Capturing dynamic fear experiences in naturalistic contexts: An ecologically valid fMRI signature integrating brain activation and connectivity

Abstract

Enhancing our understanding of how the brain constructs conscious emotional experiences within dynamic real-life contexts necessitates ecologically valid neural models. Here, we present evidence delineating the constraints of current fMRI activation models in capturing naturalistic fear dynamics. To address this challenge, we fuse naturalistic fMRI with predictive modeling techniques to develop an ecologically valid fear signature that integrates activation and connectivity profiles, allowing for accurate prediction of subjective fear experience under highly dynamic close-to-real-life conditions. This signature arises from insights into the crucial role of distributed brain networks and their interactions in emotion modulation, and the potential of network-level information to improve predictions in dynamic contexts. Across a series of investigations, we demonstrate that this signature predicts stable and dynamic fear experiences across naturalistic scenarios with heightened sensitivity and specificity, surpassing traditional activation- and connectivity-based signatures. Notably, the integration of affective connectivity profiles enables accurate real-time predictions of fear fluctuations in naturalistic settings. Additionally, we unearth a distributed yet redundant brain-wide representation of fear experiences. Subjective fear is encoded not only by distributed cortical and subcortical regions but also by their interactions, with no single brain system conveying substantial unique information. Our study establishes a comprehensive and ecologically valid functional brain architecture for subjective fear in dynamic environments and bridges the gap between experimental neuroscience and real-life emotional experience.