Separate Bayesian inference reveals model properties shared between multiple experimental conditions
-
1
Bernstein Center for Computational Neuroscience and Technische Universität Berlin, Germany
-
2
Eberhard Karls Universität Tübingen, Germany
Statistical modeling produces compressed and often more meaningful descriptions of experimental data. Many experimental manipulations target selected parameters of a model, and to interpret these parameters other model components need to remain constant.
For example, perceptual psychologists are interested in the perception of luminance patterns depending on their contrast. The model describing this data has two critical parameters: the contrast that elicits a predefined performance, the threshold, and the rate of performance change with increases in contrast, the slope. Typical experiments target threshold differences, assuming constant slope across conditions. This situation requires a balance between model complexity to perform joint inference of all conditions and the simplicity of isolated fits in order to apply robust standard procedures.
We show how separate analysis of experimental conditions can be performed such that all conditions are implicitly taken into account. The procedure is mathematically equivalent to a single Gibbs sampling step in the joint model embracing all conditions. We present a very natural way to check whether separate treatment of each condition or a joint model is more appropriate.
The method is illustrated for the specific case of psychometric functions; however the procedure applies to all models that encompass multiple experimental conditions. Furthermore, it is straight forward to extend the method to models that consist of multiple modules.
Keywords:
Bayesian inference,
gibbs sampling,
modelling,
psychometric function
Conference:
BC11 : Computational Neuroscience & Neurotechnology Bernstein Conference & Neurex Annual Meeting 2011, Freiburg, Germany, 4 Oct - 6 Oct, 2011.
Presentation Type:
Poster
Topic:
data analysis and machine learning (please use "data analysis and machine learning" as keyword)
Citation:
Dold
HM,
Fründ
I and
Wichmann
FA
(2011). Separate Bayesian inference reveals model properties shared between multiple experimental conditions.
Front. Comput. Neurosci.
Conference Abstract:
BC11 : Computational Neuroscience & Neurotechnology Bernstein Conference & Neurex Annual Meeting 2011.
doi: 10.3389/conf.fncom.2011.53.00107
Copyright:
The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers.
They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.
The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.
Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.
For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.
Received:
23 Aug 2011;
Published Online:
04 Oct 2011.
*
Correspondence:
Ms. Hannah M Dold, Bernstein Center for Computational Neuroscience and Technische Universität Berlin, Berlin, Germany, ozeana@gmx.net
Dr. Ingo Fründ, Bernstein Center for Computational Neuroscience and Technische Universität Berlin, Berlin, Germany, ifruend@yorku.ca