Files
Abstract
People experience surprise when there is a conflict between their current expectations and reality. An interesting question is to what extent distinct types of expectation violations are context-general vs. context-specific. Previous research has identified a brain network model from functional magnetic resonance imaging (fMRI), the edge-fluctuation-based surprise prediction model (EFPM), which measured regional interaction dynamics (Zhang & Rosenberg, 2023). This model predicted surprise in both an adaptive learning task and a naturalistic sports game viewing context. How do we know if the same set of models predicts participants’ subjective ratings of surprise during movie watching? To address this, we applied the surprise EFPM to the Human Connectome Project (HCP) 7T dataset to predict densely sampled, continuous behavioral ratings and assess the levels of surprise experienced by independent groups of participants while watching the same movie clips. A one-sample t-test comparing the correlations from each subject against a null hypothesis mean of 0 showed a significant association, t (118) = 4.514, p < 0.001, while a circular shift correlation method between the surprise behavioral ratings and overlapping network scores showed an insignificant result, p = 0.640, suggesting that there are shared psychological processes while we experience surprise across distinct contexts. We also tested the strength of the other two networks, the overlapping edges in the surprise networks in task and video (Zhang & Rosenberg, 2023) and the sustained attention connectome-based predictive model (saCPM) (Rosenberg et al., 2016), on predicting the subjective feelings of surprise. For overlapping edges in the surprise networks, the one-sample t-test showed a significant result, t (118) = 5.3164, p < 0.001, and the resulting p-value from the circular shift correlation was 0.529. For saCPM, a one-sample -t-test showed t (118) = 10.232, p < 0.001, and the resulting p-value from the circular shift correlation was 0.041.