Psychiatry must give up mechanistic understandings of mental illness

Nour et al (2022) complain that mechanistic understandings of psychiatric treatments, such as psychotropic medication and psychotherapy, remain elusive. After three decades of intense neuroimaging research, there is still no neurobiological account of any functional psychiatric condition. Nonetheless, Nour et al remain cautiously optimistic that computational psychiatry, with a deeper engagement between psychiatric neuroimaging researchers and artificial intelligence (AI) communities, will still be able to produce clinically relevant mechanistic insights.

Similarly, the Wellcome Trust, a few years ago, recognised that there were underlying problems that needed to be addressed before mental health research could make significant progress (see previous post). What researchers find difficult to accept, though, is that attempts to ‘localise the mind’ through research into brain activities has no future as a research programme. Instead, neurobiology needs to develop its philosophical understanding.

As I mentioned in a previous post, Bennett & Hacker (2021) in the second edition of their Philosophical foundations of neuroscience have separated out a chapter on conceptual problems arising from ascribing to the brain properties that can only sensibly ascribed to the organism as a whole. This is what they call the mereological fallacy. Psychiatry must stop identifying the brain with the person. Instead, cognition needs to be understood in a dynamic, integrated, enactive way as it is embodied in the brain and the body more generally, and embedded in the environment, which is social and cultural, affording various possibilities of action to the organism (see eg. my article). 

For example, the experience and relationships of people are at the core of depression, and they cannot be identified with neuronal or molecular processes (see eg. previous post). Psychiatrists need to accept that even though physical illness is accompanied by bodily disease, primary mental illness is not accompanied by structural brain disease.

Nor is computational psychiatry the answer. People are not computers. There is a fundamental difference, and not just of degree, between the computing and adaptation capabilities of a computer system and the perceptions, insights, thinking and understanding of a human being (see previous post). A mechanistic conception of nature fails to provide a complete characterisation of living beings (see eg. another previous post).

(With thanks to Mad in America research news item by Peter Simons)