Penn–Hamburg Research Reveals How Previous Choices Shape the Evidence We Believe
NSF- and BMBF-supported collaboration combines neuroscience, behavioral experiments, and computational modeling to uncover mechanisms underlying confirmation bias in human decision-making.
A research collaboration between the University of Pennsylvania and University Medical Center Hamburg-Eppendorf (UKE) is providing new insight into a familiar feature of human behavior: once people make a decision, that choice can influence how they evaluate the information that follows.
Published in Nature Communications, the study “Confirmation Bias through Selective Readout of Information Encoded in Human Parietal Cortex” examines whether confirmation bias arises because the brain represents new information differently after a choice has been made, or because decision-makers selectively use information that supports their previous choice.
The research emerged from Choice-induced Biases in Human Decision-Making (DECINDBIAS), a U.S.–German collaborative project led by Alan A. Stocker, PhD, at Penn and Tobias H. Donner, MD, PhD, at UKE in Hamburg. The project was jointly supported through the Collaborative Research in Computational Neuroscience (CRCNS) program by the U.S. National Science Foundation (NSF) and the German Federal Ministry of Education and Research (BMBF).
Alan A. Stocker, Ph.D.
Professor of Psychology
Department of Psychology, School of Arts & Sciences
University of Pennsylvania
Secondary Appointment, Electrical and Systems Engineering
School of Engineering and Applied Science
Affiliated Faculty, Neuroscience Graduate and Bioengineering Graduate Group
Stocker leads Penn’s Computational Perception and Cognition Laboratory, where his research examines the computational principles underlying human perception and decision-making. His work investigates how sensory information interacts with prior knowledge, expectations, and uncertainty to shape what people perceive and the decisions they make.
By combining psychophysical experiments with mathematical and computational modeling, Stocker’s laboratory seeks to identify the computations the brain uses to interpret uncertain sensory information and translate it into behavior.
Tobias H. Donner, MD, Ph.D.
Professor of Integrative Neuroscience
Department of Neurophysiology and Pathophysiology
University Medical Center Hamburg-Eppendorf (UKE)
Hamburg, Germany
Donner leads an interdisciplinary research group investigating the neural mechanisms underlying perception, cognition, and decision-making. His research examines how ongoing fluctuations in brain activity interact with sensory information to influence what people perceive and how they choose.
The Donner Lab combines human psychophysics with computational modeling and systems-neuroscience approaches, including magnetoencephalography (MEG), to connect patterns of neural activity with cognition and behavior.
Together, the Stocker and Donner laboratories bring complementary expertise to DECINDBIAS. Penn contributes strengths in computational theories of perception and decision-making, while the Hamburg team contributes expertise in human neurophysiology and MEG. This combination allows the researchers to investigate decision-making simultaneously at behavioral, computational, and neural levels.
How Previous Choices Shape New Evidence
The broader DECINDBIAS project investigates two closely related consequences of decision-making: consistency bias, in which a decision influences how people remember evidence they encountered previously, and confirmation bias, in which a previous choice influences how subsequent evidence is evaluated.
For the Nature Communications study, researchers combined psychophysical experiments with MEG to measure cortical activity as participants evaluated sequences of visual evidence and made perceptual decisions.
The researchers found that new evidence was represented accurately in visual and parietal regions of the brain regardless of whether it agreed with a participant’s previous choice. The bias emerged at a later stage, when that information was used to form another judgment: evidence consistent with an earlier decision exerted greater influence on the subsequent choice than evidence that contradicted it.
The findings suggest that confirmation bias may arise not because the brain fails to register contradictory information, but because people selectively use information that supports decisions they have already made. Distinguishing between how information is represented and how it is subsequently used provides a more precise framework for understanding the neural and computational origins of confirmation bias.
The findings may also help researchers investigate whether these biases can be modified. If contradictory information remains accurately represented in the brain but is weighted differently during decision-making, interventions involving feedback, incentives, or deliberate cognitive strategies may be able to influence how that information is used.
Building a U.S.–German Research Partnership
DECINDBIAS demonstrates how coordinated international research funding can bring together complementary expertise and research infrastructure. The project was jointly supported by the U.S. National Science Foundation (NSF) and the German Federal Ministry of Education and Research (BMBF) through the Collaborative Research in Computational Neuroscience (CRCNS) program. Additional support came from NSF and the U.S. Department of Defense through the NSF AI Institute for Artificial and Natural Intelligence (ARNI).
The partnership connects Penn’s research community with neuroscience expertise and infrastructure in Hamburg, creating opportunities to exchange experimental methods, computational approaches, and theoretical perspectives while pursuing shared questions about human cognition. Preliminary findings from the collaboration were also presented at the Vision Sciences Society Annual Meetings before publication.
Through support for interdisciplinary research and international engagement, the Office of the Vice Provost for Research (OVPR) helps strengthen partnerships that connect Penn researchers with complementary expertise and research infrastructure around the world. Collaborations such as DECINDBIAS demonstrate how coordinated U.S.–German research investment can create sustained scientific relationships while advancing fundamental understanding of the human brain and behavior.
Read the Nature Communications study:
https://www.nature.com/articles/s41467-025-61010-x
Learn more about the DECINDBIAS research:
https://tobiasdonner.net/new-paper-out-on-confirmation-bias/