A Bayesian brain model of adaptive behavior: An application to the\n Wisconsin Card Sorting Task

Adaptive behavior emerges through a dynamic interaction between cognitive\nagents and changing environmental demands. The investigation of information\nprocessing underlying adaptive behavior relies on controlled experimental\nsettings in which individuals are asked to accomplish demanding tasks whereby a\nhidden state or an abstract rule has to be learned dynamically. Although\nperformance in such tasks is regularly considered as a proxy for measuring\nhigh-level cognitive processes, the standard approach consists in summarizing\nresponse patterns by simple heuristic scoring measures. With this work, we\npropose and validate a new computational Bayesian model accounting for\nindividual performance in the established Wisconsin Card Sorting Test. We embed\nthe new model within the mathematical framework of Bayesian Brain Theory,\naccording to which beliefs about the hidden environmental states are\ndynamically updated following the logic of Bayesian inference. Our\ncomputational model maps distinct cognitive processes into separable,\nneurobiologically plausible, information-theoretic constructs underlying\nobserved response patterns. We assess model identification and expressiveness\nin accounting for meaningful human performance through extensive simulation\nstudies. We further apply the model to real behavioral data in order to\nhighlight the utility of the proposed model in recovering cognitive dynamics at\nan individual level. Practical and theoretical implications of our\ncomputational modeling approach for clinical and cognitive neuroscience\nresearch are finally discussed, as well as potential future improvements.\n