless, there was a significant three-way interaction between peak, valence, and context in the left ventral central sulcus and the right pars opercularis and precentral sulcus (q < 0.05). This interaction was driven by troughs in the high context and negative valence conditions. General linear tests showed more activity for troughs in the precentral gyrus and sulcus, central sulcus, inferior frontal gyrus, and supplementary motor area for both the left and right amygdala for high context and negative valence. Conclusion: These results suggest that highly predictable emotional words result in deactivation of the amygdala and that this result is at least in part driven by the engagement of speech production systems. This is in line with the aforementioned meta-analyses, where engagement of speech production systems results in deactivation of the amygdala on average. This could be because high contextual constraints engage the emotional processes associated with the amygdala in a predictive manner, explaining deactivation of the amygdala at the time the predictable emotional words occur. As the amygdala is only transiently engaged, it is not reactivated by the predicted words. This leads to the paradoxical suggestion that generally engaging speech production systems results in lower amygdala reactivity which might explain some of the variance in ‘affect labelling’, ‘reappraisal’, or even why psychotherapy works.
Topic Areas: Speech Perception, Control, Selection, and Executive Processes