Perhaps the most enlightening psychology experiment ever done, establishing what amounts to one of the few laws of nature in psychology, involved the punishment of lab rats. You set up a conditioning routine where you get a rat very interested in pressing a lever (by reinforcing it intermittently). After establishing the routine, you simply unhook the lever from all consequences that might interest the rat, and the rat presses the lever 100 times (a lot at first and a little as time goes on) before giving up and losing all interest in the lever. You know now how to create an interest in lever-pushing that is quantifiable: 100 presses to extinction.
Now, instead of doing nothing after establishing the routine, you punish the rat for pressing the lever (shock it, for example, by electrifying the lever); then you disconnect it from all relevant consequences as before. For a while, the rat avoids the lever, just as you avoid a lamp that has given you a shock. After a while, the association between the shock and the lever wears off, and the rat tries the lever again, finding that there is no longer a shock but, as before, neither is there any reinforcement associated with pressing it. How many times will it press the lever? One hundred times, just as before. This time, it goes slow at first with caution, then fast, and then tapering off to extinction.
The reliability of this phenomenon demonstrates that punishment does not change the tendency to engage in the behavior that was punished. Instead, it makes the person or the rat want to avoid the source of punishment. As soon as the child thinks it’s not being watched (as soon as the situation seems different in some way), the tendency to engage in the behavior will reassert itself. Punished children do what was punished behind their parents’ backs, or as soon as they get to college. Sure, I suppose you could arrange for a totalitarian state to ensure that the person is always feeling watched and thereby inhibit the behavior permanently under an umbrella of anxiety, numbness, and hate (the emotions that punishment produces). But even then, the tendency (or desire) to engage in the punished behavior will not change.
Sometimes, of course, punishment is necessary, like when you stop a child from running into a busy street. But if you want it to stick, you have to reinforce a behavior that competes with running into the street (like stopping and waiting for the light). You cannot count on punishment alone, or your kids will run into the street when you are not with them.
Why, then, do we punish children? We do so for two main reasons. The first is that punishment looks like it works even though it doesn’t. Because the child is inhibited in your presence, it’s easy to think they would be inhibited in your absence. Punishment produces politeness, not morality. Thus, the inhibited, obedient child inadvertently reinforces the parent’s punitive behavior by acting obedient (for the sorts of parents who find obedient children reinforcing).
The second main reason we punish children is because we are angry at them. Anger is the emotional state of finding damage to the object of one’s anger reinforcing. Uncontrolled and disobedient children make everyone angry some of the time and most people angry much of the time. Punishment makes them cry or look upset, and those tears and that look are reinforcers when parents are angry. (For most parents, too much damage to the child would be aversive.) If we weren’t so confused about aggression, we could acknowledge our anger at our children and engineer a constructive application of it instead of pretending we are not angry and letting it loose as punishment. Instead, we fool ourselves into thinking that we are not angry at our children, that we are merely instructing them, and we then get to enjoy hurting them with punishments without feeling bad about recognizing what we are enjoying.