Hierarchical Bayesian inference for concurrent model fitting and comparison for group studies

Abstract

Computational modeling plays an important role in modern neuroscience research. Much previous research has relied on statistical methods, separately, to address two problems that are actually interdependent. First, given a particular computational model, Bayesian hierarchical techniques have been used to estimate individual variation in parameters over a population of subjects, leveraging their population-level distributions. Second, candidate models are themselves compared, and individual variation in the expressed model estimated, according to the fits of the models to each subject. The interdependence between these two problems arises because the relevant population for estimating parameters of a model depends on which other subjects express the model. Here, we propose a hierarchical Bayesian inference (HBI) framework for concurrent model comparison, parameter estimation and inference at the population level, combining previous approaches. We show that this framework has important advantages for both parameter estimation and model comparison theoretically and experimentally. The parameters estimated by the HBI show smaller errors compared to other methods. Model comparison by HBI is robust against outliers and is not biased towards overly simplistic models. Furthermore, the fully Bayesian approach of our theory enables researchers to make inference on group-level parameters by performing HBI t-test.

References 37

1. 1.

   O’Doherty JP, Hampton A, Kim H. Model-Based fMRI and Its Application to Reward Learning and Decision Making;1104:35–53.

   Not in our library10.1196/annals.1390.022

2. 2.

   Daw ND, Doya K. The Computational Neurobiology of Learning and Reward;16(2):199–204.

   Not in our library10.1016/j.conb.2006.03.006

3. 3.

   Maia TV, Frank MJ. From Reinforcement Learning Models to Psychiatric and Neurological Disorders;14(2):154–162.

   Not in our library10.1038/nn.2723

4. 4.

   Montague PR, Dolan RJ, Friston KJ, Dayan P. Computational Psychiatry;16(1):72–80.

   Not in our library10.1016/j.tics.2011.11.018

5. 5.

   Friston KJ, Stephan KE, Montague R, Dolan RJ. Computational Psychiatry: The Brain as a Phantastic Organ;1(2):148–158.

   Not in our library10.1016/s2215-0366(14)70275-5

6. 6.

   Huys QJM, Maia TV, Frank MJ. Computational Psychiatry as a Bridge from Neuroscience to Clinical Applications;19(3):404–413.

   Not in our library10.1038/nn.4238

7. 7.

   Friston KJ, Harrison L, Penny W. Dynamic Causal Modelling;19(4):1273–1302.

   Not in our library10.1016/s1053-8119(03)00202-7

8. 8.
   Ten simple rules for dynamic causal modeling · 2010

   No notice
9. 9.

   Cohen JD, Daw N, Engelhardt B, Hasson U, Li K, Niv Y, et al Computational Approaches to fMRI Analysis;20(3):304–313.

   Not in our library10.1038/nn.4499

10. 10.
    Disentangling the Roles of Approach, Activation and Valence in Instrumental and Pavlovian Responding · 2011

    Corrected
11. 11.
    Bonsai Trees in Your Head: How the Pavlovian System Sculpts Goal-Directed Choices by Pruning Decision Trees · 2012

    No notice
12. 12.
    HDDM: Hierarchical Bayesian estimation of the Drift-Diffusion Model in Python · 2013

    No notice
13. 13.

    Stephan KE, Penny WD, Daunizeau J, Moran RJ, Friston KJ. Bayesian Model Selection for Group Studies;46(4):1004–1017.

    Not in our library10.1016/j.neuroimage.2009.03.025

14. 14.

    Dempster AP, Laird NM, Rubin DB. Maximum Likelihood from Incomplete Data via the EM Algorithm;39(1):1–38.

    Not in our library

15. 15.

    Piray P, Zeighami Y, Bahrami F, Eissa AM, Hewedi DH, Moustafa AA. Impulse Control Disorders in Parkinson’s Disease Are Associated with Dysfunction in Stimulus Valuation but Not Action Valuation;34(23):7814–7824.

    Not in our library10.1523/jneurosci.4063-13.2014

16. 16.

    Jordan MI, Ghahramani Z, Jaakkola TS, Saul LK. An Introduction to Variational Methods for Graphical Models In: Jordan MI, editor. Learning in Graphical Models. NATO ASI Series. Springer, Dordrecht;. p. 105–161. Available from: https://link.springer.com/chapter/10.1007/978-94-011-5014-9_5.

    Not in our library

17. 17.

    Bishop CM. Pattern Recognition and Machine Learning. Springer.

    Not in our library

18. 18.

