Rewards build habits, but they are not what keeps a formed habit running. Once an action has been repeated often enough in the same setting, the setting itself sets it off, and whether the action still pays off matters far less than the popular version of the habit loop suggests.

In the account of how habits work that the research supports, you repeat an action in a stable context, usually because it gets you something you want, and your memory gradually links the context to the action. After enough repetition, the context alone brings the action to mind and starts it, with little thought and limited regard for the reward.¹

Repetition in a stable setting is how habits work

Habits work by association, according to a 2016 Annual Review of Psychology article by Wendy Wood and Dennis Rünger of the University of Southern California. Each time you repeat an action in the same context, the mental link between the two strengthens a little, until simply noticing the context activates the action.¹

Their example is making coffee: while pursuing the goal, you repeatedly pair cues such as the coffee filter with actions such as measuring out the grounds. The reward shapes this learning early on. Midbrain dopamine signals, triggered when a reward differs from what was expected, help stamp in the link between the action and the cues around it, and those signals become less active as the reward recurs. The review concludes that habits formed this way are insensitive to later shifts in the reward’s value.¹

Rewards build habits, but they are not what keeps a formed habit running.

A field test of that conclusion used popcorn. In a 2011 field experiment at a university cinema, people with a strong habit of eating popcorn at the movies ate about 63 percent of their box whether the popcorn was fresh or a week stale, even though they rated the stale popcorn worse.²

The study

Limited evidence

Week-old popcorn at a campus cinema: two field experiments, 2011

Moviegoers were randomly given fresh popcorn or popcorn popped 7 days earlier, then watched 15 minutes of trailers. People who rarely ate popcorn at the movies ate much less of the stale batch, while those with strong habits ate about as much stale as fresh, and neither hunger nor liking the popcorn changed that pattern. The habit held only in its usual setting: watching music videos in a campus meeting room, strong-habit participants ate less stale than fresh popcorn. In a second experiment, moviegoers with moderate habits who ate with their nondominant hand ate less stale than fresh popcorn; for strong-habit eaters the drop fell just short of statistical significance.²

The lesson for anyone trying to change a habit is that the setting and the familiar way of doing the action, not the taste of the reward, carried the behavior; in the authors’ reading, disrupting either let people’s preferences guide eating again. The caveat is scope: one snack, two experiments by one group, and habit strength rated from a single question about how often people usually ate popcorn at the movies.²

The three parts of the habit loop: cue, routine, reward

Charles Duhigg’s 2012 book The Power of Habit made the loop famous: a cue, a routine and a reward, which he presents as the neurological loop behind every habit, based on work by researchers at MIT. Research supports all three parts, though it gives them somewhat different jobs from the book’s.³

1. 1Cuea place, time, person or preceding action
2. 2Routinethe action the cue brings to mind
3. 3Rewardstrengthens the cue-routine link while the habit forms

Then repeat from “Cue”

The cue

Duhigg sorts cues into five kinds: a location, a time, an emotional state, other people and the action you have just finished.³ Wood and Rünger’s review lists much the same triggers, including features of the physical setting, other people and preceding actions in a sequence.¹

What makes a cue useful is consistency. In a habit-formation study led by Phillippa Lally at University College London, volunteers attached a new behavior to an event that happened every day and only once a day, such as eating a piece of fruit with lunch.⁴ Gardner, Lally and Wardle suggest choosing a cue inside an existing daily routine, such as going on a lunch break, because it is a stable starting point.⁵

The routine

A habit is not quite the same as the behavior it produces. In a 2015 review of how health researchers use the word, Benjamin Gardner, then at University College London, defined habit as the process by which a cue generates an impulse to act, learned through a cue-response link. That impulse can compete or combine with conscious decisions, so it does not always turn into behavior.⁶

Gardner also separates starting from doing. A habit may get you out of the door to exercise, while the workout itself still takes thought; for simple actions, such as drinking water, starting and doing are hard to separate.⁶

The reward

Rewards help a habit form, but how they arrive matters. Animal research summarized by Wood and Rünger finds that habits develop most readily when rewards come only after some time has passed, so each action does not feel tied to its payoff. The same review reports that planning and deliberate choice during repetition can hinder habit formation.¹

Obvious rewards may not even be necessary. Lally’s volunteers were paid only for taking part, not for the new behavior, which the authors take to show that external rewards are not required, although the self-chosen behaviors were probably rewarding in themselves.⁴

1. Cue: a place, a time, other people or the action you just finished
2. Routine: the action the cue brings to mind
3. Reward: while the habit forms, it strengthens the link between cue and routine
4. Formed habit: the cue can start the routine even when the reward has lost its appeal

Where the popular loop goes further than the evidence

The popular loop claims more than habit research supports on three points: that craving the reward is what makes a habit automatic, that swapping the routine is the way to change a habit, and that the loop has been demonstrated for human habits in general.

