THE REAL REASON YOU REREAD THE SAME EMAIL FIVE TIMES. BEFORE SENDING IT — AND WHY THE FIFTH READ MAKES IT WORSE.
You Are Not Proofreading. You Caught The Typos On Read One.
What you are doing on reads three, four, and five is trying to convert an unknown into a certainty — and the format cannot deliver it. Worse: the checking itself is what makes another check feel necessary. For anyone who has spent eleven minutes on a four-line message and could not have told you, at any point, what they were still looking for.
THE THING YOU ARE TRYING TO RESOLVE IS NOT ON THE SCREEN.
Reread one is legitimate. Reread two is defensible. What happens after that is a different activity entirely.
Reread one is legitimate. You are checking accuracy — names, numbers, the attachment. Reread two is defensible. You are checking tone, which is a real variable and one people misjudge in writing.
Reads three through five are a different activity entirely, and misunderstanding which activity it is keeps people trapped in it. By then you are no longer scanning the text for defects. You are attempting to determine how this message will land with a specific person whose reaction you cannot observe, in a context you do not control, at a moment you cannot predict.
That is not an error you can find. It is an unknown, and the drive to eliminate it is one of the most studied phenomena in the field. Reviews of the contemporary literature identify intolerance of uncertainty — and, underneath it, a more fundamental fear of the unknown — as a common core running through anxiety and mood presentations, with the discomfort arising from the unknown itself rather than from any particular predicted outcome.1 Elevated intolerance of uncertainty is not a disorder. It is normally distributed across the general population.2 Most high performers sit somewhere on it, and it activates hardest around consequential communication.
Which is why the loop will not close. You are running a search for information that is not in the document, and no number of passes will produce it.
"You are not looking for a mistake. You are looking for a guarantee — and you are looking for it in a file that does not contain one."
REPEATED CHECKING RELIABLY LOWERS CONFIDENCE IN WHAT YOU CHECKED. IT BARELY TOUCHES ACCURACY.
This is the part that makes the loop close on itself, and the evidence is unusually clean.
A meta-analysis of the repeated-checking paradigm — 28 experiments, 1,662 participants — found that repeated checking produced large decreases in memory confidence, vividness, and detail, with only small reductions in memory accuracy.3 A later systematic review and meta-analysis reached the same structural conclusion: confidence deteriorates substantially with repeated checking while accuracy holds up far better.4
Read the direction of that carefully. Checking does not make you more accurate in any meaningful way. It makes you feel less sure. The proposed mechanism is familiarity: as a stimulus is repeatedly examined, perceptual processing declines, the representation loses vividness and detail, and confidence — which is partly built on vividness — falls with it.
So the fifth read produces a dimmer, less detailed impression of the email than the second read did. The reduced confidence then registers as a reason to read it again. The behaviour manufactures its own justification.
Catches the real errors. Names, figures, the missing attachment. High value.
Catches tone. A genuine variable that is easy to misjudge in writing. Reasonable value.
Searches for information the document does not contain — how it will be received. Zero yield, by construction.
STOPPING IS AN EXECUTIVE ACT, AND EXECUTIVE CAPACITY IS NOT CONSTANT.
Notice when the loop is worst. It is rarely at 9am on a clear morning.
It is at 6pm, after a day of demanding decisions, on a message that is not even the most important thing you handled.
There is a plausible mechanism for that timing. Research using magnetic resonance spectroscopy across an approximate workday found that participants performing high-demand cognitive control tasks — but not those doing low-demand versions for the same duration — showed elevated glutamate concentration in the lateral prefrontal cortex by the end of the day. The behavioural signature accompanying it was a measurable shift in preference toward low-effort, immediately available options.6 The proposed account is that sustained control exertion produces a metabolic cost that makes further mobilisation of that same control system harder.
Hold that loosely — it is one strand of a developing literature, not settled fact. But it aligns with something observable: the capacity required to override an urge and declare a task finished is the same capacity that has been running all day. Rereading is cheap, immediate, and momentarily relieving. Deciding to stop is effortful. Late in the day, the cheap option wins more often.
The instinctive fix is to gather more certainty: run the message past a colleague, redraft it a third time, search your history for how you phrased something similar before. Each of these produces a brief drop in discomfort, which is precisely the problem — the relief reinforces the checking, and the underlying unknown is untouched, because it lives in another person's head. Intolerance of uncertainty is well established as a transdiagnostic treatment target, and the interventions that reduce it do not work by supplying certainty. They work by building tolerance for its absence.2 The move is not to resolve the unknown. It is to send while it is still unresolved.
SET THE NUMBER BEFORE YOU WRITE. NOT WHILE YOU ARE REREADING.
Every effective intervention here shares one property: it removes the stopping decision from the moment when your capacity to make it is lowest.
Pass one for accuracy. Pass two for tone. Then send. The limit has to be set before you open the draft, because deciding "that's enough" mid-loop is exactly the judgement the loop has degraded.
Before pass three, ask what specifically you are still looking for. If you cannot name it in a sentence, it is not in the document. If you can name it, fix that one thing and send.
The unease at send is not a signal that the message is wrong. It is the residue of an unknown you were never going to resolve. Treating it as data is what starts pass six.
If stopping requires control capacity, spend it when you have it. A consequential message drafted at 6pm will cost more passes than the same message drafted at 10am.6
The fifth read is not diligence. It is a search for a guarantee the format cannot issue, running on a system that gets less confident every time you run it. Two passes. Then send it.
"Why Your Brain Treats Ambiguity Like Danger (And What That Costs You Daily)."
The long-form companion — on uncertainty as a threat signal, and why the brain would rather have bad news than no news.
Back To Articles & Resources- Carleton RN. Into the unknown: A review and synthesis of contemporary models involving uncertainty. Journal of Anxiety Disorders. 2016;39:30–43; and Carleton RN. Fear of the unknown: One fear to rule them all? Journal of Anxiety Disorders. 2016;41:5–21. ↩
- On the normal distribution of intolerance of uncertainty across the general population, and its status as a well-established transdiagnostic risk factor and treatment target reliably reduced through cognitive behavioural therapy, see the review literature building on Carleton (2016), including systematic reviews summarising intervention effects on intolerance of uncertainty and associated symptom severity. ↩
- van den Hout MA, van Dis EAM, van Woudenberg C, van de Groep IH. OCD-like checking in the lab: A meta-analysis and improvement of an experimental paradigm. Journal of Obsessive-Compulsive and Related Disorders. 2019;20:39–49. ↩
- Abbasi Jondani J, Yazdkhasti F, Abedi A. Memory confidence and memory accuracy deterioration following repeated checking: A systematic review and meta-analysis. Journal of Behavior Therapy and Experimental Psychiatry. 2023;81:101855. ↩
- Burns DJ, Dalterio CH, Burns SA, Coelho GV. Not all checking decreases memory confidence: Implications for obsessive-compulsive disorder. Behaviour Research and Therapy. 2020. ↩
- Wiehler A, Branzoli F, Adanyeguh I, Mochel F, Pessiglione M. A neuro-metabolic account of why daylong cognitive work alters the control of economic decisions. Current Biology. 2022;32(16):3564–3575.e5. See also Blain B, Hollard G, Pessiglione M. Neural mechanisms underlying the impact of daylong cognitive work on economic decisions. PNAS. 2016;113(25):6967–6972. ↩