Symposia
Cognitive Science/ Cognitive Processes
William J. Villano, PhD
University of Miami
Coral Gables, Florida
Noah Kraus, B.S.
Research Associate
University of Miami
Coral Gables, Florida
Travis Reneau, B.S.
Graduate Student
Washington University in St. Louis
St. Louis, Missouri
Brittany A. Jaso, Ph.D.
Postdoctoral Fellow
Boston University Center for Anxiety and Related Disorders
Boston, Massachusetts
Anthony Otto, Ph.D.
Assistant Professor
McGill University
Montreal, Quebec, Canada
Aaron Heller, Ph.D.
Associate Professor
University of Miami
Coral Gables, Florida
In uncertain contexts, people and animals are theorized to learn from surprising outcomes (i.e., prediction errors [PEs]) to more accurately predict future outcomes. Laboratory-based work demonstrates that over repeated outcomes, humans use PEs to adjust future expectations, and recent work has linked variation in this PE learning process, such as overlearning from negative relative to positive PEs, to internalizing psychopathology. Yet, it is unclear whether variation in PE learning is a symptom or a risk factor in psychopathology, and whether learning mechanisms identified in the laboratory generalize to real-world settings. Using experience sampling methods from a prior study of emotional responses to PEs (Villano et al., 2020), we assessed 740 college students’ expected exam grades and computed grade PEs as the difference between expected and actual exam grades. Here, we demonstrate that individuals learned to predict grades more accurately after just four exams by updating their expectations for future exams in accordance with grade PEs. Moreover, individuals with elevated neuroticism, a personality trait linked to increased stress reactivity and risk for internalizing psychopathology, were more inaccurate and pessimistic in their grade expectations. Neurotic individuals decreased their expectations more after negative PEs but also made larger updates after small PEs of either valence. This resulted in both valence-dependent and valence-independent learning biases that together could promote increasingly inaccurate and pessimistic expectations over time, which could lead to internalizing disorders. However, we found that the inaccurate expectations observed in neurotic individuals arose specifically from the tendency to overlearn from small PEs. In conclusion, we find support for learning biases in neurotic individuals and demonstrate that overlearning from small PEs may lead to inaccurate expectations for real-world outcomes that promote psychopathology risk.