Sleeping on a decision helps, though not for the reason usually given

A piece of advice with a theory attached

Almost everybody accepts that important decisions should be slept on, and most people can point to an occasion when the morning produced a clearer view. Attached to this ordinary advice, for about two decades, has been a much stronger scientific claim: that for complex decisions, deliberately not thinking about the problem produces better choices than thinking about it, because unconscious processing has greater capacity.

Those are different claims. The first is a piece of practical experience. The second is a specific theory with a specific prediction, and it is the second that has run into difficulty.

What happened to the strong version

The original experiments compared people who deliberated, people who were distracted by an unrelated task, and people who answered immediately, on choices with many attributes. The distracted group was reported to choose better on the complex problems, which was surprising enough to attract enormous attention.

Subsequent replication attempts, including coordinated efforts across multiple laboratories with substantially larger samples and pre-specified analyses, have largely failed to reproduce the advantage. Reviews of the accumulated literature have also raised concerns about how sensitive the original results were to particular materials and analysis choices. The theory has defenders, and the fair description is that its central prediction has not survived well.

Incubation, which is a narrower and better-supported idea

A related but separate research area concerns incubation in problem solving: setting a problem aside and returning to it. Pooled analyses of this work generally find a real but modest benefit, concentrated in certain problem types, particularly ones where an initial wrong approach has to be abandoned.

The proposed mechanism here is unglamorous. A break allows an unhelpful line of attack to fade, so the problem is re-encountered without the stuck framing. That requires no claim about superior unconscious computation, and it fits the common experience of seeing the obvious answer immediately on returning.

The mechanisms that clearly do operate

Several are uncontroversial and they explain most of what people notice. Strong emotion subsides, and a decision made while angry or elated weights the immediate feeling heavily. New information arrives, sometimes because the delay prompts somebody to mention something. The set of options changes, since alternatives that were not visible on the first pass often surface on the second.

And you get to read your own reasoning cold. Notes written yesterday are encountered as a stranger’s argument, which is the closest most people come to reviewing their own thinking without knowing what they meant. That alone justifies the practice, and it requires no theory of unconscious deliberation whatsoever.

Sleep itself may add something on top of the interval, since there is reasonable evidence that memory is reorganised during it and that some kinds of problem benefit from a night rather than an equivalent period awake. The size and generality of that benefit are still argued about, and it is not required for any of the above. A day would do most of the work.

What a delay costs

The obvious cost is that options expire, and a decision deferred past its deadline has been made by default. The subtler cost is that delay is the most respectable form of avoidance available, since it can be described as prudence indefinitely.

The distinguishing test is whether the delay is doing anything. A delay that has a purpose — waiting for a specific piece of information, waiting for a feeling to pass, waiting to reread the case — has an end condition attached. A delay with no end condition is not deliberation, and calling it deliberation is how a decision quietly becomes an omission.

A practice that survives the evidence

Separate the stages and put a boundary between them. Gather and think in one session, ending with the reasoning written down. Return to it after an interval, read what you wrote before reconsidering anything, and note specifically what looks different. Then decide, and record what you expect to happen.

Nothing here depends on the contested claim. The interval works through emotion, information, fresh option generation and cold reading, all of which are ordinary and all of which are enough. If somebody tells you the benefit comes from your unconscious mind performing superior computation while you were busy, that specific claim has not held up under replication, and the advice is fine without it.