So far, we've talked mostly of learning from success. But consider that when you succeed, you must already have had the necessary means within your grasp. If so, then making changes in your mind might only make things worse! As people often say, You shouldn't argue with success. For whenever you try to improve an already working procedure, you risk damaging whichever other skills depend on that same machinery.
Accordingly, it may be more important that we learn from how we fail. What should you do if some well-established method — call it M — has failed to reach a certain goal? One policy would be to alter M, so it won't make the same mistake again. But even that might be dangerous because it might cause M to fail in other ways. Besides, we might not know how to change M to remove the error. A safer way to deal with this would be to modify M by adding special memory devices called censors and suppressors (we'll discuss this in detail later), which remember particular circumstances in which M fails and later proceed to suppress M when similar conditions recur. Such censors would not tell you what to do, only what you shouldn't do; still, they prevent your wasting time by repeating old mistakes.
Learning has at least two sides. Some parts of our minds learn from success — by remembering when methods work. But other portions of our minds learn mainly when we make mistakes, by remembering the circumstances in which various methods failed to work. Later we'll see how this can teach not only what we shouldn't do, but also what we shouldn't think! When that happens, it can permeate our minds with prohibitions and taboos of which we're entirely unaware. Thus, learning from success tends to aim and focus how we think, while learning from failure also leads to more productive thoughts, but in a less directed way.
We would not need to deal with exceptions and censors if we lived in a universe of simple, general rules with no exceptions, as in the lovely mathematical worlds of arithmetic, geometry, and logic. But perfect logic rarely works in the real worlds of people, thoughts, and things.
This is because it is no accident that there are no exceptions to the rules in those mathematical worlds: there, we start with the rules and imagine only objects that obey them. But we can't so willfully make up the rules for objects that already exist, so our only course is to begin with imperfect guesses — collections of rough and ready rules — and then proceed to find out where they're wrong.
Naturally, we tend to prefer learning from success rather than from failure. However, I suspect that confining ourselves to positive learning experiences alone leads to relatively small improvements in what we can already do. Probably, there is no way to avoid at least a certain degree of discomfort when we make substantial changes in how we think.