A reduced sense of agency is a primary symptom of depression. This study investigates how agency is learned and perceived in a multi-agent environment, comparing depressed and non-depressed individuals. Participants explored their control over an on-screen outcome via button presses across multiple trials, while observing a simulated agent’s button presses. They rated each agent’s control after each block of trials. Experimental conditions varied which agent had control and the type/direction of control (positive/excitatory, negative/inhibitory, or none). We applied an active inference model to the behavioural data to understand action and perceptual processes in forming agentic beliefs. Results showed all participants identified the controlling agent but incorrectly attributed control to the non-controlling agent in the opposite direction. Depressed participants consistently perceived themselves and others as having less agency, while viewing the other agent as having positive control across all conditions. The model suggested stronger prior beliefs about agents having opposing directions of control. Depressed participants, however, perceived their control to match that of external agents, while non-depressed participants perceived their control to be independent of the external agent. Despite perceiving reduced agency, depressed participants increased their agentic action over time, suggesting heightened but potentially aimless environmental sampling. The model reflected this as a higher prior for acting and a lower preference for observing outcomes. This study provides novel insights into how individuals with and without depression perceive agency differently within the same environment through differing biases in perceptual and action processes. Depressed individuals may not learn about other agents in the same way as non-depressed individuals. These findings could inform future therapeutic strategies and deepen our understanding of depression.

Modelling Agency Perception in a Multi-agent Context in Depression Using Active Inference

Folesani, Federica;Murri, Martino Belvederi;
2025

Abstract

A reduced sense of agency is a primary symptom of depression. This study investigates how agency is learned and perceived in a multi-agent environment, comparing depressed and non-depressed individuals. Participants explored their control over an on-screen outcome via button presses across multiple trials, while observing a simulated agent’s button presses. They rated each agent’s control after each block of trials. Experimental conditions varied which agent had control and the type/direction of control (positive/excitatory, negative/inhibitory, or none). We applied an active inference model to the behavioural data to understand action and perceptual processes in forming agentic beliefs. Results showed all participants identified the controlling agent but incorrectly attributed control to the non-controlling agent in the opposite direction. Depressed participants consistently perceived themselves and others as having less agency, while viewing the other agent as having positive control across all conditions. The model suggested stronger prior beliefs about agents having opposing directions of control. Depressed participants, however, perceived their control to match that of external agents, while non-depressed participants perceived their control to be independent of the external agent. Despite perceiving reduced agency, depressed participants increased their agentic action over time, suggesting heightened but potentially aimless environmental sampling. The model reflected this as a higher prior for acting and a lower preference for observing outcomes. This study provides novel insights into how individuals with and without depression perceive agency differently within the same environment through differing biases in perceptual and action processes. Depressed individuals may not learn about other agents in the same way as non-depressed individuals. These findings could inform future therapeutic strategies and deepen our understanding of depression.
2025
9783031771378
9783031771385
active inference; computational psychopathology; depression; multi-agent; sense of agency;
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11392/2604053
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? 0
social impact