FIGURE 2.

An embodied, predictive, and interoceptive framework to osteopathy and mental health. A patient with chronic pain and comorbid depression is lying in supine position, while the therapist applies osteopathic treatment strategies (e.g., providing touch-based interventions to the symptomatic bodily area). Arguably, the belief (prior) of the patient predicts physical and mental states associated with pain and depression to be the likely causes of uncertain sensory information (likelihood). However, if the provided treatment (sensory input) is not linked to these physical and mental states, this surprising mismatch between expected and actual interoceptive information generates interoceptive prediction errors. These interoceptive prediction errors are subsequently minimized using active and perceptual inference processes. If, in chronic pain and depression, high precision is afforded to the belief (prior) and low precision is afforded to the sensory information (likelihood), active inference processes are engaged which produce symptoms resembling the predicted physical and mental states associated with pain and depression through autonomic nervous system activity. However, in a healthcare setting, this might not sufficiently reduce and explain interoceptive prediction errors. Consequently, perceptual inference processes are engaged to update the prior (belief) based on the likelihood (sensory information) thus revising the generative model holding the belief and issuing the prediction. These processes, arguably, underpin osteopathic treatment and putatively reduce (the belief about and prediction of) physical and mental states associated with pain and depression by updating persistent and noisy interoceptive prediction errors (which maintain symptoms through active inference) with surprising and precise interoceptive prediction errors (which alleviate symptoms through perceptual inference).