A podcast episode turns computational psychiatry into something you can say to a patient
- PsyDactic episode 72, "Functional Neurological Disorder, Predictive Processing and Active Inference," hosted by T. Ryan O'Leary, MD (a child and adolescent psychiatry fellow in the National Capital Region), released March 20, 2025, runs 24 minutes 37 seconds, and is free with a full transcript on its Buzzsprout page.
- The episode argues that FND runs on a false Bayesian inference - the brain generates a symptom-producing prediction and weighs it over the sensory evidence that contradicts it - a mechanism traced to a 2012 paper in Brain, distinct from an "it's psychological" account.
- It grounds the model in specific circuitry (prefrontal cortex, anterior cingulate cortex, amygdala, noradrenaline as the precision-weighting signal, cortisol as a marker of allostatic load) and treats stress itself as uncertainty, or entropy, following Peters et al. (2017); Hoover's sign, a bedside test for functional leg weakness, illustrates precision-weighting directly.
- Named clinical implications: physical therapy (with attentional diversion), CBT, and hypnosis all work, in this framework, by recalibrating the brain's faulty predictive model - and the episode is also available on Apple Podcasts, Amazon Music, and YouTube.
Most computational-psychiatry ideas stay theoretical: a framework for grant applications, not a tool for the exam room. Functional neurological disorder is the exception this episode picks. PsyDactic episode 72 walks through predictive processing and active inference not as abstractions but as an explanation a clinician can hand a patient - and the reasoning behind treatments already in use.
The Mechanism Behind the Symptom
The episode's central claim, drawn from a 2012 paper in Brain, is that FND runs on a false Bayesian inference: the brain's prediction of a symptom outweighs the sensory evidence that would correct it. Hoover's sign - a patient with "paralyzed" leg extension that fires normally when the other leg pushes down against resistance - becomes the bedside demonstration of precision-weighting, the mechanism by which the brain decides which signal to trust. Peters et al. (2017) supplies the second piece: stress functions here as uncertainty, or entropy, that the predictive system has to resolve one way or another.
The circuit gives the theory a treatment
The episode ties this to a specific circuit - prefrontal cortex, anterior cingulate cortex, amygdala - with noradrenaline doing the precision-weighting and cortisol marking allostatic load. That circuit-level account is what lets the episode name a mechanism for treatments FND clinicians already reach for: physical therapy with attentional distraction, CBT, and hypnosis all work, in this frame, by recalibrating a faulty predictive model, not by whichever mechanism is usually cited for each therapy on its own.
Hoover's sign is not a trick to catch a faker - it is precision-weighting, visible at the bedside.
The source is a podcast episode, not a peer-reviewed publication, and the host discloses that AI tools assisted in preparing the material - a disclosure worth flagging, not a reason to discount the content. The host also states directly that hypnotizability does not predict treatment outcome, so the model explains a mechanism without yet offering a way to predict who will respond to therapy.