    Neal RM, Hinton GE. A View of the Em Algorithm That Justifies Incremental, Sparse, and Other Variants In: Jordan MI, editor. Learning in Graphical Models. NATO ASI Series. Springer, Dordrecht;. p. 355–368. Available from: https://link.springer.com/chapter/10.1007/978-94-011-5014-9_12.

    Not in our library

19. 19.

    Rigoux L, Stephan KE, Friston KJ, Daunizeau J. Bayesian Model Selection for Group Studies—Revisited;84:971–985.

    Not in our library10.1016/j.neuroimage.2013.08.065

20. 20.

    Friston K, Mattout J, Trujillo-Barreto N, Ashburner J, Penny W. Variational Free Energy and the Laplace Approximation;34(1):220–234.

    Not in our library10.1016/j.neuroimage.2006.08.035

21. 21.
    Observing the Observer (I): Meta-Bayesian Models of Learning and Decision-Making · 2010

    No notice
22. 22.
    VBA: A Probabilistic Treatment of Nonlinear Models for Neurobiological and Behavioural Data · 2014

    No notice
23. 23.

    Daw ND. Trial-by-Trial Data Analysis Using Computational Models In: Delgado MR, Phelps EA, Robbins TW, editors. Decision Making, Affect, and Learning: Attention and Performance XXIII. Oxford University Press;. p. 3–38.

    Not in our library

24. 24.

    Piray P, Ly V, Roelofs K, Cools R, Toni I. Emotionally Aversive Cues Suppress Neural Systems Underlying Optimal Learning in Socially Anxious Individuals;39(8):1445–1456.

    Not in our library10.1523/jneurosci.1394-18.2018

25. 25.

    Frank MJ, Moustafa AA, Haughey HM, Curran T, Hutchison KE. Genetic Triple Dissociation Reveals Multiple Roles for Dopamine in Reinforcement Learning;104(41):16311–16316.

    Not in our library10.1073/pnas.0706111104

26. 26.

    Piray P. The Role of Dorsal Striatal D2-like Receptors in Reversal Learning: A Reinforcement Learning Viewpoint;31(40):14049–14050.

    Not in our library10.1523/jneurosci.3008-11.2011

27. 27.
    Model-Based Influences on Humans' Choices and Striatal Prediction Errors · 2011

    No notice
28. 28.

    Piray P, Toni I, Cools R. Human Choice Strategy Varies with Anatomical Projections from Ventromedial Prefrontal Cortex to Medial Striatum;36(10):2857–2867.

    Not in our library10.1523/jneurosci.2033-15.2016

29. 29.

    Frank MJ, Seeberger LC, O’reilly RC. By Carrot or by Stick: Cognitive Reinforcement Learning in Parkinsonism;306(5703):1940–1943.

    Not in our library10.1126/science.1102941

30. 30.

    Rutledge RB, Lazzaro SC, Lau B, Myers CE, Gluck MA, Glimcher PW. Dopaminergic Drugs Modulate Learning Rates and Perseveration in Parkinson’s Patients in a Dynamic Foraging Task;29(48):15104–15114.

    Not in our library10.1523/jneurosci.3524-09.2009

31. 31.

    Robbins H. An Empirical Bayes Approach to Statistics. The Regents of the University of California;. Available from: https://projecteuclid.org/euclid.bsmsp/1200501653.

    Not in our library

32. 32.

    Casella G. An Introduction to Empirical Bayes Data Analysis;39(2):83–87.

    Not in our library

33. 33.

    Kruschke JK. Bayesian Estimation Supersedes the t Test;142(2):573–603.

    Not in our library10.1037/a0029146

34. 34.

    Daunizeau J. Variational Bayesian Modelling of Mixed-Effects.

    Not in our library

35. 35.

    Raman S, Deserno L, Schlagenhauf F, Stephan KE. A Hierarchical Model for Integrating Unsupervised Generative Embedding and Empirical Bayes;269:6–20.

    Not in our library10.1016/j.jneumeth.2016.04.022

36. 36.

    Yao Y, Raman SS, Schiek M, Leff A, Frässle S, Stephan KE. Variational Bayesian Inversion for Hierarchical Unsupervised Generative Embedding (HUGE);179:604–619.

    Not in our library10.1016/j.neuroimage.2018.06.073

37. 37.

    Stephan KE, Schlagenhauf F, Huys QJM, Raman S, Aponte EA, Brodersen KH, et al Computational Neuroimaging Strategies for Single Patient Predictions;145:180–199.

    Not in our library10.1016/j.neuroimage.2016.06.038

Cited by papers we hold 1

• Ten simple rules for the computational modeling of behavioral data · 2019
  No notice

Reads

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.