How do you discover things about the world? Just look and see! It seems simple — but it's not. Each casual glance employs a billion brain cells to represent the present scene and to summarize its differences from records of other experiences. Your agencies formulate little bits of theories about what happens in the world and then make you do small experiments to confirm or reformulate those conjectures. It only seems simple because you're unaware of what is happening.
How do you discover things about your mind? You use a similar technique. You make up little bits of theories about how you think, then test them with tiny experiments. The trouble is that thought-experiments don't often lead to the sorts of clear, crisp findings that scientists seek. Ask yourself what happens when you try to imagine a round square — or when you try to be happy and sad at the same time. Why is it so hard to describe the results of such experiments or draw useful conclusions from them? It is because we get confused. Our thoughts about our mind-experiments are mind-experiments themselves — and therefore interfere with one another.
Thinking affects our thoughts.
People who program computers encounter similar problems when new programs malfunction because of unexpected interactions among their parts. To find out what's happening, programmers have developed special programs for debugging other programs. But just as in thought-experiments, there is a danger that the program being watched might change the one that's watching it. To prevent this, all modern computers are equipped with special interruption machinery that detects any other program's attempt to alter a debugging program; when this happens, the culprit is frozen in its tracks so that the debugging program can examine it. To do this, the interruption machinery must be supplied with a private memory bank that can store enough information to make it possible, later, to restart the frozen program as though nothing had happened.
Are brains equipped to do similar things? It was easy to build self-examination systems into computers that did only one thing at a time, but it would be much harder to do in a system that, like the brain, engages many processes at once. The problem is that if you were to freeze only one process without stopping the others, it would change the situation you're trying to examine. However, if you were to stop all those processes all at once, you couldn't experiment on how they interact.
Later, we'll see that consciousness is connected with our most immediate memories. This means that there are limits on what consciousness can tell us about itself — because it can't do perfect self-experiments. That would require keeping perfect records of what happens inside one's memory machinery. But any such machinery must get confused by self-experiments that try to find out how it works — since such experiments must change the very records they are trying to inspect! We cannot handle interruptions perfectly. This doesn't mean that consciousness cannot be understood, in principle. It only means that to study it, we'll have to use the less direct methods of science, because we cannot simply look and see.