In the previous post, I reviewed one recent study that examined how the contents of thought change from waking, to falling sleep, to dreaming. The main finding was that, whereas in waking life we tend to be in control of our thoughts to some extent, as we fall asleep these thoughts seem to suddenly appear without our having intended. In psychology, we refer to thoughts that we are conscious of creating and directing as ‘meta-cognitive’, and it is precisely a lack of metacognition which so frequently punctuates the dreaming world.

Another recent paper explored this concept on a new level; the authors aimed to find out what the brain is doing when we have spontaneous thoughts during waking and sleeping, and whether this brain activity can be linked to ‘metacognition’. Specifically, they expected there may be some common brain activity occurring during waking and sleeping thoughts, but that perhaps the area of activity relevant to metacognition may be decreased during sleep.

For the study, the authors sampled participants’ thoughts during quiet wakefulness, NREM stage 2 sleep (N2), and REM sleep. Sixty-nine healthy individuals between 25-64 were recruited to participate in the experiment.

Collecting thoughts during wake and sleep

In order to sample the thoughts of participants during waking, the participants sat in front of a computer and focused on a cross in the center of the screen for 30 minutes. Every once in awhile (about once a minute, at random intervals), a sound would signal to the participant to immediately report what was going through their mind. They would then rate how ‘thought-like’ the report was and how much effort they were putting into the thought. For example, if someone was merely observing the fact that they were hungry, this would not be very thought-like and also would not require much effort, whereas if someone was trying to list everything they had to do after the experiment, this would be both thought-like and effortful.

To collect thought patterns during sleep, participants were awakened multiple times throughout the night during specific stages of sleep (around 20 minute intervals during only Stage 2 and REM sleep). There were again asked to report what was going through their mind and then responded to the same thought-like and effort scales.

The participant reports were later scored by judges, who first classified whether the reports could be considered ‘low-thought’ or ‘high-thought’ experiences, and then scored the extent of ‘metacognition’ in the reports.

In total, 869 waking reports, 157 N2 reports, and 107 REM sleep reports were scored. High-thought trials were more frequent in waking reports (75%), than either N2 sleep reports (45%) or REM sleep reports (62%). Further, waking thoughts were rated significantly higher on ‘metacognition’ than thoughts during either N2 or REM sleep.

Neural Correlates of Thought

For the brain imaging portion of the study, 13 participants were selected to analyze high-density EEG activity during wake and NREM sleep; and 10 participants were selected to analyze high-density EEG during REM sleep. This was done in order to compare brain activity between low-thought and high-thought reports, and between thoughts low vs high in metacognition.

During wake, high-thought reports were associated with decreased delta power (1-4 Hz) compared to low-thought reports over several brain areas, including the midcingulate and posterior cingulate cortex, premotor cortex, and the medial prefrontal cortex. A similar pattern was found for NREM stage 2 sleep, with decreased delta power for high-thought reports in several of the same areas, especially midcingulate cortex, but not in the prefrontal cortex. In REM sleep, decreased delta power for high-thought reports was again greatest over the midcingulate cortex, but did not show other similarities to wake and N2.

Of note, across all three states the midcingulate cortex was found to differentiate high-thought from low-thought reports, indicating this region as a common neural substrate of thought across waking and sleeping.

When examining reports high in metacognition, the authors found decreased delta power over the medial prefrontal cortex compared to those low in metacognition. This pattern is similar to that seen for high-thought vs. low-thought reports, but seems specific to the prefrontal cortex, as opposed to the midcingulate cortex.


Overall, the results indicate that high levels of thought were associated with activation of the midcingulate region across all three states. This finding suggests that thoughts do share a certain neural substrate across different states of wake and sleep, despite these states having very distinct neurophysiological profiles.

Nevertheless, waking thoughts were found to have higher metacognitive content than dreaming thoughts, which corresponds with our tendency during wake to consciously think about our thoughts, feelings, and behavior, whereas during dreaming thoughts seem to be more spontaneous. This form of metacognition in wake seemed to be associated with activity in the prefrontal cortex, which is reduced during sleep.


Perogamvros, L., Baird, B., Seibold, M., Riedner, B., Boly, M., & Tononi, G. (2017). The Phenomenal Contents and Neural Correlates of Spontaneous Thoughts across Wakefulness, NREM Sleep, and REM Sleep. Journal of Cognitive Neuroscience.

You are reading

Dream Factory

How to Have Lucid Dreams

A recently published study compares techniques for inducing lucid dreams.

Animating Dreams and the Future of Dream Recording

Recent research uses body signals during sleep to power an animated avatar.

Dream Deprived: A Modern Epidemic?

A recent paper reviews how poor sleep habits are robbing us of our dreams.