Also, with punishment generally, there’s a problem that people almost inevitably overestimate its effectiveness. Punishment generally follows exceptionally bad behavior, exceptional behavior is, obviously, exceptional, so punishment would be expected to be followed by behavior which is not exceptionally bad just because of regression to the mean, even if punishment were totally ineffective or even mildly counter-productive. But, unfortunately, people are almost totally oblivious to regression to the mean, and so what should be the expected result regardless becomes for them evidence of the effectiveness of punishment.
Because punishment tends to immediately terminate the punished behavior, while not preventing it in the long run, punishment reinforces the punisher. The person doing the punishment is an agent under conditioning as well, but most folks fail to notice this.
I don’t agree with this, but rewarding is clearly better, of course. It’s easy to get tunnel vision when thinking about punishment, many of us have memories of injustice done in its name.
What do you mean by this? Rewarding everyone who fails to exhibit bad behavior? This has two problems:
1) Since most people behave well most of the time, this gets expensive quickly.
2) Human psychology is such that if there is a regularly given reward, people will simply readjust their baseline and thus will perceive being deprived of it as a punishment anyway.
I think you could have done the steelmanning yourself, but here you go:
1) Rewards can be social, in other words virtually free. You don’t have to reward all good behaviour, in fact that probably makes it less effective. (like you already said yourself?)
2) Here’s how I would do it: Always reward exceptionally good behaviour, sparingly reward ordinarily good behaviour.
This doesn’t mean punishment should be never used, but it’s difficult to build a positive relationship with someone you’re punishing constantly.
Punishment generally follows exceptionally bad behavior, exceptional behavior is, obviously, exceptional, so punishment would be expected to be followed by behavior which is not exceptionally bad just because of regression to the mean
I don’t think treating human behavior as a simple random variable is a good model. See here for a better model.
If you have a point to make, I think it can be made more effectively than “Read this article”.
I can identify behaviors that please me more than others, creating an ordinal structure on the set on the set of possible behaviors. I can also observe a frequency distribution of those behaviors. From the frequency distribution and the ordinal structure, I can identify a median. From there, it’s not too difficult to identify reasonable assumptions such that the frequency of a bad behavior being followed by a worse behavior is less than the frequency of a bad behavior being followed by a better behavior, where “bad behavior” is “behavior that is worse than the median”.
Also, with punishment generally, there’s a problem that people almost inevitably overestimate its effectiveness. Punishment generally follows exceptionally bad behavior, exceptional behavior is, obviously, exceptional, so punishment would be expected to be followed by behavior which is not exceptionally bad just because of regression to the mean, even if punishment were totally ineffective or even mildly counter-productive. But, unfortunately, people are almost totally oblivious to regression to the mean, and so what should be the expected result regardless becomes for them evidence of the effectiveness of punishment.
Because punishment tends to immediately terminate the punished behavior, while not preventing it in the long run, punishment reinforces the punisher. The person doing the punishment is an agent under conditioning as well, but most folks fail to notice this.
Upvoted for the insight.
I don’t agree with this, but rewarding is clearly better, of course. It’s easy to get tunnel vision when thinking about punishment, many of us have memories of injustice done in its name.
What do you mean by this? Rewarding everyone who fails to exhibit bad behavior? This has two problems:
1) Since most people behave well most of the time, this gets expensive quickly.
2) Human psychology is such that if there is a regularly given reward, people will simply readjust their baseline and thus will perceive being deprived of it as a punishment anyway.
I think you could have done the steelmanning yourself, but here you go:
1) Rewards can be social, in other words virtually free. You don’t have to reward all good behaviour, in fact that probably makes it less effective. (like you already said yourself?)
2) Here’s how I would do it: Always reward exceptionally good behaviour, sparingly reward ordinarily good behaviour.
This doesn’t mean punishment should be never used, but it’s difficult to build a positive relationship with someone you’re punishing constantly.
Of course, that means that unless you also punish bad behavior, it won’t stand out from the ordinary good behavior.
It’s also difficult to build a positive relationship with someone who is constantly engaging in bad behavior.
This is actually the biggest problem with torture, in my opinion.
I don’t think treating human behavior as a simple random variable is a good model. See here for a better model.
If you have a point to make, I think it can be made more effectively than “Read this article”.
I can identify behaviors that please me more than others, creating an ordinal structure on the set on the set of possible behaviors. I can also observe a frequency distribution of those behaviors. From the frequency distribution and the ordinal structure, I can identify a median. From there, it’s not too difficult to identify reasonable assumptions such that the frequency of a bad behavior being followed by a worse behavior is less than the frequency of a bad behavior being followed by a better behavior, where “bad behavior” is “behavior that is worse than the median”.
But Eugine made a point and his point was:
He then backed up his point to give context to suggest what a better model might be, i.e. one that models a human as a temporal process with habits.