Interoception in social cognition under uncertainty: a computational and psychophysiological approach

Interoception - the perception and monitoring of physiological states - plays a critical role in affective processes, intuitive decision-making, and social interactions. However, how interoceptive processing influences emotional and social processes under uncertainty remains empirically underexplored. Within a predictive processing framework, this dissertation combined behavioural, computational, and electrophysiological methods to investigate how individuals integrated interoceptive and external cues to infer others' emotions and adapt empathic responses in dynamic and uncertain environments.

In the first study (Chapter 2), questionnaires and behavioural experiments were employed, along with Structural Equation Modelling and Diffusion Drift Modelling, to explore if intolerance of uncertainty (IU) and alexithymia mediated the relationship between interoceptive sensibility and anxiety, as well as empathic distress and emotion perception. The second study (Chapter 3) further investigated how emotion perception is influenced by interpersonal contextual uncertainty. Specifically, it explored how emotion egocentricity biases in emotion recognition are modulated by learned interpersonal emotion contingencies and perceptual ambiguity. Computational modelling using the Hierarchical Gaussian Filter (HGF) illustrated how individuals adapted their beliefs about interpersonal emotional congruency, with learning trajectories modulated by individual differences in interoceptive sensibility and physiological reactions.

In the third study (Chapter 4), we measured heart-evoked potentials (HEPs), a cortical index of cardiac interoceptive processing, to probe interoceptive predictive processing during an adaptive empathy task. Results indicated that HEP amplitudes tracked precision-weighted prediction errors during social feedback, while showing that interoceptive modulations were associated with autistic traits. HEPs also predicted empathic accuracy during decision-making in this task. Lastly, in chapter 5, transcutaneous vagus nerve stimulation was shown to modulate adaptive empathic learning, providing causal evidence for the role of afferent vagal input in social interactions.

This research advances the understanding of how interoception shapes social and emotional functioning. It uncovers neural, cognitive, and physiological mechanisms underlying empathy and interoceptive processing using computational approaches. These findings have significant implications for our understanding of the role of interoceptive processing on social perception under uncertainty and, potentially, for the development of targeted interventions to improve emotional resilience and social functioning in clinical and non-clinical populations. This thesis enriches emerging perspectives of social cognition as an embodied anticipatory process, where interoception and uncertainty dynamically shape how we perceive, learn, and adapt to the social world.