If you could control your dreams, what kind of dreams would you dream about?

Some dreams feel so revelatory—if only returning to sleep would take us back there. It turns out, however, that our ability to shape our dreams is better than mere chance. In the blockbuster movie Inception, Dom Cobb (Leonardo DiCaprio) and his compatriots use drugs and psychological profiles to trigger specific dreams in people.

Although the heavy sedation and level of detail incited are far-fetched, dream control isn't entirely a Hollywood fantasy. Techniques to control, or at least influence, our dreams have been shown to work in sleep experiments. We can strategize to dream about a particular subject, solve a problem or end a recurring nightmare.

With practice we can also increase our chances of having a lucid dream, the sort of "dream within a dream" that Inception's characters regularly slip into. The ability to influence other people's sleep worlds is still crude. But emerging technologies raise the prospect that, at the very least, we'll get an idea of what others are dreaming about in real time.

We asked Deirdre Barrett, author of the book The Committee of Sleep: How Artists, Scientists and Athletes Use Dreams for Creative Problem-Solving—and How You Can, Too (Crown, 2001) and assistant clinical professor of psychology at Harvard Medical School, about what dream-control strategies do and don't work—and why. An edited transcript of the interview follows. We're all familiar with dreams, but what's the scientific definition?

The literal definition is a narrative experience that occurs during sleep. A few people will define it as a REM (rapid eye movement) sleep experience but, actually, the research doesn't support that. Some things that seem to look like dreams occasionally occur in other stages of sleep.

Why do most dreams seem to occur in REM, and what's happening during that sleep phase that seems to produce dreams? REM is generally the only time during sleep that most of the cortex is pretty much as active as it is when we're awake. During this phase, there are rhythmic bursts of activity in the brain stem.

There's one school of thought that this rhythmic firing is the sole cause of dreaming and all the upper cortical activity is a simple response to that. It just doesn't look that way. It looks like the lower brain stem activity wakes the cortex up and then the cortex does a lot of organized, meaningful thinking once it's activated.

The thing that is very frustratingly not neat and clean is that every once in awhile when you wake somebody out of a non-REM period, they report something that looks pretty much like the elaborate narrative of a dream. This is especially common in people who have big traumas and shift workers who have their sleep disrupted, so it may be that it happens mainly when something isn't operating completely properly with the regular sleep cycle. During dreams, are certain regions more active than others or does that depend on what you're dreaming about?

It's sort of halfway in between the extreme version of either of those. On average, there are several areas that are more active than they would be during the waking state.