Prosser et al ^{Reference Prosser, Helfer and Leucht1} cogently argue that psychotherapeutic treatment is no less ‘biological’ than pharmacotherapy – a point also made by Bowlby, who argued, from an ethological perspective, that behaviour is shaped by evolutionary processes no less than anatomy. ^{Reference Holmes2} However, in linking Friston's ‘free energy’ principle with cognitive–behavioural therapy (CBT), they fall into the trap of ‘brand nominalism’ (e.g. Hoover/vacuum cleaner, Coke/soda, Kellogg's/breakfast cereal, etc.). There is a lot more to psychotherapy than CBT; arguably, psychodynamic therapy fits the Friston bill rather better than Beck's baby.

As I (a non-mathematical psychotherapist) understand it, the essence of the free-energy model is the brain's Bayesian shaping of sensory input into experience according to a probabilistic calculus. In healthy psychological functioning, discrepancies between prediction and input are resolved by action to reduce uncertainty and update probabilities. Underlying many psychiatric disorders are relational difficulties arising from outmoded free-energy-reducing models. For the psychologically unwell, both action and the tolerance of uncertainty entailed in updating these models are inhibited.

Psychodynamic therapy addresses this in a number of ways, by: (a) creating a trusting attachment relationship, thereby instating an ‘epistemic superhighway’, ^{Reference Fonagy and Allison3} which, by ‘borrowing’ therapists' brains' free-energy reduction (i.e. their trained Bayesian skills), reduces clients' need to cling to free-energy minimisation at all costs; (b) offering an ambiguous stimulus via the neutral transference-promoting stance of the therapist; (c) promoting ‘action’, not in the CBT sense of ‘experiments’, but in exploring the resulting fears and fantasies – i.e. prior predictions – that arise in vivo with the therapist; (d) tolerating the free energy liberated by abandoning these predictions; (e) instating more adaptive relational probablisms; and (f) reinforcing the capacity for action and updating in the living/learning everyday world, whether ‘natural’ or culturally created (e.g. therapeutic day hospitals, group therapy, etc.).

Psychodynamic therapy is thus not, as Freud dubbed it, ‘the impossible profession’, but the ‘improbable profession’, in that it helps its clients revise their predictions and, in collaboration with the trusted secure base, to live with, and put to good use, the ‘surprise’ associated with liberated free energy. These processes are far from exclusively ‘top down’ as Prosser et al suggest, but implicate the amygdala as much as the prefrontal cortex and need to be seen in the context of the synchronous ‘social brain’ of client and therapist acting in concert.