Monday 9 July 2018

We're Wrong About Basically Everything

Pavlov's dogs were a cultural touchstone of my youth.
Everyone knew the story
. Pavlov had a bunch of dogs. Before serving food, he'd ring a bell. Later, he observered that if he rang the bell they would start salivating, even if no food was present. The dogs were "conditioned" to have a physical response that is supposed to be associated with food in response to an audio cue.

This is part of the development of the theory of classical conditioning. By repeatedly artificially pairing stimuli in the environment, a thing with
a brain
can be made to pair those stimuli within their body. Turn on a light and puff air in face enough times and eventually I'll flinch when the light comes on even if the air puff doesn't come. The mechanism for this has a well understood neurological basis. Classical conditioning works. You can produce a behaviour by associating it with a reward and extinguish a behaviour by associating it with a punishment.

In the 1970s a new kind of "conservativism" rose in the English-speaking world. Maybe in the rest of the world too, but it certainly happened in the USA, the UK, Canada and Australia. There were people who were very concerned about the debts that were being run up by societies. They believed that being frugal and living within our means was an important value. They didn't want to pass down the growing debt to future generations.

But it wasn't just a complaint about debt, it was a new way of solving the problem. Thanks to a new school of economic thinking, they believed that they could reduce government deficits by reducing taxes. The notion, expressed by the "Laffer Curve" was that when taxes go above a certain threshold, the incentive to avoid taxes becomes too high. When taxes are lower, the reward for working is greater. Therefore, people will work more and pay more taxes. At the same time, social programs should be reduced, as they provide incentive to be lazy.

"People respond to incentives" is a key tenet of economics. This isn't meant to be a a grade-school simplication of economic theory. Try typing it into a search engine. You'll find articles written by serious people, you'll find free university courses, you'll find it's a real idea that real economists rely on to think about the world. Take continuing education classes at universities and you'll have economists telling you this as an important part of economics.

So drawing on simple conditioning, providing incentives, we eliminated national debts over the 1980s and stopped passing on the burden of debt to our children. Now people live within their means as individuals and as societies.

Okay, so that last part is the opposite of what happened. What went wrong?

Let's go back to Pavlov's dogs for a moment. First of all, it's time to hate Pavlov, because he didn't use bells, he used various stimuli including electric shocks. But I don't say that just to slag some guy who died close to 100 years ago. I say that because it's worth remembering that at the time there would have been no ethical guidelines that told you not to torture animals to do psychological experiments. I'm sure that in part there were no such guidelines because the whole idea of being an ethical scientist didn't really catch fire until the latter half of the 20th century. But that aside, people apparently just weren't horrified or repulsed by the idea of shocking dogs.
You have to wonder whether they
thought of dogs as sentient beings that had their own experiences.

That's sort of important to the whole process. In psychology and neurology there are people running into people making substantial chicken-egg problems out of neurological changes and psychological changes. The dogs can be described as being "conditioned" to associate two stimuli in a very mechanistic way. Then, when the knowledge of conditioning is used on human subjects and it works the same way, we might be tempted to write off our conscious thoughts on the matter as after-the-fact rationalizing of a conscious mind that believed itself to be the boss of the body.

Pavlov's experiments started because he noticed that dogs salivated when the people who were in charge of feeding them showed up. The white lab coat was a trigger that food was coming.
It's weird to me
that this would inspire someone to think of automaton-like behavioural modification when there is another considerably more obvious explanation: dog's aren't stupid.

Thinking about food makes you salivate. Writing about dogs salivating while thinking about food is making me salivate because I'm thinking about food. Unless the dogs were stupid beyond belief, it would be impossible for them to not recognize the people who feed them, to not know where their food comes from. Unless the dogs are essentially capable of no thought at all, they will think of food when they see people whose job it is to bring them food. The entire process is explained by saying, "Well, if someone rings the dinner bell at dinner time, then when the bell is rung, you will think, 'Oh, it's dinner time.'"

It's not just intelligence. It's also trust. Dogs don't just know that dinner comes when the bell rings. They know that the humans bring dinner when the bell rings. If the humans ring the bell and dinner doesn't come, the dogs will be confused. If it happens several times, they might start to think the humans aren't really all that reliable at delivering dinner. They might wonder if the bell meant dinner or if it meant something else and dinner just used to happen around the same time. They might be angry at the humans. And of course they might still salivate because one way or another they are thinking about dinner, even if it's just, "I remember we used to get dinner when we heard that bell."

I'm going off speculating that dogs would think of these interactions the way that humans would. I have no reason to believe that. Regardless of the actual mental processes of dogs, the point is that a dog doesn't simply react to a signal - or to a reward or punishment - it also has feelings about that signal, reward or punishment. And humans certainly do. The element of how you feel about the person offering you incentives plays a big role in how you choose to react to incentives.

The idea that we can condition humans to behave in a certain way by exposing them to stimulus gets far too much credit. Of course it can be done. But the moment the human becomes aware that another human is providing the stimulus for the purpose of conditioning them, all bets are off. The recipient of the conditioning then has feelings about whether they care to be conditioned or not. They may intentionally rebel just to prove a point, or they may start looking for ways to game the system to get the reward without doing the desired behaviour. They may do exactly what is expected of them, but only because they have thought about whether they want to or not and have decided the incentive is worth it. It becomes a contest between the conditioner and the conditionee and there is no guarantee at all that the conditioner is smarter or better at the contest.

It's a pretty good bet you can get someone to do something they don't really mind doing by paying them some amount of money. To extrapolate from that to the idea that you can generally control or predict human by providing the right incentives is idiotic and provably false.

I had
a friend
who grew up in India in the 80s. Corporal punishment of children was common; when you misbehaved you got the strap. This person told me that they and their friends got the strap every day. That was India in the 80s but it was also many other places in the world for huge swaths of history. The same "bad kid" got the same corporal punishment every single day, sometimes multiple times a day.

Every day they misbehaved, every day they were punished for it. Every day the administrators gave the strap, every day they got the same misbehaviour. Why didn't the disincentive of punishment stop the kids from misbehaving? Why didn't the disincentive of misbehaving kids stop the administrators from giving the strap?

Anyone who has paid attention to children know that's just not how it works. And we don't change that much when we grow up. We easily get struck in maladaptive cycles with one another instead of changing our behaviour to create a different outcome. We don't respond to the punishments and rewards that others try to impose on us, we emulate their behaviour, imposing punishments and rewards back on them. People do respond to incentives, they respond to incentives by learning that offering incentives is the right way for humans interact with one another. They emulate the incentive-offering behaviour, and regard one another as dogs, rats, insects or mould - however far down the mental-capacity ladder we have to go to get something that has behaviour that can actually be programmed with electric shocks.

So I have my own Pavlovian reaction. When I hear "people respond to incentives" I immediately roll my eyes and/or clench my fists in rage. It's utterly stupid. People who say, "people respond to incentives" can't use that information to successfully predict anything. It's an empty phrase that predisposes us to treat each other badly.