The Psychology of Punishment
I was having a discussion with a friend, over e-mail, concerning the value of spanking as a training tool, which branched out into a wider discussion of punishment and criminal justice. Yesterday, I wrote this enormous reply on the subject, putting together my experience with all of the theory and research I've read on the subject. I'm feeling terribly satisfied for having finally articulated this, so I decided to share it with a slightly wider audience. It's a long read, really an eight-page essay, but perhaps some of you may find it useful and interesting. Commentary and questions welcome.

> now, with regards to spanking as an activity, i think it take a
> different view to it from you. i'm not sure where the difference
> comes from, but...

::cracks knuckles::
::stretches typing fingers::
::grins::
All right, here goes...

> i suspect you'll give me good reasons it is Wrong(tm), but what i
> really need is demonstrations that it is ineffective. i don't
> care much about what is right and wrong -- just what works and
> what does not work.

This is snipped from further on, but I think it's really the central point of contention.

For starters, I know you've probably gotten that kind of response before, but I don't think I'm usually that much of an irrational moralist -- I'm still very much an evolutionary psychologist at heart!

There is certainly an ethical dimension here, but I consider good ethics to be primarily determined by good psychology. Reward-and-punishment as a primary training method is very poor psychology. There are reasons why I think spanking is a particularly unacceptable form of punishment, and I'll get to those later, but the thing that ultimately makes it so vile to me is that punishment like this isn't even all that effective as a training tool. Thus it becomes unnecessary cruelty.

Please note: For brevity's sake (or at least to keep it readable), I'm not going to explain all of the background psychology I'm leaning on. If at any point you don't follow my logic or accept the psychology I'm using, please ask me for clarification; I can probably elaborate to your satisfaction.

> i was spanked once as a child. once, and never again.
> and i never forgot the experience, even though i've long since
> forgotten what it was for. it made a profound impact.

Nearly everyone I've asked who was spanked as a child says this same thing -- that they can still remember the experience, but can't remember what provoked it. Spanking makes a profound impact, yes, but I think this detail is a telling commentary on what kind of impact it's making. Because ultimately, the lesson it was supposedly intended to teach is not the one that's being remembered.

The behaviorist view is tempting in its simplicity and obviousness, but taken alone, it's a gross oversimplification of human psychology -- and, I'm guessing, the psychology of any truly social animal, puppies included. Used as a training method, reward-and-punishment is crudely effective, but lacks the nuancing capabilities of other methods, and brings with it a host of unintended consequences, some of which can negate the intended effects.

The first practical problem with the reward-and-punishment model is that it's maddeningly imprecise. It works by creating an association, linking a moment or situation with strong positive or negative feeling. Because of the expansive way the human mind forms associations, this effect cannot be focused; the feeling becomes linked not only with the target behavior, but with the whole sphere of activity, and often with various incidental contextual elements as well. The more often the punishment is repeated, the stronger its effects, and the more global its associations become.

To take an example from my own experience: When my grandparents finally bought a stereo system, my grandmother began turning the radio to her favorite station while she was cleaning the house. My grandfather disliked her choice of music, and, being a curt bastard, he reprimanded her several times for putting that station on. The result? After a few tries, he succeeded; she never put that radio station on again, even when he was out of the house. In fact, she never turned the stereo on at all. She never touched the stereo cabinet again, even to dust it.

It's an extreme example, but I suspect that if you think about it, you'll be able to come up with similar stories from your own life.

Sure, punishment can still be a fairly effective deterrent, and it certainly has its place as an emergency method of curbing highly problematic behavior. If you just want to, say, prevent your kid from trying to climb the insanely hot radiators, and for some reason there's no way to physically separate them from the radiators, go ahead and yell at them each time they do it -- as long as you're willing to accept that they'll also become reluctant to go near the radiators, climb things when you're looking, or play with whatever object they were trying to reach in the first place, and that after a while they may get strangely uneasy when you put on that particular cologne.


