Devoodooifying Psychology

[Epistemic status: very low. Total conjecture based on insufficient evidence.]

“Voodoo death” refers to supposed cases where people died after being cursed by witch doctors. The theory goes that even though witch doctors don’t have real magic, if their victims come from a culture that believes in witchdoctory then they’ll be so scared that they gradually waste away out of fear and die anyway.

For a while psychologists believed that this absolutely happened, a testament to the powers of the mind. Maybe it was because of toxic levels of the stress hormone adrenaline or something.

Now there’s some more controversy. A lot of these cases turned out to be primitive tribesmen who said it totally happened to a friend of a friend of a cousin or something. In others, witch doctors placed curses on people who already had some kind of serious disease, then took credit. In still others, the curse victim became so upset that they stopped eating or drinking and died of dehydration – which, while technically a death, doesn’t really testify to the powers of the mind so much as the power of blood osmolality. And after some more thought, everyone agreed the adrenaline theory probably didn’t apply since adrenaline spikes kill suddenly but voodoo victims waste away over the space of weeks.

So now voodoo death looks a lot more complicated than a simple progression of curse -> adrenaline -> you were killed by your own mind. Lester gives a relatively sympathetic view of the evidence here, but even he can only find two good cases – one of which involved a patient who died of preexisting asthma, the other of which did not involve voodoo at all.

I find this interesting because so much of psychology seems basically voodoo-ish.

Take the placebo effect. This is basically the voodoo effect in reverse. Instead of a witch doctor saying you’ll get worse, and you do, a regular doctor tells you you’ll get better, and you do. For a while, people were claiming all sorts of amazing effects for placebo – placebo can activate the immune system to fight infections, placebo can slow the growth of cancer, placebo can make bedridden invalids start dancing jigs. But the best studies now suggest that the placebo effect is probably very weak and limited to controlling pain. The vaunted power of mind over body, of belief over reality, doesn’t look nearly as impressive as we thought.

Or take stereotype threat. Again, this is sort of a voodoo curse. If people make you think you’re going to do bad on a test, then you’ll do bad on the test. Again, widely believed, held up as an example of the power of perception. Again, doesn’t replicate well in large studies, has a very suspicious funnel plot, and is starting to inspire doubt even among top researchers in the area.

Or take self-esteem. Again, a reverse voodoo curse. By believing that you’re a good person and likely to do well, good things will happen to you. Again, very popular in the ’90s, but it hasn’t aged well. Similarly, self affirmations failed to replicate results showing their effectiveness.

This is all oversimplified; there are still lots of unrebutted studies supporting all of these things. Maybe some of the studies that seem to debunk them have themselves been debunked, and I don’t know about it. Still, it seems to me that things that sound like voodoo – that is, which argue that our optimistic or pessimistic beliefs about how well we will do can mysteriously and directly affect how well we will do – are faring unusually badly as psychologists get better at trying to replicate things.

(This kind of thing is why I’m so skeptical of growth mindset, despite so little hard data supporting my skepticism)

This isn’t to say that nothing like this is true. At the very least, my belief that I can’t swim the English Channel makes me not try to swim the English Channel, and so if that belief is wrong it’s voodoo-ish-ly making me fail at Channel-swimming. But this is a lot less mysterious than the thing where you repeat “I will do well at school” every day and so do better at school.

Let me introduce a second category of things.

First, the name preference effect. This is where you’re positively predisposed to things that sound like your name. For example, people named Dennis are more likely to become dentists, people named Bob are more likely to be bakers or butchers or barristers, and people with three letters in their name are more likely to live in Three Forks. People believed this for years until finally somebody did a statistical reanalysis and found that it was totally false.

Second, unconscious social priming. Supposedly people who heard the word “retirement” walked more slowly for a while afterwards, because “retirement” primed their thoughts of old people, and old people primed their thoughts of being slow, and so for a while they themselves behaved like an old person. This sort of thing inspired an entire field of psychology showing similar results (like the infamous study showing that an earthquake increased divorce rates by priming thoughts of instability), but it very much failed to replicate and is now the archetypal example of a formerly-accepted finding now believed to be false. Dr. Primestein’s work in this area is also a must-read.

Third, and from just last month: Artificial surveillance cues do not increase generosity: two meta-analyses. You know how if there was a picture of eyes or something, people would be nicer and more law-abiding, because deep down they felt like they were being watched? Yeah, turns out that’s not true.

