Christopher Nolan’s Inception is a film about a time when we have the power to enter into each other’s dreams, and actively steer the dream’s course to implant an idea in the dreamer.

The film raises the issue of how much we understand about the neuroscience of dreams. Due to its need for invasive experiments, neuroscience typically works with non-human animals, which raises a significant difficulty: how do you know that a rat is dreaming? You can’t wake it up from REM sleep and ask. (Well, you can, but don’t expect a cogent response.) There’s no accepted objective indicator that a person or animal is having a dream, as opposed to sleeping. But, we can still learn something useful by looking at the neuroscience of sleep.

The neuroscience of sleep has told us a few important things over the years. For example, we know that our pattern of sleep and wakefulness (the “circadian rhythm”) has much of its basis in the activity of the suprachiasmatic nucleus, a rice-grained-sized group of cells just above where the optic nerves from our eyes crossover. We know that our free running rhythm—what we go to if we are completely in the dark, with no indicator of solar activity—is slightly over 24 hours, and that the length of the rhythm can be affected by things like cannabinoids found in pot. We know that the brain activity of a person dreaming is very similar to that of an awake person—were it not for the fact that our body is paralyzed during dreaming, we’d probably do a lot of things we’d regret.

While we’ve made a lot of progress in understanding sleep, we’ve a long way to go to understand dreaming. What makes it a challenge, perhaps as big a challenge as understanding consciousness itself, is the subjective aspect of dreaming. For example, we know that vivid dreaming occurs during REM sleep in humans. We also know that other animals have REM sleep. Do they also dream? How can we know, since, as I mentioned above, we can’t wake them during REM sleep and ask (the way we determined this fact with humans)? How we can go from objective facts like the presence of REM sleep to subjective ones, like a dream of a pink elephant bouncing down along a high tension power line (from one of my own dreams) is as unclear as how we get from neurons firing to awareness. Nonetheless, significant work has occurred on some of the neuronal correlates of REM sleeping in rodents and songbirds.

The most intriguing result from recent work is that during sleeping, the brain appears to “play back” patterns of activity that occurred during the day. For example, Matt Wilson and colleagues have found that patterns of “place cell” activity — brain cells that light up, like crumbs left on Hansel and Gretel’s path in the woods, corresponding to a specific path that the animal (in these experiments, a rodent) took during the day — and this playback seems to be integral to the animal learning the path it took. In birdsong, from work by Dan Margoliash and others, we’ve learned that birds playback patterns of activity almost identical to singing while they sleep, and again, it seems to be integral to the bird learning its large repertoire of over a million syllables. Why does the brain play back patterns of daytime activity at night? It isn’t completely understood, but some backstory on memory research helps motivate one hypothesis.

It’s been known for some time that a structure called the hippocampus is responsible for acquisition of new memories. Without it, we still have our memories, but anything new that happens is completely lost (think of the movie Memento, one of Christpher Nolan’s previous films) — we are stuck in the continual present. Real-life patient HM taught us this many years ago, after he had this structure removed as part of an experimental operation to cure his epilepsy. He, and many similar cases, lose all memory but for those events that happened some time before the loss of their hippocampus, typically a few months. Over time, the idea has emerged that perhaps the hippocampus “trains” the neural networks in other regions of the brain to store memories through repeated playback during sleep. Like crickets trying to attract females in the night, in the world of memory nothing succeeds like persistent repetition.

So if in REM sleep the brain is repeating patterns of activity from periods of wakefulness, perhaps that process helps the brain to remember, over the long term, the items that are temporarily stored in the hippocampus.

What is not understood from these studies, which were done in rodents and song birds after all, is the basis of all the strange subjective aspects of dreaming — such as how or why in our dreaming we seem to borrow from real experience while adding a good dollop of stuff from elsewhere. This aspect of dreaming seems like it would be crucial in order to have any hope of building a dream experience a la Inception. There is not a whole lot of creative potential in simply regurgitating the day’s brain patterns.

Until these and many other mysteries of dreaming are solved, what the research is showing is that the best way to architect a dream is to architect the experience you have during wakefulness, since dreaming appears to be a lot about learning patterns you were exposed to while awake. Our understanding of the coupling is not clear enough to think about designing dreams by structuring our awake behavior, but perhaps with further research we will come to that point where we can do inception of ideas into our own heads.

Image: Midlands Rat Club
Corrections: Aug 11, 2010: “…neuroscience typically works with animals, rather than humans…” adjusted to “…neuroscience typically works with non-human animals.” Reference to length of circadian rhythm also adjusted.