Technical Program

Paper Detail

Paper: PS-2A.46
Session: Poster Session 2A
Location: H Lichthof
Session Time: Sunday, September 15, 17:15 - 20:15
Presentation Time:Sunday, September 15, 17:15 - 20:15
Presentation: Poster
Publication: 2019 Conference on Cognitive Computational Neuroscience, 13-16 September 2019, Berlin, Germany
Paper Title: Fear Generalization of Emotional Stimuli Can Be Explained By a Bayesian Inference Model
Manuscript:  Click here to view manuscript
License: Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 Unported License.
Authors: Lukas Neugebauer, Christian Büchel, University Medical Center Hamburg-Eppendorf, Germany
Abstract: The amount of generalization that organisms show from learned associations to new stimuli can often be explained by perceptual dissimilarity, i.e. distance in psychological space. However, this doesn’t seem to be true using emotionally relevant stimuli like fearful faces. We propose that this can be understood in a Bayesian framework in which the organism infers a mapping of psychological space onto outcome probabilities by integrating prior assumptions with new information. This approach allows for the incorporation of domain specific prior knowledge. We employed face stimuli that differ on one fear relevant (emotional expression) and one fear irrelevant (identity) dimension in combination with Pavlovian conditioning to investigate generalization at several time points. We can show that generalization is skewed towards the fear relevant pole in the beginning, but gravitates towards the actually reinforced stimulus over time. Our Bayesian model that comprises a prior belief state about the structure of the predictive relationship between the psychological space and an aversive outcome can reproduce the experimental data.