This is equally true, incidentally, for rewarding behavior. I put less stress on this because positive associations generally have less problematic collateral-damage potential than negative ones, but things can get wacky even here. And lest you wonder, following punishment with reward for correct behavior will reinforce the training, and can reduce some of the blanket aversion that a purely punishing method will create, but it doesn't necessarily fine-tune the associations any better, and it does encourage dependency, as I'll explain later.

And all of that is assuming the reward and punishment follow fairly swiftly on the heels of the provoking incident. They're often used quite differently:

> sometimes you can't be there
> to do it at the time, so the method used is simple and arguably
> brutal: show the dog the offending excrement, paddle them, and
> then absolutely do not forget to praise them when they
> demonstrate correct behavior next time!

If it's more than a few minutes after the fact, the effectiveness of reward or punishment becomes laughable. This dog isn't forming negative associations with the act of defecating on the floor, they're forming negative associations with the experience of having their owner haul them by the collar to a pile of excrement in the living room and shove their noses in it. This will likely cause them to avoid you the next time they (or another household dog) have an accident in the living room, but it won't housebreak them in any direct way.

Now, it could be argued that there is some associative usefulness in making the dog smell the excrement and then immediately pushing them out the back door, but the paddling is adding nothing to this particular situation. The persistence with which people apply this tactic is a testament to our stubborn traditionalism, deep dislike for excrement, and unwillingness to accept that we are powerless in the face of any given unpleasant incident.

> cruel and unusual? yes, but it turns out well house-
> broken dogs quickly, and while other forms of training are tried
> perennially, nothing seems to be as effective as this, and the
> dogs don't appear to be screwed up by it. as long as this is
> applied consistently, it works.

I'd be curious to know if you've done the research on this -- it sounds suspiciously like the kind of commonplace wisdom that has minimal grounding in actual data. I honestly don't know what kind of other methods of dog training have been tried, or what problems they had, but I know of a similar example from horse training. That example also provides an eloquent caution about assuming that an animal is not being "screwed up" by a training method when you've never seen an example of an animal who was trained by other methods.

Horses, thoroughly social animals by nature, have traditionally been broken to the saddle through a fairly rigorous reward-and-punishment system. It is crude and cruel but highly effective, and no one questioned it for a very long time. Not all that long ago (see the film "The Horse Whisperer"), someone began to try a gentler method, one that adapted to the communicative and social cues that horses naturally respond to. The horses accepted the saddle just as quickly (if not more so), and with less fuss -- and the resulting horses were easier to work with, healthier, and far less nervous and inclined to spook at shadows. The new method wasn't just more humane -- it ultimately produced a more useful horse.

Still, there are plenty of horse trainers who still insist on using the traditional methods. This is partly because of the standard conservative logic that "this is the way it has always been done, and it's worked just fine so far," and partly because it's just less effort for them. It's always easier for people who've been trained by reward and punishment to think in those terms. Adopting a new training method completely enough to be effective involves a high degree of thoughtfulness and conscious planning, and a willingness to curb old knee-jerk reactions. Otherwise, a person just slips back into their old patterns the minute something makes them tense, and the resulting inconsistency means they're not using either method effectively. It's easy, at that point, to decide that the new method just doesn't work, when it was never really given a chance.

And that's all just my first point.

The second major problem with the reward-and-punishment model is that it doesn't encourage specific behavior nearly so much as it encourages kids (or puppies, or adults) to try to please authority figures.

You see, one of the contextual elements that is inevitably linked with the positive or negative feeling is the person who's delivering the reward or punishment. I have heard spanking advocates say that they are trying to create an impersonal cosmic cause-and-effect feeling, and it's a nice thought, but it's entirely unrealistic. Even a puppy knows who's paddling it, and you'd better believe the puppy is rapidly learning what kind of an owner it's got, and what kind of relationship it has with that owner.

Absolute consistency in meting out reward and punishment is simply impossible, unless you have a cult-like situation where everyone's life is under close 24-hour watch by people who have nothing better to do. Otherwise, there will always be incidents that occur unnoticed, and times when the authority figure has more pressing things to attend to. So the association between an act and the parental response will never be completely consistent -- but the parent will be there every time reward or punishment is meted out, so that association becomes by far the strongest and the most consistent.

