A Computational Account of Approach-Avoidance Conflict in a Foraging-Inspired Search Paradigm
Abstract
An approach-avoidance conflict (AAC) arises when the same action promises reward but also carries potential harm. Here, we examined how humans initiate and sustain engagement under such conflict using an online foraging-inspired AAC task. Eighty-seven participants repeatedly encountered visual search fields associated with four instructed reward-punishment states and decided whether to enter each field and how extensively to search it, allowing us to distinguish passive avoidance as withheld entry from active within-field avoidance as early search termination. Participants committed to search more often and explored more extensively when expected rewards were high and punishments were low. Agent-based behavioral modeling indicated that initial and sustained engagement could be characterized at different levels of behavioral granularity: commitment to search was best explained by additive reward-punishment evaluation with state-independent trial drift, whereas search extent was best captured by a click-wise stopping process inspired by the marginal value theorem, a foraging account of when to leave a current patch for alternatives. These findings suggest that avoidance under conflict is not a single decision, but unfolds across separable decisions about whether to engage, how long to persist, and when to leave.
Related
- The diminishing returns of distance: Computational analysis of approach-avoidance behavior under threat of predation
- Psychosis-associated disruptions to semantic foraging in narrative recollections: Integrating language model embeddings and topic modeling
- Repeated Social Defeat Stress Induces Individual Differences in Effort-Based Motivation Dissociable from Social Avoidance in Mice
- Steroid avoidance or withdrawal for kidney transplant recipients
- Prospective associations between perceived school climate and psychological distress among Chinese adolescents: a three-wave longitudinal study of school belonging, social avoidance, and psychological resilience
BibTeX
@misc{usee2026computational,
title = {{A Computational Account of Approach-Avoidance Conflict in a Foraging-Inspired Search Paradigm}},
author = {Franziska Usée and Christiane A. Melzig and Dirk Ostwald and Anne Collins},
howpublished = {PsyArXiv},
year = {2026},
month = sep,
doi = {10.31234/osf.io/7rbng_v1},
url = {https://osf.io/preprints/psyarxiv/7rbng_v1/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - A Computational Account of Approach-Avoidance Conflict in a Foraging-Inspired Search Paradigm
AU - Franziska Usée
AU - Christiane A. Melzig
AU - Dirk Ostwald
AU - Anne Collins
PY - 2026
DA - 2026/09/13
PB - PsyArXiv
DO - 10.31234/osf.io/7rbng_v1
UR - https://osf.io/preprints/psyarxiv/7rbng_v1/
AB - An approach-avoidance conflict (AAC) arises when the same action promises reward but also carries potential harm. Here, we examined how humans initiate and sustain engagement under such conflict using an online foraging-inspired AAC task. Eighty-seven participants repeatedly encountered visual search fields associated with four instructed reward-punishment states and decided whether to enter each field and how extensively to search it, allowing us to distinguish passive avoidance as withheld entry from active within-field avoidance as early search termination. Participants committed to search more often and explored more extensively when expected rewards were high and punishments were low. Agent-based behavioral modeling indicated that initial and sustained engagement could be characterized at different levels of behavioral granularity: commitment to search was best explained by additive reward-punishment evaluation with state-independent trial drift, whereas search extent was best captured by a click-wise stopping process inspired by the marginal value theorem, a foraging account of when to leave a current patch for alternatives. These findings suggest that avoidance under conflict is not a single decision, but unfolds across separable decisions about whether to engage, how long to persist, and when to leave.
ER -