Many environments do not immediately provide information that is necessary for successful learning and decision-making (e.g., where should I invest my money?). Instead, we often need to combine information across many experiences to learn the outcome contingencies of an environment. This can be particularly challenging under uncertainty, strongly depends on the affective state of an organism, and differs across clinical and age groups. See our recent review and chapter on Adaptive Learning under Uncertainty published in Communications Psychology and on Decision-Making under Uncertainty in the Encyclopedia of the Human Brain.

Learning under perceptual uncertainty 

We investigate the computational principles of reward-based learning under perceptual uncertainty. In this project, we have developed and tested formal models that algorithmically describe how humans take perceptual uncertainty into consideration. In an ongoing project led by Prashanti Ganesh, we study the role of pupil-linked arousal systems that might help the brain adjust learning to perceptual uncertainty. See our most recent work on this topic:

Ganesh, P., Cichy, R. M., Schuck N. W., Finke, C., & Bruckner, R. (2024). Adaptive integration of perceptual and reward information in an uncertain world. eLife, 13:RP99266. <Link>

Ganesh, P., Donner, T. H., Cichy, R. M., Schuck N. W., Finke, C., & Bruckner, R. (2024). Pupil-linked arousal encodes uncertainty-weighted prediction errors. <Link PsyArXiv>

Adaptive learning in changing environments

We study how humans dynamically adjust to changing environments. Humans and animals have to flexibly regulate how they combine newly arriving information with information from past experiences during learning in these environments. In one line of work led by Hashim Satti, we examine learning in changing environments under threat. In particular, we try to better understand how trait anxiety and the proximity of threat changes human abilities to take uncertainty into account during aversive learning and decision-making. See our most recent conference contribution on this topic:

Satti, M. H., Wille, K., Nassar, M. R., Cichy, R. M., Schuck, N. W., Dayan, P., & Bruckner, R. (2024). Absence of systematic effects of trait anxiety on learning under uncertainty. Proceedings of the conference on Cognitive Computational Neuroscience 2024. <Link>

Moreover, we have proposed that resource rationality—adapting belief updating to cognitive limitations—can explain age-related differences in learning. Using a computational model and lifespan data, we show that children and older adults adopt a more frugal sampling strategy, leading to systematic biases such as perseveration and anchoring. Our findings suggest that these biases may not just be deficits but adaptive adjustments to cognitive resource constraints.

Bruckner, R., Nassar, M. R., Li, S.-C., & Eppinger, B. (2025). Differences in learning across the lifespan emerge via resource-rational computations. Psychological Review. Advance online publication. <Link>

< Go back to the landing page to find out more about the group >