We thank Holmes, & Michaels and Datta, for their interest in our editorial. ^{Reference Prosser, Helfer and Leucht1} Both highlight that the link between the free-energy framework ^{Reference Friston2} and psychodynamic psychotherapy was not discussed and – by this omission – suggest that we may believe that psychotherapy equals cognitive–behavioural therapy (CBT). Psychotherapy undoubtedly encompasses more than CBT, but we focused on CBT for two main reasons. First, CBT is arguably the most empirically supported psychotherapy and is commonly used for a wide range of mental and personality disorders, as well as in correctional settings. ^{Reference Hofmann, Asnaani, Vonk, Sawyer and Fang3,Reference Lipsey, Landenberger and Wilson4} Therefore, we believe that our editorial's CBT focus maximises the impact of our main conclusion: that psychotherapy is a biological treatment.

Second, CBT theories resonate very deeply with the free-energy framework because of the centrality both place on top-down processes and learning. ^{Reference Friston2} Both consider these processes to be fundamental – in the sense of basic and necessary – to understanding the pathogenesis of psychopathology. In our view, these concepts have tended to not be as central to psychodynamic theories as they are to CBT theories. More importantly, this example of converging lines of evidence from well-established research fields means that our thesis is not ‘highly speculative’ and not without value, which Michaels & Datta merely proclaim without any evidence for these charges.

Regarding Michaels & Datta's other criticisms, the fact that reliable biomarkers for mental disorders are difficult to identify does not necessarily mean that psychopathology does not have a neurobiological basis, as they suggest. This is because the diagnostic systems used in neuroscientific studies performed to date are woefully inadequate, since the preponderance of evidence suggests that psychopathology is dimensional, rather than forming discrete categories. ^{Reference Haslam, Holland and Kuppens5} Accordingly, a more plausible explanation for this biomarker issue is that we have been using inadequate independent variables in our analyses (i.e., the DSM/ICD categories). Thus, the biomarker issue is likely only a methodological/measurement, rather than a substantive, issue for the neurobiological model of psychopathology.

Michaels & Datta also misunderstand the free-energy framework by suggesting that our thesis is just another grossly oversimplified reductionistic theory. The free-energy framework provides a biopsychosocial explanation of the pathogenesis of psychopathology. This is because experience-dependent synaptic plasticity is fundamental to free-energy minimisation, ^{Reference Friston2} which means that the (social, familial, etc.) environment has a central causal role in neural functioning and its pathologies.

Finally, while we agree that psychodynamic theory is vitally important, Holmes is setting up a false dichotomy between CBT and psychodynamic theory. In brief, the reason is that once you start to understand the pathogenesis of psychopathology using the free-energy framework, core CBT and psychodynamic constructs become integrated into a framework that unifies these psychotherapeutic approaches.

The reason is straightforward: if you want to formalise psychodynamic constructs within the free-energy framework, you will necessarily have to do so in terms of top-down processes and learning, because these are fundamental to how the brain works. ^{Reference Friston2} Thus, psychodynamic constructs will have to be formalised using concepts traditionally associated with CBT. Equally, we find that two core ideas from psychodynamic theory emerge from the free-energy framework. First, according to the free-energy framework, top-down processes and learned responses are not purposeless; rather, they always serve an adaptive function for the brain (i.e., to minimise prediction error). ^{Reference Friston2} This means that some top-down processes and learned responses can be likened to the psychodynamic construct of defence mechanisms, since – in essence – these processes defend against uncertainty (i.e., prediction error). Second, since inferences/predictions drive and modulate neural activity at multiple spatial and temporal scales in the brain, ^{Reference Friston2} much of an individual's mental life is unconscious.