The thing a child learns fastest from this treatment is that if they piss off mom or dad, life gets unpleasant, and if they please mom or dad, they get cookies (or praise, or whatever). Figuring out what behavior actually pisses off or pleases mom and dad becomes a secondary task, in which the behaviors themselves have little intrinsic merit beyond their effect on mom and dad. Exclusively used, a reward and punishment system turns out a dog (or child, or adult) who is intensely aware of the approval or condemnation of authority figures, and chooses their behavior primarily based on how they think an authority figure will react.

In the best-case scenario, where reward-and-punishment is used effectively and consistently, such a dog (or child, or adult) will toe the line admirably, even compulsively, always watching the authorities to make sure someone is noticing, and sometimes getting confused and anxious if no reward is forthcoming for previously-rewarded good behavior. We might be perfectly happy to have this quality in our dogs, even if it does make them weirdly dependent (and I've met a few stunning examples), but this is not the kind of person most of us want to produce when we're raising our children. This is, incidentally, the training method preferred by most cults. It discourages thinking while encouraging rigid conformity, and overwrites prior conditioning with remarkable haste.

If the method is used exclusively but inconsistently, as is almost inevitable, it turns out a dog (or child, or adult) who will toe the line when under watch, misbehave when they think they can get away with it, and rebel as soon as they think they're big enough and powerful enough to buck the authority successfully, and then later turn into clones of the authority figures who trained them (sound familiar? You probably know a few dozen of these). This is a less desirable result from a training perspective, though it may sound more desirable from a personal-independence perspective. But more importantly, this person (dog, child) is still choosing their behavior purely by the response of authority figures -- they're just not necessarily going for a positive response. Thinking and independent choice is not being encouraged here, and neither is any real internalization of the values that are being imposed.

So, you may be wondering why, if this is the case, everyone you know doesn't fit that description. The truth is, it's very rare for the reward-and-punishment method to be used exclusively; we tend instinctively to mix in training methods that work better with our psychology. Thus, the effects of the strict reward-and-punishment model will usually show up in diluted form, visible but not definitive.

You see, the major facet of our psychology that behaviorists overlook is our overwhelming desire to integrate ourselves socially. Kids (and adults, and yes, even puppies) really want to be loved and respected and responsible and trustworthy and anything else their peers value, and it doesn't take a deep knowledge of evolutionary psychology to see why. We want to internalize our culture's values, we will internalize our culture's values, and we don't have to be hit with sticks to do it. The real question is how we will conduct ourselves with regard to those values.

The problem is, not only do we have to learn what our culture wants from us, we also have to learn what we are, and how we can fit that together with our social standing. A person isn't born knowing their own temperament, their own aptitudes, their own predispositions -- indeed, a person can spend a lifetime figuring themselves out. But a baseline of self-understanding is necessary for an individual to learn how to successfully adapt to the social group in which they find themselves. We are incessantly hungry for information about ourselves, how others perceive us, what we're good at, how we think and behave as seen through others' eyes.

And when we are given feedback, data on ourselves, we internalize it, use it to shape our self-images, and adapt our behavior accordingly.

This provides us with our single most powerful tool for teaching and training children (and adults, when necessary). It's even possible to use on dogs and other social animals, albeit much harder, since we can't communicate as directly.

When you want to teach a child to share, you reinforce their self-image as a person who shares. If they're not a very generous sort of child, you wait for any incident that could possibly constitute sharing, and make sure to drop a passing comment to the effect of, "you're such a generous person, Jimmie," or, "why, that was a very generous thing to do." It's sort of a form of praise, but it doesn't have to be gushy or even glowing, and it doesn't have to come with a cookie -- in fact, it works better if it doesn't. It just has to give them a glimpse of themselves as you want to see them, and they will begin to live up to it.

