Sweet dreams are made of this

For decades, neuroscientists have dreamed about unlocking the mystery of dreams. What brain state(s) is involved in evoking vivid, intense and often bizarre dreams?

We now know that dreams occur during a state called rapid-eye movement (REM) sleep, in which our brain functions as a closed loop, shutting off out interaction with the environment. All sensory perception eventually goes to the thalamus, whose activity is inhibited during REM, effectively cutting off (gating) external input; meanwhile, motor output is also inhibited by the brain stem, sending our bodies into a state called atonia (“no-(muscular)-tone”) , which usually prohibiting any acting out of dreams.

On the other hand, some cortical/limbic regions, such as the visual association areas, go into overdrive during REM, while regions such as the dorsal-lateral prefrontal cortex lower their metabolism. These states are somehow integrated in the cortex to produce internally-generated perceptions, which manifest as partially coherent dreams (a very crude analogy would be the brain coming up with its own stories to tell itself).

But what neural processes dictate dream content? Research in the past has hinted that dreamed motor actions in a specific task involve activations of similar motor regions during the same awake actions. Furthermore, patients with REM sleep disorders - in which they don’t have atonia- tend to “act out” dream fragments. However, it has been extremely difficult to directly study specific dream content, as mental motor movement in spontaneous dreams cannot be experimentally controlled.

In this paper, the authors try to assess dream content by utilizing a phenomenon called lucid dreaming. It is a rare but trainable state of sleep in which the sleeper becomes aware of the experience of dreaming, has full access to memory, and can control dream actions. Lucid dreamers can communicate their state by predefined eye-movements, which can be monitored by the electrooculogram (EOG).

Six highly experienced lucid dreamers were recruited, and instructed to make series of mental left and right hand clenching movements separated by sets of left-right-left-right eye movements to signal that a) they were in a state of lucid dreaming, and b) they were about the switch hands. During the task they were monitored by EOG, combined electroencephalography (EEG)-fMRI and EEG- NIRS (near-infrared spectroscopy). EOG measures eye-movement, EEG monitors brain waves to verify their REM sleep state, and both fMRI and NIRS measures brain activity by looking at blood-oxygen levels (energy consumption) and blood flow (hemodynamic responses), respectively. An actually executed hand-clenching task and an imagined one where also performed as control conditions.

Being hooked up to half a dozen machines can’t be too comfortable – unfortunately, only two subjects managed to achieve lucid dreaming and perform the predetermined mental motor task, one subject for each monitoring (fMRI or NIRS) method. This severely limits the generalizability of the data acquired. Nonetheless, with fMRI, researchers found increased activity in the sensorimotor cortex contralateral to the indicated movement side, which is what also happens during wakefulness. However, compared to acting out hand-clenching, the same mental movement in lucid dreaming activated very specific clusters and showed less fluctuations. This result was confirmed by NIRS data, where a dream state elicited smaller hemodynamic response compared to awake. Surprisingly (as this was not seen in fMRI), the supplementary motor cortex (SMA) – which is involved in planning motor sequences- showed the same amount of activation in mental movement in lucid dreaming and actual performance in awakefullness.

So what do these results tell us?  In general, the data supports the notion that the pattern of activity in motor imagery during a dreaming state largely overlaps with activity corresponding to motor execution. This may help explain why in some studies, subjects show enhanced motor performance of a specific task after a nap. For importantly, the current study also acts as a proof-of-concept that neuroimagining can be used to measure the neural correlates of specific dream content rather than a general dream state. Furthermore, the technique has the potential of being used to inversely infer specific dream content, or to put it in a scifi term, “dream reading”. Although preliminary, it is an improvement over previously used methods such as self-reports of dreams, which can be subject to distortions and inaccuracies.

However, it is important to note the limitations of this study. The subject pool was extremely small; hence results could be strongly influenced by individual differences. Actual hand twitches were not measured directly, making the results more difficult to interpret. Furthermore, as the authors noted, while lucid dreaming has all the defining markers of sleep and basal dream features, “it is still different from nonlucid dreaming in its metacognitive insight into the hallucinatory nature of the dream state and full access to cognitive capabilities”. Thus it is difficult to assess whether neural activity during lucid dreaming can be translated into that of nonlucid dreams.

While we are stills miles off from actual “dream reading”, this study provides one potential direction that might lead to dream recording machines.

Quick note: It is possible that the media might pick this study up and blow it out of proportions. As a cautionary tale, check out neuroscientist Morgan Cerf’s story on how his imagining research led to a media fiasco, and the so-called “dream machine”. On the Story Collider.

Dresler et al. (2011). Dreamed movement elicits activation in the sensorimotor cortex Current Biology DOI: 10.1016/j.cub.2011.09.029