There is such a thing as a rapist type. A little over half of rapists are repeat offenders, with six victims on average. This group is also more likely to slap or choke people they have sex with, and to hit children. (And also to commit sexual assault, but at this point that's obvious.)

I understand there's some questionable statistics here, but I have to admit that that's reasonable enough that I don't care if it should really be one-third or something. My model predicts that there is indeed such a cluster. Doesn't mean they're "'predators disguising as people' until they shed their social pretense and let loose their inner evil upon unsuspecting victims." as DaFranker so helpfully put it.

The remaining group, of one-time rapists, probably matches your model.

They all match my model, except the actual psychopaths and such. Who, while probably over-represented among rapists, are by no means the norm.

I'm pretty sure that's false, assuming we're counting fuck-or-die situations (where both parties are being raped, anyway) and messing with meds as contrived. To stretch your metaphor horribly, until Obama was first elected he wasn't president, but he was the type of person who wants a political career and has positions that fit in a party's platform and can give good public speeches and raise money to campaign and so on, in the way that most people aren't.

That doesn't change the fact that rape, while it obviously selects somewhat, does not rewire you into a cartoon villain.

To take a N=1 sample, I can't think of a non-contrived situation where I would rape. What's more, when I model myself being born in the kind of environment that would lead me to rape, the person stops being recognizably me long before puberty. Whereas other unfortunately-raised mes remain me well into invading Poland or going postal or torturing heretics.

... and that doesn't indicate to you that your model may be inserting magical personality-rewrites into Evil Mutants? It sure as hell would to me. In fact, it has. Always proved right so far. If you can't empathize with them, you don't understand them. Might not work on aliens, but it sure as hell works on humans.

There are certainly some people, possibly more than I think (but way less than the whole of humanity), who can lose control. And as they're still the same people, once they regain control of themselves, they are horrified and atone as much as they can.

Or are you thinking of something like war, where (or so I heard) people go berserk and kill and rape indiscriminately? That I'll buy. (It does change people fundamentally, but doesn't replace them with evilbots.) But that's nowhere near what we're discussing here.

I wouldn't say they "kill and rape indiscriminately". They just kill and rape the enemy. It's not like they're people, right? Because if they were, that might make killing them wrong. Can't have our soldiers thinking that, can we? (I understand the US army, at least, has switched away from demonizing their enemies in Basic Training for precisely this reason.)

Wait, is the jump from "anyone who chooses to do this is evil, I do this, I know ways to stop doing this, I'm not taking them" to "I'm evil" far-fetched? Isn't that as basic as cognitive dissonance gets?

Calling people "evil" is tricky. There are a lot of conflicting metaethics and definitions floating around the issue. But I think it's fair to say that "not maximizing morality" is different from "evil". You ever spend money on chocolate? Fuck you. That money could have gone to charity. Saved lives. And you spent it on chocolate instead? Who the hell do you think you are, to put your pleasure above people dying? Of course, this goes for all nonessential expenditure. Humans don't maximize morality, because if we did we wouldn't be able to compete. We might want to, but if we somehow learned to defeat the layers of akrasia and bias and simple hypocrisy then we would no longer be human. Humans aren't FAIs, we're evolutionarily adapted to a specific niche.

You claim that I model rapists as Evil Mutants, but I don't know what you mean by that. Can you name one false prediction of my model?

Edited; previous version was:

It would help if you named one false prediction that my model of rapists as "Evil Mutants" makes.

4Nornagest8yThis is a complicated subject. To begin with, it's pretty hard to get more than a small percentage of soldiers to kill people at all: until after WWII, somewhere in the neighborhood of 80% of combatants in any given battle didn't fire their weapons, and the great majority of shots fired weren't aimed. Modern training methods aim to reduce this through associative conditioning, making training as realistic as practical, and a variety of other techniques that sometimes include dehumanization of the enemy. Now, there's a spectrum running from battlefield killing to full-blown atrocity, but atrocity's also got some unique features. If you're ordered to execute prisoners, for example, the prisoners end up dead or they don't: you can't shoot over their heads, or run ammunition or tend to the wounded instead of fighting, as you could in pitched battle. Because of this atrocity can be used as a tool of policy: soldiers who've committed war crimes have no choice but to justify them to themselves and each other, distancing them from their enemies and bonding them with a shared rationalization. This is bidirectional, of course; committing atrocity makes it easier to commit further atrocities, and a war where many gray-area cases come up (engaging enemy fighters in civilian clothes, for example, or mistakenly shooting a surrendering soldier) is one in which deliberate atrocity becomes more likely. The US army in the last few wars has tried very hard to draw a line, partly for PR reasons and partly because atrocity isn't well suited to recent strategic models, but the psychology involved doesn't lie entirely within institutional hands -- though institutions can of course exploit it, and many do.

Can the Chain Still Hold You?

by lukeprog 3 min read13th Jan 2012360 comments


Robert Sapolsky:

Baboons... literally have been the textbook example of a highly aggressive, male-dominated, hierarchical society. Because these animals hunt, because they live in these aggressive troupes on the Savannah... they have a constant baseline level of aggression which inevitably spills over into their social lives.

