%A Bauer,Robert %A Gharabaghi,Alireza %D 2015 %J Frontiers in Neuroscience %C %F %G English %K reinforcement learning model,Bayesian Control Rule,Classification Accuracy,Neurofeedback,Neurorehabilitation,Brain-Computer-Interface,Brain-Machine-Interface (BMI),brain-robot interface,reinforcement learning,Functional restoration %Q %R 10.3389/fnins.2015.00036 %W %L %M %P %7 %8 2015-February-12 %9 Original Research %+ Dr Robert Bauer,Division of Functional and Restorative Neurosurgery and Division of Translational Neurosurgery, Department of Neurosurgery, Eberhard Karls University Tuebingen,Tuebingen, Germany,mail@guggenberger.consulting %+ Dr Robert Bauer,Neuroprosthetics Research Group, Werner Reichardt Centre for Integrative Neuroscience, Eberhard Karls University Tuebingen,Tuebingen, Germany,mail@guggenberger.consulting %+ Prof Alireza Gharabaghi,Division of Functional and Restorative Neurosurgery and Division of Translational Neurosurgery, Department of Neurosurgery, Eberhard Karls University Tuebingen,Tuebingen, Germany,mail@guggenberger.consulting %+ Prof Alireza Gharabaghi,Neuroprosthetics Research Group, Werner Reichardt Centre for Integrative Neuroscience, Eberhard Karls University Tuebingen,Tuebingen, Germany,mail@guggenberger.consulting %# %! Restorative brain-computer interfaces and reinforcement learning %* %< %T Reinforcement learning for adaptive threshold control of restorative brain-computer interfaces: a Bayesian simulation %U https://www.frontiersin.org/articles/10.3389/fnins.2015.00036 %V 9 %0 JOURNAL ARTICLE %@ 1662-453X %X Restorative brain-computer interfaces (BCI) are increasingly used to provide feedback of neuronal states in a bid to normalize pathological brain activity and achieve behavioral gains. However, patients and healthy subjects alike often show a large variability, or even inability, of brain self-regulation for BCI control, known as BCI illiteracy. Although current co-adaptive algorithms are powerful for assistive BCIs, their inherent class switching clashes with the operant conditioning goal of restorative BCIs. Moreover, due to the treatment rationale, the classifier of restorative BCIs usually has a constrained feature space, thus limiting the possibility of classifier adaptation. In this context, we applied a Bayesian model of neurofeedback and reinforcement learning for different threshold selection strategies to study the impact of threshold adaptation of a linear classifier on optimizing restorative BCIs. For each feedback iteration, we first determined the thresholds that result in minimal action entropy and maximal instructional efficiency. We then used the resulting vector for the simulation of continuous threshold adaptation. We could thus show that threshold adaptation can improve reinforcement learning, particularly in cases of BCI illiteracy. Finally, on the basis of information-theory, we provided an explanation for the achieved benefits of adaptive threshold setting.