Adaptive Stimulus Selection to Estimate Model Parameters in the Rod-And-Frame Task
Keywords
No Thumbnail Available
Authors
Issue Date
2018-06-17
Language
en
Document type
Journal Title
Journal ISSN
Volume Title
Publisher
Title
ISSN
Volume
Issue
Startpage
Endpage
DOI
Abstract
In the rod-and-frame task human subjects are asked to judge the orientation of a rod in a frame
(clock or counter clockwise) relative to the subjects perceived gravity. The orientations of both the
frame and the rod are altered throughout the experiment. In addition the experiment is conducted
with different head orientations which affect the perception of gravity. The response of a subject
to a stimulus is used to update the knowledge about the psychometric model of the subject. Using
this new information it is possible to calculate, using a Bayesian model, the stimulus that will give
the most information on the next trial. And every trial the algorithm updates its estimation of a
subjects psychometric model. By using such an adaptive method I hope to get the same results, as
an experiment done with a ’fixed stimulus’ paradigm, in less trials. Thereby reducing the costs of
doing experiments and reducing the probability of a subject getting tired or bored and therefore
giving less reliable data. In the end the results were inconclusive because of implementation
errors or non-optimal choices during the development of the algorithm. Furthermore in some
tests it took 3 seconds to calculate the next stimulus therefore it is not applicable to a normal
experimental situation. Nevertheless I am still convinced of this paradigm and therefore I think
that more research has to be done to find my mistakes and improve the algorithm.
Description
Citation
Supervisor
Faculty
Faculteit der Sociale Wetenschappen