Scientists have never observed a baboon troupe that wasn't highly aggressive, and they have compelling reasons to think this is simply baboon nature, written into their genes. Inescapable.

Or at least, that was true until the 1980s, when Kenya experienced a tourism boom.

Sapolsky was a grad student, studying his first baboon troupe. A new tourist lodge was built at the edge of the forest where his baboons lived. The owners of the lodge dug a hole behind the lodge and dumped their trash there every morning, after which the males of several baboon troupes — including Sapolsky's — would fight over this pungent bounty.

Before too long, someone noticed the baboons didn't look too good. It turned out they had eaten some infected meat and developed tuberculosis, which kills baboons in weeks. Their hands rotted away, so they hobbled around on their elbows. Half the males in Sapolsky's troupe died.

This had a surprising effect. There was now almost no violence in the troupe. Males often reciprocated when females groomed them, and males even groomed other males. To a baboonologist, this was like watching Mike Tyson suddenly stop swinging in a heavyweight fight to start nuzzling Evander Holyfield. It never happened.

This was interesting, but Sapolsky moved to the other side of the park and began studying other baboons. His first troupe was "scientifically ruined" by such a non-natural event. But really, he was just heartbroken. He never visited.

Six years later, Sapolsky wanted to show his girlfriend where he had studied his first troupe, and found that they were still there, and still surprisingly violence-free. This one troupe had apparently been so transformed by their unusual experience — and the continued availability of easy food — that they were now basically non-violent.

And then it hit him.

Only one of the males now in the troupe had been through the event. All the rest were new, and hadn't been raised in the tribe. The new males had come from the violent, dog-eat-dog world of normal baboon-land. But instead of coming into the new troupe and roughing everybody up as they always did, the new males had learned, "We don't do stuff like that here." They had unlearned their childhood culture and adapted to the new norms of the first baboon pacifists.

As it turned out, violence wasn't an unchanging part of baboon nature. In fact it changed rather quickly, when the right causal factor flipped, and — for this troupe and the new males coming in — it has stayed changed to this day.

Somehow, the violence had been largely circumstantial. It was just that the circumstances had always been the same.

Until they weren't.

We still don't know how much baboon violence to attribute to nature vs. nurture, or exactly how this change happened. But it's worth noting that changes like this can and do happen pretty often.

Slavery was ubiquitous for millennia. Until it was outlawed in every country on Earth.

Humans had never left the Earth. Until we achieved the first manned orbit and the first manned moon landing in a single decade.

Smallpox occasionally decimated human populations for thousands of years. Until it was eradicated.

The human species was always too weak to render itself extinct. Until we discovered the nuclear chain reaction and manufactured thousands of atomic bombs.

Religion had a grip on 99.5% or more of humanity until 1900, and then the rate of religious adherence plummeted to 85% by the end of the century. Whole nations became mostly atheistic, largely because for the first time the state provided people some basic stability and security. (Some nations became atheistic because of atheistic dictators, others because they provided security and stability to their citizens.)

I would never have imagined I could have the kinds of conversations I now regularly have at the Singularity Institute, where people change their degrees of belief several times in a single conversation as new evidence and argument is presented, where everyone at the table knows and applies a broad and deep scientific understanding, where people disagree strongly and say harsh-sounding things (due to Crocker's rules) but end up coming to agreement after 10 minutes of argument and carry on as if this is friendship and business as usual — because it is.

But then, never before has humanity had the combined benefits of an overwhelming case for one correct probability theory, a systematic understanding of human biases and how they work, free access to most scientific knowledge, and a large community of people dedicated to the daily practice of CogSci-informed rationality exercises and to helping each other improve.

This is part of what gives me a sense that more is possible. Compared to situational effects, we tend to overestimate the effects of lasting dispositions on people's behavior — the fundamental attribution error. But I, for one, was only taught to watch out for this error in explaining the behavior of individual humans, even though the bias also appears when explaining the behavior of humans as a species. I suspect this is partly due to the common misunderstanding that heritability measures the degree to which a trait is due to genetic factors. Another reason may be that for obvious reasons scientists rarely try very hard to measure the effects of exposing human subjects to radically different environments like an artificial prison or total human isolation.

When taming a baby elephant, its trainer will chain one of its legs to a post. When the elephant tries to run away, the chain and the post are strong enough to keep it in place. But when the elephant grows up, it is strong enough to break the chain or uproot the post. Yet the owner can still secure the elephant with the same chain and post, because the elephant has been conditioned to believe it cannot break free. It feels the tug of the chain and gives up — a kind of learned helplessness. The elephant acts as if it thinks the chain's limiting power is intrinsic to nature rather than dependent on a causal factor that held for years but holds no longer.

Much has changed in the past few decades, and much will change in the coming years. Sometimes it's good to check if the chain can still hold you. Do not be tamed by the tug of history. Maybe with a few new tools and techniques you can just get up and walk away — to a place you've never seen before.