The E ect of Detail on Predictive Models A search for the optimal granularity
Keywords
No Thumbnail Available
Authors
Issue Date
2020-03-01
Language
en
Document type
Journal Title
Journal ISSN
Volume Title
Publisher
Title
ISSN
Volume
Issue
Startpage
Endpage
DOI
Abstract
Predictive processing is a theory in Cognitive Neuroscience which states that informa-
tion processing in the brain is performed via predicting future inputs and comparing
them to the actual observations. Recently it has been proposed that knowing what
determines the level of detail of the predictions is important for understanding the
predictive models in our brain. For this research project, a simulation was built in
which multiple agents made predictions at di erent levels of detail. The results of this
simulation were analysed to see if an optimal level of detail could be found, where the
prediction error is low while predictions are su ciently informative. In the end the
optimum could not be determined, and it remains an open question how exactly the
brain balances expected prediction error with informativeness of predictions.
Description
Citation
Supervisor
Faculty
Faculteit der Sociale Wetenschappen