Residency thesis · BAP04-B-2026-6538 · in preparation
Supervisor · Prof. Dr. Gökhan Sarısoy · Ondokuz Mayıs University
Research
Three open questions and one study already presented, all built on the same method. All of them are about how a brain weighs what it already believes against what it is currently being told.
Belief updating and sensory attenuation in schizophrenia
Two accounts of delusion sit side by side in the literature and are rarely tested in the same people. One says beliefs are revised on too little evidence. The other says self-generated sensation is not attenuated, so ordinary actions carry the weight of external events.
The study runs both: a beads task modelled with the Hierarchical Gaussian Filter, and a force-matching task on a device built for this protocol. Model and parameter recovery are run before data collection. The analysis plan is written; public registration is pending.
Akathisia as active inference
Prediction error can be resolved two ways: revise the model, or act on the world until the sensory evidence matches it. One account treats delusion as a pathology of the first route, though that remains an active debate rather than a settled finding. Akathisia may be a pathology of the second — an inner restlessness that may be the felt quality of an error no action can close.
Aripiprazole makes this testable. A D2 partial agonist does not simply lower dopaminergic signalling; it compresses it toward an intermediate set point, damping rather than removing its dependence on endogenous tone. If dopamine encodes the precision of prediction errors, partial agonism acts as what I will call a precision clamp — the term is mine, not the literature’s — and the prediction is a loss of context-dependent adjustment rather than a shifted baseline.
Effort and reward in multiple sclerosis
Fatigue and depression in MS overlap on every questionnaire that tries to tell them apart. Effort-based decision making offers a behavioural separation: whether the cost of action or the value of the outcome is what has changed.
Belief updating in generalized anxiety disorder
The same method, applied earlier and to a different disorder: belief updating in generalized anxiety disorder, modelled with the Hierarchical Gaussian Filter. Oral presentation, with Selçuk Özdin. This is a separate study from the schizophrenia thesis above, not a preliminary stage of it — it is where the modelling approach the thesis depends on was first worked out.