The Mind Is Bayesian, Not Freudian

By Graham Peterson

Defensiveness, as I understand the term, is an emotion motivated by stored up traumas.  Traumas drive people to react disproportionately to external events, that are in a current context tangential or unrelated to those internal fears.  “You’re just being defensive,” as it goes.

Note (again), that’s an unfalsifiable accusation.  Any defense against the accusation goes toward proving it’s true. “No, I’m not.”  “See, I told you that you were being defensive.” Psychoanalytic claims are accusatory cannon fodder — and are not measurable, checkable, scientific claims.  Moreover, they’re a recipe for division and paranoia.

But enough Freud bashing.  Let’s think about an alternative theory of mind.

I’ve known some defensive people in my life.  I know a guy who basically always got his way because challenging him, even on the small stuff, was asking for a fight.  That guy was also beaten — mercilessly — by his parents, until a foster parent took him in around age ten.  So is it fair to accuse him of being defensive?  I don’t think so.  The phrasing has all kinds of prejudicial weight.  It means he carries some burden or baggage in his subconscious that he is unable to access or control.  It means he’s irrational.

But let’s think harder.  To this guy, who has experienced routine and disproportionate sanctions for his behaviors, constant criticism and attacks from people who love him — to him the correct inference to make, conditioned on a signal like “X is getting critical with me,” is that a humiliating beating is coming.

If we think of people making perfectly rational predictions about the likelihood of future events, based on a probability distribution they’ve constructed with memories and prior information, we have a Bayesian theory of mind.  Defensiveness is in this theory merely incorrect inferences one makes after switching contexts.  Mistakes in inference and behavior come from external changes in the social and physical environment.

The mind, in this view, is a machine built to interact with an environment and other people, not a machine built to manage its own internal logic, relative to some prehistorically given drives and urges.  There are three advantages to the social, Bayesian model of mind, as opposed to the individual, psychoanalytic theory of mind.  The first is scientific, the second is therapeutic, and the third is normative.

First, a Bayesian theory of mind is scientifically attractive.  It makes generalizable and mathematically tractable, probabilistic predictions about how people will react externally.  Both stimulus and response are externally observable, and are connected by an internally unobservable mechanism that makes ordered predictions.  Observing the order of responses, with respect to the order of stimuli, reveals the logic of the unobservable mechanism.

The Freudian theory of mind is on the other hand, literally, untestable story telling about a tragic struggle for control among characters inside people’s heads.  The Bayesian theory of mind is falsifiable; the psychoanalytic theory of mind is not.

Someone either learns a probability distribution based over events, and responds predictably from draws over that distribution, or they don’t.  The only way to observe the narrative struggle among the alleged characters in the Freudian story is to ask people to openly reflect.  But remember that the entire theory is predicated on the idea that they are lying to themselves and us.  “You’re in denial.”  “No, I’m not.”  “See.”  “You’re being defensive.”  “No, I’m not.”  “See.”  So there go our external observations and verifiability.

Second, a Bayesian theory of mind  recommends a dramatically different treatment regiment for folks who are having trouble.

If people are actually Bayesian, then we wouldn’t want to put people on couches and dredge up their subconscious, hoping for release.  We wouldn’t treat therapy like a tabloid expose of internalized drama.  We wouldn’t try to catch them in their own bullshit or tell them to catch themselves.  (I can’t imagine a more destructive thing to do to someone with low self esteem, who is afraid, than give her even more reason to degrade her faith in herself.)

We would talk about their histories and memories in order to establish the frequency with which patients (reasonably) expect to experience traumatizing (and joyful!) events.  We would then discuss with a patient whether or not those expectations, given the current environment the patient is in, are in fact reasonable.  If not, we would understand that he is making mistakes in estimating probabilities, not lacking some amorphous “strength” or “maturity” to “manage” his internal drives.

So we would encourage him to go out into the world and take more samples of current context, in order to update the information on his priors, not sit at home and pick himself apart, second guessing himself, hoping to reveal and unlock a big box of built up trauma.  This is in essence what I understand Cognitive Behavioral Therapy to be, and is probably why it works so well.

People recognize that their fears (priors) are irrational (incongruent with their current environment), and go out and get new experiences to retrain themselves (their priors) to react differently (make better predictions) about current stimuli.

Finally, the advantages are normative.  The Bayesian theory of mind paints a much nicer picture of people.  It assumes that people are essentially are rational and intelligent, not irrational animals trying to keep a lid on their erections and rage.  It takes for granted that people are usually in equilibrium with their environments, responding proportionally and intelligently to events.

In its role as a theory that people use to anticipate and interact with one another, the Bayesian theory generates empathy and understanding.  It encourages us to try and understand the distribution of priors that one another are working with, all slightly different, but for understandable reasons.  We stop seeing one another as nuclear reactors of internal power struggles, waiting to “take out” our “built up” issues.  And we start believing in each others, on balance, good intentions.

We stop being paranoid of one another’s subconscious demons and traumas, and most importantly — we stop being paranoid of our own subconscious demons and traumas.  We realize that we are built as inference machines, linked up to a social and physical environment.  We realize that one another and our minds are social, and that we reason together in groups, about one another, and solve problems with one another.

The Bayesian theory of mind is hopeful, charitable, and testable.  The psychoanalytic theory of mind is an untestable, derogatory cudgel.

2 thoughts on “The Mind Is Bayesian, Not Freudian”

    1. Hey thanks! I pulled most of this from intuitions, but chapter 3 of John Levi Martin’s Explanations of Social Action is a great takedown of Freud. I’m pretty sure economists already consider the mind to be a Bayesian prediction and decision machine when they talk about expected utility. A google scholar search turned up a few hits for bayesian theory of mind too. I’m mostly shooting from the hip here though.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s