And again, don’t trust me too much – there are a lot of studies I could have mentioned here, and three don’t necessarily make a pattern. But if I’m right that these are representative examples, then they seem to share a pattern. I’m not sure they all technically qualify as “social priming” (though they might, and that field has already been pointed out as particularly bad) but they all have the same feeling of tiny cues that you don’t think about causing big unconscious changes to behavior. This seems to be another category that is faring unusually badly.

Implicit association tests probably don’t work (1, 2, 3, 4). That is, people who have “implicit racial biases” according to the tests are not more racist in everyday life than people who don’t. If this were true – and if it reflected a general failure of implicit racial biases to affect explicit actions – it’s hard to overestimate how much it would change psychology. We wouldn’t have to worry about how the wrong character on TV would accidentally bias people toward having certain stereotypes. We wouldn’t have to worry about subconscious racism affecting hiring decisions even among people who are trying hard to be fair and neutral.

Does this fall into the previous patterns? It’s not exactly about self-fulfilling prophecies, or tiny stimuli having oversized effects on behavior. But it seems to have a certain kinship with them.

And a few days ago, a friend posted a quote on Tumblr:

There is no sovereign sanctuary within ourseles which represents our real nature. There is nobody at home in the internal fortress. Everything we cherish as our ego, everything we believe in, is just what we have cobbled together out of the accident of our birth and subsequent experiences. With drugs, brainwashing, and other techniques of extreme persuasion, we can quite readily make a man a devotee of a different ideology, the patriot of a different country, or the follower of a different religion

I was only too happy to be able to reply with Gwern’s research on how “brainwashing” mostly doesn’t work. Does this suggest the post is wrong about the lack of a “real nature”? Does this relate to any of the previous patterns?

A single thread seems to run through all of these examples: a shift away from the power of the unconscious. The unconscious doesn’t make you succeed or fail proportionately to your belief in yourself. The unconscious doesn’t change your behavior because of insignificant environmental cues. The unconscious doesn’t make you racially discriminate despite your own better nature. The conscious mind is strong enough to hold onto its preferred beliefs despite brainwashing techniques intended to force it otherwise.

So maybe we should update in general towards less of a role for the unconscious mind?

I remember my first freshman psychology class. After studying learned helplessness, I realized that was what I was feeling: study after study of crazy things, everything depended on your beliefs, the first letter of your name could affect your life outcome, stuff like that. The end result was a lot like in the quote above – nobody has any control over their lives, we’re all at the mercy of vast unconscious forces, society’s hidden assumptions and bases are vital in shaping our future. What if all of that was wrong? What if people mostly make decisions based on reasonable factors, succeed or fail based on things like ability or random luck, and social assumptions are relatively powerless beyond a common sense level? Wouldn’t that be great?

But maybe this is going too far. Once again, all I have is a few data points, curated by my own biases. Certainly not all of the studies showing creepy unconscious effects have been disproven; probably fewer than 10% of them have. Certainly not all the studies that have been disproven show creepy unconscious effects – ego depletion is a very mechanistic biological idea, but it’s done no better than priming. You have to kind of squint to see the pattern, then take it on faith that it’s real and that it continues throughout all the data that haven’t been checked.

And how do you tell the baby from the bathwater here? Some results – like cognitive biases, sales techniques or eyewitness unreliability – sort of fall under the heading of “power of unconscious effects”, but seem subtly different – maybe because of a less agentic unconscious? I don’t know. But I would be surprised if those followed the same pattern.

(on the other hand, a first draft included the Asch conformity experiments in that list, but apparently those never said what I thought they did)

Of course, this post is really about Freudian psychology. When I presented a Freudian friend with information on the general irrelevance of childhood factors and family composition, he countered by saying that the only thing that could really harm his belief in psychoanalysis was to learn that the unconscious wasn’t very powerful.

I don’t think we’re anywhere close to there yet. And I don’t think it’s meaningful to “deny the unconscious” – the unconscious is everything that happens except for conscious stuff, and that seems like a lot. But maybe our concept of the unconscious, or certain things that we attributed to the unconscious, was overly broad. And I think there’s an interesting project in trying to make explicit exactly what that means and what sort of smaller concepts we can get away with.