The Journal

Why reminders stop working, and the ones that last

The first week a new reminder works. By the fourth it can be swiped away without being read. That is not a failing of character, with ADHD or without it: it is habituation, a close relative of what hospitals call alarm fatigue. What follows is what the research says a prompt has to do to survive a month, and which parts of that are evidence and which are inference.

Published 2 November 2026

---

The first week works, and that is the problem

You set a reminder for 7pm: take the bins out. On Monday it works. By the following Thursday your thumb has dismissed it before your eyes have finished reading it, and the bins are still in the kitchen. A month later you no longer notice it arrive.

The question is often asked about ADHD. But the mechanism is not specific to ADHD, and none of the studies below recruited people with ADHD. Nothing here is medical advice or a substitute for assessment; if attention problems are affecting your work, studies or relationships, the NHS advises seeing a GP.

What happened to the 7pm reminder has a name and a well-described shape. It is worth knowing, because the fixes that follow from it are different from the usual advice to try harder or set a louder alarm.

Alarm fatigue: what it is, and who it was measured in

The clearest real-world record of people learning to ignore alerts comes from hospitals, where it is called alarm fatigue. Sendelbach and Funk, reviewing the problem in AACN Advanced Critical Care in 2013, report that research has found "72% to 99% of clinical alarms are false", and describe alarm fatigue as sensory overload that "can result in desensitization to alarms and missed alarms".

The Joint Commission, which accredits American hospitals, put a figure on the scale in its 2013 Sentinel Event Alert on alarm safety: "between 85 and 99 percent of alarm signals do not require clinical intervention", and "as a result, clinicians become desensitized or immune to the sounds".

Two things in that record matter at home. First, the people who tuned out were trained professionals whose job was to respond. Tuning out is not a sign of not caring. Second, the trigger was volume of alerts that turned out to need nothing. A phone that buzzes all day with messages, offers and news is the domestic version, and in one week-long field study of 15 Android users, Martin Pielot and colleagues logged an average of more than sixty notifications a day per person. Your bin reminder is competing with all of them.

The limit: ICU monitors are not phones, and a nurse is not a household. The hospital record shows that people stop responding to alerts that rarely matter; it does not measure your phone.

Habituation, defined properly

Habituation is the decline in a response to a stimulus that is repeated. In 2009 fourteen researchers who study it across a wide range of species revised the standard list of its characteristics, published by Rankin and colleagues in Neurobiology of Learning and Memory. Four of the ten matter here, in their words as reproduced in a later open-access review:

"Repeated application of a stimulus results in a progressive decrease in some parameter of a response to an asymptotic level." In plain terms, the response fades and settles low.

"More frequent stimulation results in more rapid and/or more pronounced response decrement." The more often it comes, the faster it fades.

"Within a stimulus modality, the less intense the stimulus, the more rapid and/or more pronounced the behavioral response decrement." Gentle signals fade faster.

"Within the same stimulus modality, the response decrement shows some stimulus specificity." What has faded is the response to that particular stimulus; a different one gets through. A related characteristic, dishabituation, is that a new stimulus can restore the faded response.

These characteristics come mostly from laboratory work, much of it in animals. They describe a general property of nervous systems, not anything about reminders or apps.

Why a fixed daily reminder is the textbook case

Put the 7pm reminder against that list. It arrives daily, which is frequent. It is quiet: one buzz, one line of text. And it is identical every time: same sound, same words, same moment. On each of those characteristics it is close to the textbook case of a stimulus that will fade.

There is human evidence that changing a repeated alert slows the fading. Anthony Vance and colleagues, in MIS Quarterly in 2018, measured attention to computer security warnings with brain imaging and eye tracking over a five-day working week. Attention declined across the week and recovered only partly between days, and a "polymorphic" design that changed the warning's appearance "substantially reduced habituation of attention". In a three-week field experiment with app permission warnings on phones, adherence fell for everyone but fell "at a substantially lower rate" for people shown the changing version.

What this piece could not find is direct evidence that varying the time of a reminder, rather than its appearance, resists habituation in people. The habituation characteristics concern how often and how strongly a stimulus arrives and whether it changes, not how predictable its timing is. So the case for an irregular time is an inference from stimulus specificity and from the security-warning results, not a finding. It is a reasonable one, and it costs nothing to try.

'Tidy the kitchen' is ignored; 'put the blue bin out' is not

Habituation explains why a reminder stops being seen. A second problem is a reminder that is seen and still not acted on. Pielot's participants typically viewed notifications "within minutes", whether or not the phone was on silent. Seeing was not the bottleneck.

A reminder that says "tidy the kitchen" asks the reader to work out what that means, at a moment chosen by the phone rather than by them. One that says "put the blue bin out" names a single physical action. Why starting is harder than doing sets out the evidence that concrete, specific plans are easier to begin than outcomes, and it applies with more force to a prompt you will glance at for one second.

Write the reminder as the first action, not the goal. If the first action is not obvious, the job is not ready to be a reminder yet; it needs the ten minutes at a table that turns it into one.

The snooze that is really a re-decision

Snoozing feels like a small delay. Functionally it is a decision not to do the job now, taken again each time the prompt returns, and each repeat is another presentation of the same stimulus. A reminder snoozed four times has been seen five times, which on the habituation account makes it fade faster.

The honest options when a reminder arrives are three: do it, move it to a specific later time on purpose, or delete it because it was never going to happen. Each ends the loop. A notification that offers a button for "done" and another for "tomorrow" is closer to that than one that offers only "snooze 10 minutes", because it forces the choice rather than deferring it. This is reasoning from the mechanism above, not a tested result.

What a prompt has to do to survive a month

Pulling the evidence and the inference together, a household reminder has a better chance of still working in week four if it meets these conditions. The first two rest on the research above; the rest are this piece's inference from it.

It is rare. Fewer prompts means each one is a less frequent stimulus. Remove every reminder you have learned to ignore; each is one more alert competing with the ones that matter.

It names one physical action, not an outcome.

It can be finished from the prompt itself, or ended with a deliberate choice of a new time, rather than snoozed.

It does not always arrive at the same minute, and ideally does not always look the same.

It is not a substitute for a fixed time where a fixed time is the point. The bin goes out on bin night, and a reminder that wanders past the collection is worse than none.

Barriers builds the irregular part in for one kind of task. A task marked as a Quick task is brought up a few times at most during the day, at a random minute inside windows between morning and early evening, never during the hours you have marked as quiet, and the notification names the job and carries Done and Tomorrow buttons that work without opening the app. A task with a deadline keeps its own reminder for that deadline.

---