An interesting idea in this book was that most people who try to do “behaviorism” are actually doing it wrong (such as delivering the rewards and punishments too late, so they are associated with something else instead). And yet the people who do it wrong will defend their approach as scientifically proved.
As a rule of thumb, if your approach uses a lot of punishment, chances are that you are actually only rewarding yourself (by giving yourself a feeling of high status when you deliver the punishment). Which is the true reason why approaches using punishments are so popular (among the people who use them).
What about the claims in “Maintaining behavior” that you do need consistent aversives (punishment), but only inconsistent rewards? That seems to say the exact opposite of the earlier stance: it says that you should use lots of punishments (every time the subject gets something wrong), and few rewards.
I’m confused as to what the book actually wants you to do.
One of the chapters deals with getting rid of behaviors you don’t want, with eight methods (some of which she doesn’t recommend). For example, training an incompatible behavior: if don’t want your dog to beg at the table during dinner, train your dog to lie down someplace else during dinner. Or “shape the absence”—reinforce everything that’s not the unwanted behavior.
I’m confused as to what the book actually wants you to do.
I recommend reading the book. It is one of the best things I have ever read. A short review or a comment cannot explain everything. (Also, I don’t fully remember everything; it was a few years ago.)
First, there is a difference between teaching a behavior, and unteaching a behavior. (Is “unteach” a proper English word?) Second, there is a difference between creating a new habit, and maintaining the existing habit.
On the topic of unteaching, the important thing is that instead of “don’t do X” it is often easier to teach an alternative “in situation Z (instead of X) do Y”. But if you want to use punishments, the important thing is that they come immediately and consistently. A small punishment that comes always and immediately after the act, works much better than a large punishment that comes only sometimes and several days after the act.
When you start teaching a new habit using rewards, again the important thing is to deliver the reward immediately and consistently… at the beginning. But after the habit is established, you gradually reduce the size and frequency of the rewards. If you stop rewarding suddenly, the animal will give it a few more attempts, and then give up… and maybe occassionally try again, just to see if the rewards have returned. But if you gradually make the rewards rare, the animal will keep doing it, and the occassional reward will be enough to keep the habit. If you deliver the rare rewards regularly (e.g. only once a day, or always once per 100 attempts), the animal will notice the regularity, and after receiving a reward will slow down (because it means that the next reward is far away). But if you deliver the rare rewards unpredictably, the animal will keep trying all the time.
An interesting idea in this book was that most people who try to do “behaviorism” are actually doing it wrong (such as delivering the rewards and punishments too late, so they are associated with something else instead). And yet the people who do it wrong will defend their approach as scientifically proved.
As a rule of thumb, if your approach uses a lot of punishment, chances are that you are actually only rewarding yourself (by giving yourself a feeling of high status when you deliver the punishment). Which is the true reason why approaches using punishments are so popular (among the people who use them).
tl;dr—all you need is love (and clicker)
Concise, to the point, love it.
What about the claims in “Maintaining behavior” that you do need consistent aversives (punishment), but only inconsistent rewards? That seems to say the exact opposite of the earlier stance: it says that you should use lots of punishments (every time the subject gets something wrong), and few rewards.
I’m confused as to what the book actually wants you to do.
One of the chapters deals with getting rid of behaviors you don’t want, with eight methods (some of which she doesn’t recommend). For example, training an incompatible behavior: if don’t want your dog to beg at the table during dinner, train your dog to lie down someplace else during dinner. Or “shape the absence”—reinforce everything that’s not the unwanted behavior.
I recommend reading the book. It is one of the best things I have ever read. A short review or a comment cannot explain everything. (Also, I don’t fully remember everything; it was a few years ago.)
First, there is a difference between teaching a behavior, and unteaching a behavior. (Is “unteach” a proper English word?) Second, there is a difference between creating a new habit, and maintaining the existing habit.
On the topic of unteaching, the important thing is that instead of “don’t do X” it is often easier to teach an alternative “in situation Z (instead of X) do Y”. But if you want to use punishments, the important thing is that they come immediately and consistently. A small punishment that comes always and immediately after the act, works much better than a large punishment that comes only sometimes and several days after the act.
When you start teaching a new habit using rewards, again the important thing is to deliver the reward immediately and consistently… at the beginning. But after the habit is established, you gradually reduce the size and frequency of the rewards. If you stop rewarding suddenly, the animal will give it a few more attempts, and then give up… and maybe occassionally try again, just to see if the rewards have returned. But if you gradually make the rewards rare, the animal will keep doing it, and the occassional reward will be enough to keep the habit. If you deliver the rare rewards regularly (e.g. only once a day, or always once per 100 attempts), the animal will notice the regularity, and after receiving a reward will slow down (because it means that the next reward is far away). But if you deliver the rare rewards unpredictably, the animal will keep trying all the time.