Symposium: Computational Models in Affective Science

Computational models are often used to formalize and study fluctuations of affect over time. A central question to the creation of such models is which characteristics a computational model should possess in order to adequately describe affect dynamics. In this regard, evidence for the presence of nonlinearity in affect dynamics accumulates. However, it is not yet clear where this nonlinearity comes from: It might either represent an inherent characteristic of affect or it might be an artefact due to environmental effects. In this talk, I will present the Affective Ising Model (AIM) – a nonlinear model of affect dynamics – and detail several studies in which we compared its viability against linear competitor models. By accounting for external events in these studies, we were able to investigate whether the observed nonlinearity in affect is indeed due to external events, or due to affect being nonlinear in nature. Results from each study indicate that the AIM outperforms its competitors, even when accounting for external events. This suggests that nonlinearity is a defining feature of affect and should, consequently, be accounted for in our analyses. Submitted for the symposium “Computational models in affective science” with Kenny Yu, Alan Voodla, Lei Zhang, and Niels Vanhasbroeck.

This is an in-person presentation on July 20, 2023 (09:00 ~ 09:20 UTC).

Decisions are often accompanied by feelings of positive or negative valence with some intensity, also called affect. It has been proposed that affect functions as a monitoring signal, recruiting subsequent regulatory control processes. However, it's unclear what are the mechanisms that generate affect in decision-making. Inspired by control process theory (Carver, 2015), we model affect as the difference between expected and actual progress in an evidence accumulation framework. Actual progress is mapped onto the drift-rate parameter and expected progress onto a novel expected drift-rate parameter during a perceptual decision. Affect is computed as the difference between the expected and actual amount of evidence in a trial. We then test predictions of this model in a perceptual decision-making experiment, where expected and actual progress are experimentally manipulated. We find that affect reflects the sum of actual and expected progress, but not their discrepancy as predicted by control process theory. Comparing the empirical data with model predictions, we find that the model is able to simultaneously account for choice, reaction times, and affect in perceptual decisions.

This is an in-person presentation on July 20, 2023 (09:20 ~ 09:40 UTC).

Human generalization research aims to understand the processes underlying the transfer of prior experiences to new contexts. Generalization research predominantly relies on descriptive statistics, assumes a single generalization mechanism, interprets generalization from mono-source data, and disregards individual differences. Unfortunately, such an approach fails to disentangle various mechanisms underlying generalization behavior and can readily result in biased conclusions regarding generalization tendencies. Therefore, we combined a computational model with multi-source data to mechanistically investigate human generalization. By simultaneously modeling learning, perceptual and generalization data at the individual level, we revealed meaningful variations in how different mechanisms contribute to generalization behavior. The current research suggests the need for revising the theoretical and analytic foundations in the field to shift the attention away from forecasting group-level generalization behavior and toward understanding how such phenomena emerge at the individual level. This opens the possibility of having a mechanism-specific differential diagnosis in generalization-related psychiatric disorders. Symposium title: Computational models in affective science Confirmed presenters: Alan Voodla, Lei Zhang, Kenny Yu, and Niels Vanhasbroeck

This is an in-person presentation on July 20, 2023 (09:40 ~ 10:00 UTC).

One of the main challenges in social affective neuroscience originates from the fact that humans do not make decisions alone, but rather, are influenced by their social environment. However, few studies have inspected the underlying neurocomputational processes, in particular when learning from oneself and learning from others coexist in the same environment. Here, I will present a real-time multi-player goal-directed learning paradigm, where, within each group of five individuals, one participant was scanned with MRI. Leveraging reinforcement learning models and fMRI we captured nuanced distinction between direct valuation through experience and vicarious valuation through observation, and their dissociable, but interacting neural representations in the ventromedial prefrontal cortex and the anterior cingulate cortex, respectively, respectively. Connectivity analyses revealed increased functional coupling between the right temporoparietal junction (rTPJ) representing instantaneous social information and the putamen, when individuals made behavioral adjustment as opposed to when they stuck with their initial choice. Together, these data provide a comprehensive behavioral and neurocomputational mechanism of social influence in goal-directed learning and the potential associated social specificity.

This is an in-person presentation on July 20, 2023 (10:00 ~ 10:20 UTC).