Jun 2, 2011

Wolves, incentives, and maybe even parenting

Like any good economist, I train my dog using incentives. Lots of incentives. This includes treats, of course, but also belly rubs and a healthy dose of gushing remarks about how adorable he is.

In some sense, the incentives have worked wonders. Homedog can do all kinds of tricks, including but not limited to fetching a can out of the fridge *and* shutting the door behind him. But here’s the thing: In the areas where it matters most, like listening and obeying in dangerous situations, nobody who knows him would call Khan the best-behaved of canines. Strangeness is his vice: He’s not good in strange situations or around strange people. He’s not even that good on a leash.

Naturally, I was humiliated when I read Mark Rowlands’s account of how he trained a wolf. It’s been said that you can’t train a wolf to walk on a leash. Mark Rowlands did. More impressively, he trained the wolf to walk by his side off the leash. More impressively still, he trained the wolf to walk by his side off the leash in a field of sheep(!).

Obviously I’m doing something wrong. I can’t even train Khan to walk on a leash in a field of squirrels.

I think the fact that a wolf can be trained to walk calmly and passively in a field of sheep says that we are missing something important if we are conceptualizing behavior purely in terms of incentives. The wolf has a huge incentive to go and get him some free and easy sheep flesh, so what’s preventing him from doing so? I cannot imagine that it would be treats, no matter how delicious, nor punishments, no matter how cruel. Nor can I imagine that it has anything to do with a desire to please. Something else is going on here.

We economists tend to think along behaviorist lines, believing that if you reward a behavior you get more of it, and if you punish a behavior you get less of it. There is a lot of truth to that, but I’m gradually realizing that it’s far too simple. Dangerously simple. Like holy-shit-I-really-need-to-rethink-things simple.

Here’s Mark Rowlands in The Philosopher and the Wolf:

It is a mistake to think that your dog’s obedience can be obtained by rewards. The rewards can take different forms. Some people obsessively pop treats into the mouths of their dogs for accomplishing even the easiest of tasks. The most obvious result of this is a fat dog that will refuse to obey its owner when it suspects there is no treat around to be offered, or when it is distracted by something – a cat, another dog, a jogger, etc. – that it deems more interesting than a treat.

More often, however, the ‘reward’ takes the form of an inane chatter they insist in carrying on with their dog. ‘Good boy’ … ‘What a clever dog you are’ – and so forth. And they often accompany this chatter with nagging little tugs on the lead to, as they see it, help reinforce their message. This is, in fact, precisely the way not to train a dog – and it hasn’t a snowball’s chance in hell of working with a wolf. If you’re continually talking to your dog, or half-heartedly tugging on his lead, he has no need to watch you. In fact, he has no reason to give a fig about what you’re doing. He can do what he likes in the sure knowledge that you will let him know what’s happening – and that he can act on or disregard this information as he chooses.

The problem with this approach is that the dog has his own desires:

People who think that their dog’s obedience can be bought are people who think – and how often have I heard this – that their dog basically wants to do as his ‘master’ wants – he always aims to please – and simply needs to have explained to him precisely what this is. And this is, of course, non-sense. Your dog doesn’t want to obey you any more than you want to obey anyone else.

Rowlands doesn’t mention this, but apart from the flawed logic of incentives as teaching devices, psychologists have long known that incentives can be are counterproductive. For example, if you amply reward a kid for reading, the kid will, as expected, read more (at least for a short while). But the perverse part is this: In the absence of the reward, the kid will read even less than they did at baseline.

So what do we do if not layer on the incentives? Rowlands explained that the opposite approach – training through ego, a battle of wills – is equally as flawed because it will result in a dog that will grow up, in all likelihood, to be not very nice at all.

The key to training, he says, is this:

Make him think he has no choice in the matter. This is not because he is made to feel the loser in a battle of wills, but because of an attitude of calm but remorseless inevitability that you must bring to your training. In a battle of wills, what you are saying to the wolf is this: You will do what I say – I am giving you no choice. But the attitude with which to train a wolf is this: You will do what the situation demands – this situation affords no other option. It is not I to whom you are responding; it is the world.

Maybe it’s scant consolation for the wolf. But it certainly helps put the trainer in his or her proper place – not as a dominant and arbitrary authority whose will is to be obeyed at all costs, but as an educator who allows the wolf to understand what the world requires of it.

This makes so much sense, and I hate that it wasn’t until now that I heard this articulated. Ideas are flitty, of course, but this seems like a rare example of an idea that could, with time and attention, be life-altering, changing not just my conscious thoughts but my behavior and my perceptions.

The way that I understand what Rowlands is saying is that a force that shapes behavior more powerfully than tastes, preferences, incentives, desires, demand functions and other such things economists like to talk about is – not sure how to put this – something like an acceptance of or a submission to constraints. For example, the reason why people remain loyal to spouses, friends, jobs, whatever, is probably often not because they have (unconsciously) concluded that the benefits outweigh the costs, but because they have accepted it as duty. You could say that there is a demand function for duty, but I think you'd be missing the point.

There are important lessons to be learned here. People will probably scoff at my saying this, but it seems that the principles of being an effective dog/wolf trainer also apply to being an effective parent, teacher, manager, whatever. Whenever you are in a position of authority, you'd do better not to act as a reward-giver nor as a dominant and arbitrary authority, but as a calm and remorseless educator, teaching what the world demands of us.

And lest you have questions about the morality of such “training,” Rowlands has a response:

This training was the greatest gift I ever gave Brenin [the wolf] – a shining example of one of the few things in my life I really did right. Some people think that training dogs – and, even more so, wolves – is cruel, as if you are going to break their spirit or make them permanently cowed. But far from breaking his spirit, when a dog or wolf knows exactly what is and is not expected of him his confidence, and as a result his composure, grow immensely.

It is a hard truth that, as Friedrich Nietzsche once put it, those who can’t discipline themselves will quickly find someone else doing it for them. And, for Brenin, it was my responsibility to be that somebody. But the relation between discipline and freedom is a deep and important one: Far from being opposed to freedom, discipline is what makes the most worthwhile forms of freedom possible. Without discipline there is no real freedom; there is only license.

Amen.