rewriting Reality: New Clues on REM Sleep and Memory Consolidation
Source PublicationJournal of Cognitive Neuroscience
Primary AuthorsMcDevitt, Kim, Turk-Browne et al.

You turn the corner, expecting the familiar red door of your favourite bookshop. It is gone. A grey concrete wall stands in its place. Your brain reels for a microsecond. This is a prediction error—a glitch in your internal matrix. To stop you from making the same mistake tomorrow, your mind must sever the old link between ‘corner’ and ‘red door’. It needs to update the file. But when does this critical editing actually happen?
It appears the cutting room floor is located in our dreams. Researchers at the intersection of neuroscience and psychology designed an experiment to trap this fleeting process. They taught participants a simple rule: image A predicts image B. Once the volunteers trusted this sequence, the scientists broke it. Image A appeared, but B did not. The prediction failed. The brain was left holding an erroneous map.
The role of REM sleep and memory consolidation
The participants were then divided. Some remained awake. Others napped but stayed in the quiet shallows of non-REM sleep. A third group entered the chaotic, darting-eye phase known as REM. Later, inside the fMRI scanner, the stakes of that nap became clear.
The data, specifically from the right hippocampus, told a story of separation. In the group that achieved REM sleep, the neural patterns for the cue and the missing item had drifted apart. The brain had differentiated them. Conversely, those who stayed awake or missed out on REM sleep showed less of this separation. Their neural representations remained muddled.
While the researchers note that these findings in the dentate gyrus were uncorrected for multiple comparisons—meaning they require replication—the implication is striking. The study suggests that we do not merely sleep to rest. We sleep to correct our worldview. When the world surprises us, REM sleep may be the mechanism that ensures we are not fooled twice.