A new approach to the study of mental disorder—called computational psychiatry—uses Bayesian inference to explain where people with problems are going wrong.
Bayesian inference is a method of statistical reasoning used to understand the probability of a hypothesis and how to update it as conditions change.
The idea is that people with schizophrenia, for example, are doing a bad job at inferring the reasonableness of their hypotheses. This happens because schizophrenics seem to be less likely to put enough weight on prior experience (a factor in Bayesian reasoning).
Somewhat similarly, “sensory information takes priority [over previous experience] in people with autism.” (Bayesian reasoning implicated in some mental disorders)
Distorted calculations — and the altered versions of the world they create — may also play a role in depression and anxiety, some researchers think. While suffering from depression, people may hold on to distorted priors — believing that good things are out of reach, for instance. And people with high anxiety can have trouble making good choices in a volatile environment… (Ibid)
The key problem with autism and anxiety is people with these conditions have trouble updating their expectations—a major component of Bayesian reasoning—and thus make many mistakes.
These mistakes, of course, compound and further increase a sense of anxiety or alienation.
Like several of the researches quoted in the linked article, I find this computational approach exciting.
It speaks to me because it confirms a core hypothesis of FIML practice—that all people make many, significant inferential mistakes during virtually all acts of communication.
In this respect, I believe all people are mentally disordered, not just the ones who are suffering the most.
I think a Bayesian thought experiment can all but prove my point:
What are the odds that you will correctly infer the mental state(s) of anyone you speak with? What are the odds that they will correctly infer your mental state(s)?
In a formal setting, both of you will do well enough if the inferring is kept within whatever the formal boundaries are. But that is all you will be able to infer reasonably well.
In the far more important realm of intimate interpersonal communication, the odds that either party is making correct inferences go down significantly.
If we do not know someone’s mental state, we cannot know why they have communicated as they have. If our inferences about them are based on such questionable data, we are bound to make many more mistakes about them.