Why we forget dreams so quickly

The speed is what makes it strange. You wake holding an entire dream — people, a place, a plot, a feeling — and within seconds it is thinning out, and within a minute or two there is nothing left but the knowledge that there was something. No other experience disappears like that. Whatever else dreaming is, it is extraordinarily badly retained.

The main explanation is chemical. Laying down new long-term memories depends on a set of neuromodulators, and REM sleep is a state in which some of them are at unusual levels. Noradrenaline in particular drops to its lowest point of the entire 24-hour cycle during REM. That specific chemistry is associated with poor encoding of new long-term memories. The brain during a dream is not in a good state for recording the dream — which is a slightly absurd design, but it appears to be the design.

The consequence is that recall depends almost entirely on waking. Dream content is generally available only if you wake during the dream or immediately after it, while it is still held in short-term memory and can be transferred across into something more durable. Sleep straight through a dream and it is usually gone, unrecoverable, not stored anywhere to be retrieved later. This explains a great deal that otherwise looks mysterious: why the dream you remember is almost always the last one before the alarm, why an alarm going off mid-dream produces unusually vivid recall, why waking naturally in the night often surfaces a dream you would not otherwise have known you had.

It also explains the fragility of the minutes after waking. Reaching for the phone, getting up, starting to plan the day — any of these introduce new material into exactly the short-term store the dream is sitting in, and the dream loses the competition. The standard advice to stay still and not open your eyes when you wake is not folklore; it follows directly from how the memory is held.

People differ substantially in dream recall, and the differences are measurable rather than imagined. Some people report several dreams a week, others almost none for years. The most likely explanation is not that low recallers dream less — everyone appears to dream — but that they have fewer opportunities to catch one. Research associated with Perrine Ruby's group has found that high recallers have more brief intra-sleep awakenings, short periods of wakefulness during the night that mostly go unremembered but provide exactly the window a dream needs to be transferred. High and low recallers have also been reported to differ in brain activity measures, including responsiveness to stimuli during sleep.

Other factors are more prosaic. Attitude and interest correlate with recall — people who find dreams interesting recall more, which sounds circular but appears to work partly through intention before sleep and attention on waking. Alcohol reduces recall, both by suppressing REM early in the night and by fragmenting the second half. Severe sleep restriction reduces it too, by cutting into the REM-rich hours near morning. And keeping a dream journal reliably increases recall, one of the better replicated findings in this area — the practical article on remembering your dreams in this section covers what follows from that.

A last point worth making, because it comes up constantly. Forgetting a dream is not a sign of anything. There is no evidence that low recall indicates suppression, avoidance or a psychological problem, and no evidence that a forgotten dream was more significant than a remembered one. The selection is mechanical: dreams that coincided with a moment of waking survive, dreams that did not are gone. What you remember of your own dream life is a sample determined by the shape of your night, not a curated message.

Frequently asked

Dream symbols on this topic

• ☽ Dreaming about forgetting — meaning and interpretation

What this is based on

• Research on REM neurochemistry and memory encoding
• Perrine Ruby's group's work on high and low dream recallers
• Dream-diary research on factors affecting recall frequency