Matt Wall discusses ‘Inception’ and the science of dream-hacking.
“If you need me, me and Neil will be hanging out with the Dream King. Neil says ‘hi’, by the way…”
- Tori Amos ‘Tear in your hand’ from ‘Little Earthquakes’ (1991).
Dreams and the unconscious mind have long held a fascination for film-makers and other artists. This interest might derive partly from the stubborn refusal of the media world to progress away from a psychoanalytic perspective in their analysis (an approach which was already out of date in the 1960s in psychology, but, bafflingly, still seems to have some currency in media-studies departments), but also partly perhaps because of the limitless creative opportunities that trying to put dreams on the screen affords. A highly influential early example of dreams on film is the Salvador Dali-designed dream sequences in Hitchcock’s ‘Spellbound’ (1945), which (as you might expect) is disturbingly surrealist and seems to be what David Lynch based a substantial part of his early career on. Another well-known example is Kubrick’s ‘Eyes Wide Shut’ (1999), based on Arthur Schnitzler’s Traumnovelle (‘Dream Story’), which some have interpreted as an extended meditation on the nature of fantasy and dreams.
One of the big hits of last summer was ‘Inception’ by Christopher Nolan; a rare blockbuster in that it featured a highly convoluted plot, an ambiguous ending and the action taking place almost entirely in a dream-world. Inception is essentially a heist-movie, with the standard elements of ‘getting-the-team-together’ at the beginning and ‘oh-no-the-plan’s-all-going-a-bit-wrong’ towards the end, however the clever twist is that the operation takes place inside the victim’s head, as part of a shared dream-state. The technology which allows this multiple sharing of consciousness is never explained, but it doesn’t really matter; a great deal of thought clearly went into building up a consistent logic for the dream-world(s) and the means of operating within them, which makes for an instantly and deeply interesting and absorbing film. Use of impossible architecture like the Penrose staircase and other tropes such as zero-gravity which are not found in the real world make some really engaging sequences, and the creativity displayed in the use of visual effects (the exploding café, and Parisian streets folding in on themselves in particular) is stunning.
While the idea of being present in, and actively manipulating, someone else’s dreams is currently science fiction, a number of recent pieces of research have actually begun to take tentative steps in that direction. One of the recent advances in brain-imaging methods that’s generated a lot of excitement in the last few years is a technique called Multi Voxel Pattern Analysis (MVPA). This is a method of analysing fMRI data which relies on looking at broad patterns of activity across a brain region; a classification algorithm is ‘trained’ with some input in the form of brain images from two different categories, and can then (when given some new brain images) successfully predict which category the images belong to. So, a participant could be looking at either faces, or houses, and the algorithm (given the right examples in the training phase) could tell whether a face or a house was being presented, based on their brain activity. Some parts of the media have got quite excited about this technique and have called it ‘mind-reading’ however at the moment it’s somewhat limited by the algorithm having to go through an appropriate training phase with each participant. The technique has developed well beyond the simple two-category stage though, this press release from UC Berkeley details one of the most recent and exciting experiments in this area, where in fact movies are reconstructed from brain data. This technology is actually strongly reminiscent of another film, Wim Wenders’ sprawling, ridiculously-long, occasionally-magnificent but more often teeth-grindingly ponderous Until the End of the World (1991). In this film several of the characters become obsessed with an invention that allows them to record and play-back their dreams. Since we know the visual cortex of the brain is highly active during REM (dreaming) sleep, it’s not a huge leap to imagine that it will perhaps be possible to record at least a rudimentary version of a dreamer’s visual experience using a fMRI scan and further developments of the MVPA technique to reconstruct the content.
Another related piece of recent research concerns so-called ‘lucid’ dreamers, the small percentage of the population who manage to become aware that they are in a dream while still dreaming, and can then influence the dream environment. A team from the Max Planck institute in Germany gave specific instructions to a lucid dreamer, and then scanned his brain while he carried them out, in his dreams. The instructions were to clench first his right first, and then his left alternately for ten seconds, and this dreamed ‘movement’ was evident in characteristic alternating activation patterns in the left and right motor cortex; the participant’s actual hands never moved. This research represents an impressive leap forward as it’s the first time that the specific content of a dream (the movement) has been visualised by an external system. Another good report on this experiment is here.
Another set of researchers are coming at the idea of dream-hacking from a different angle. A project called ‘Power Dreaming’, due to launch next year, and funded by the US army, is seeking to provide therapy for the large number of veterans who suffer from nightmares as a result of Post-Traumatic Stress Disorder (PTSD). While details are somewhat sketchy at the moment, it appears the aim of the project is to reduce the power of negative, traumatic dreams through alternative soothing stimuli, generated through virtual reality and delivered through 3D goggles. A more detailed write-up on this project was recently published by Wired.
So, while ‘Inception’ remains science fiction for the moment, the dream-recording device which featured in ‘Until the End of the World’ in 1991 is now taking its first steps towards being a viable technology. In fact, I’d be very surprised indeed if more than one group of researchers was not actively working on extending the techniques mentioned above in this direction already. Within a decade or two we could all be routinely recording our dreams and playing them back over breakfast. It sounds far-fetched I know, but, you gotta dream, right?
Matt Wall is a cognitive neuroscientist currently working at UCL, specialising in fMRI. He does occasionally dream in MRI scanners, but only by accident. His main web-page is here and he also writes the blog ‘Computing for Psychologists’. Twitter: @m_wall.