And they will be doing it more or less for its own sake, because they believe they're the kind of person who does that for its own sake. They will have far more interest in the activity itself. They will cheerfully do it when no one is looking. They will internalize the values as their own. (Obviously there are other factors involved as well, but this is the general effect.) And what's more, if you reinforce their self-image as a person who thinks and does what is right -- eventually, they will learn to think, and then do what they think is right. That is something a reward-and-punishment model will never be able to teach them to do.

The tough part about this, for most people, is responding to unwanted behavior. The single fastest way to eliminate misbehavior is to fail to acknowledge its existence, while busily reinforcing the corresponding positive character traits. Alternatively, you can respond to a negative instance with a comment such as, "Well, Jimmie, that's strange. You're a very kind boy. What happened that made you want to steal Joey's toy?" The key is not to use "you're such a bad boy" language, or anything that implies it, because if you tell him he's a bad boy, he'll diligently become one.

Seen through this lens, most conventional forms of punishment simply reinforce the child's self-image as a person who misbehaves. That comes loaded with negative associations, so they won't feel good about seeing themselves that way, but that won't change the basic self-image, it will only lay a nice layer of poor self-esteem over it. And poor self-esteem is just great for encouraging good behavior. [Insert sarcasm here.]

Oh, and while I'm at it:

> as it is, we as a society are doing a great job of turning
> out career fuckups because we don't punish bad behavior early
> enough with reasonable force (mandatory sentencing is a disaster,
> ask me about that sometime...) and we don't reward good behavior
> either.

I'll agree with you on the lack of reward, but otherwise, this is nonsense.

We as a society are doing a great job of turning out career fuckups because we persistently reinforce their self-images as troublemakers, we make it difficult for them to rejoin mainstream society, we don't give them alternative ways of dealing with the circumstances they find themselves in, and we don't give them any reason to respect the law enforcement system or to want to internalize the self-stated values of the institutions that promote it. And meanwhile, we demonstrate quite clearly by our actions what our real values are.

That brings me to my third point about why punishment as a training tool is very poor psychology, particularly when it is cruel. Kids (etc.) learn how to behave first and foremost by watching how the people around them behave. They learn how to treat people by watching how the people around them treat people. They pick up on subtleties to a remarkable degree, and they will do as you do long before they will do as you say. This includes picking up on the punishing paradigm itself.

If you convey, through your actions toward the child, that the way you deal with someone doing something you don't like is by yelling at them, the child will yell at their friends when a game isn't going they way they like. If you employ public humiliation, the child will learn to humiliate their peers when they're unhappy with them. And if you strike the child, the child will learn to use physical violence to get their way.

Now, I will certainly grant that the potential for this behavior is already present, and you'll probably see young children express it from time to time on their own. But the potential for systematic genocidal rage is also present, and most kids don't express that very much. Kids (and adults, and yes, puppies) are remarkably good mimics, and they learn very fast.

That means that if you spank a child, you're teaching them, loud and clear, that the proper way to deal with people whose behavior you don't like is through violence and humiliation.

Of course, the child will quickly learn that you don't spank your friends or co-workers or even older children, and being very sensitive to such nuances, they will dutifully register that you only use violence and humiliation on people who are much weaker and less powerful than you. The next time they see a child on the playground who is a few years younger than them, they'll know what to do.

Think about it. Within adult civil society, there are acceptable forms of punishment for unacceptable behavior. Withdrawal of social connections is the most universal, followed by withdrawal of privileges, and temporary semi-isolation is also considered fairly civil. Violence is considered a last resort, and always within the context of a two-way fight.

When, in the adult world, are you allowed to pummel someone without them having the right to defend themselves? The only example I can think of is prison.

And children will notice this, and will also dutifully recognize that they, who are in precisely this situation, are clearly thoroughly powerless people, disrespected by their authority figures, with no rights over their own bodies. This will certainly not help them build a positive self-image, and it will also lead to them treating people accordingly when, at last, they are in positions of power.

The more cruel your punishments are, the more cruel you are teaching the victim to be. The implications of this in law enforcement are obvious.

I trust I make my point.

*grin*

That was fun.

Commentary welcome, and indeed invited.