Take craving first. A Random House study guide to Duhigg’s book summarizes its claim that a cue and a reward are not enough: the cue must also trigger a craving for the reward before the behavior becomes automatic.⁷ Research defines a formed habit by the opposite property. In Wood and Rünger’s review, the standard lab test is whether people keep repeating a well-practiced response after its reward has lost its value, for example after eating a food until they want no more.¹ Craving may describe some habits well, but it is not how researchers define or test a habit.

The same guide states the book’s golden rule of habit change: keep the cue and the reward, and insert a new routine.⁷ Wood and Rünger add a complication. Even after a new habit is learned, the old memory traces are not necessarily replaced, so the old habit can return when its cues reappear. Their review describes two complementary tactics, disrupting the cues that trigger an unwanted habit and repeating a new behavior until it becomes habitual, and notes that life changes such as moving house or switching jobs remove many old cues, opening a window to act on new intentions. Even if-then plans for acting on your intentions fail to control many strong habits, the review notes. They also argue that the difficulty of changing established habits does not reflect a failure of willpower.¹

The human evidence is also thinner than the loop’s confidence implies. In 2018, Sanne de Wit of the University of Amsterdam and colleagues noted that only one published study had shown that extensive training creates rigid habits in people, and reported five failed attempts to produce such habits in the lab, including two attempts to replicate that original study.⁸ Gardner’s 2015 review adds that most habit research is correlational and relies on self-report measures.⁶

Further reading

• 

  Good Habits, Bad Habits: The Science of Making Positive Changes That Stick

  by Wendy Wood

  Wood's book on how context and repetition build habits, from the researcher this article relies on most.

As an Amazon Associate WiserHours earns from qualifying purchases.

How much of daily life runs on habit

The often-quoted 43 percent comes from diary studies of US college students, not from a measure of adult life in general. In 2002, Wendy Wood, Jeffrey Quinn and Deborah Kashy asked 279 Texas A&M University undergraduates to note what they were doing each hour for one or two days. In one study 35 percent, and in the other 43 percent, of the recorded behaviors were done just about every day and usually in the same place.⁹

So the figure describes the share of recorded activities that met a definition of habit, among students, over one or two days. The same diaries showed why habits are easy to overlook: during habitual behavior, students’ thoughts often wandered to something unrelated, while during nonhabitual behavior their thoughts tended to match what they were doing.⁹

People also underrate their habits when explaining themselves. In two preregistered studies published in 2022, Asaf Mazar and Wendy Wood ran an online experiment on helping with 809 adults and followed everyday coffee drinking by 112 US college students for a week. Habit strength predicted behavior as well as or better than inner states such as mood, yet participants explained their behavior mainly by those inner states.¹⁰

The 66-day figure is a median, not a deadline

The 66 days comes from Phillippa Lally’s 2010 study at University College London and is a median, not a rule. Among the 39 volunteers whose data fitted the researchers’ model, the time for a daily behavior to reach 95 percent of its peak automaticity ranged from 18 to 254 days.⁴

The 96 volunteers, mostly graduate students with an average age of 27, chose a new eating, drinking or exercise behavior and rated daily how automatic it felt for 84 days. Of the 82 who provided enough data, the growth curve fitted well for 39, or 48 percent. The study lasted 84 days, so longer times are projections from each person’s curve. Eating, drinking and exercise behaviors showed no clear difference in time taken (medians of 65, 59 and 91 days, in groups of 10 to 15 people), and missing a single day did not materially affect progress.⁴

Myth

It takes 21 days to form a habit.

Fact

The 21 days appears to come from anecdotes about plastic surgery patients adjusting to their new looks. In a 2010 study of 39 people forming daily habits, times ranged from 18 to 254 days.

The 21 days has a much thinner origin. Gardner, Lally and Wardle trace it to anecdotal reports of plastic surgery patients who typically adjusted to their new appearance within 21 days, and suggest telling people to expect about 10 weeks of daily repetition instead.⁵ A 2024 systematic review by Ben Singh of the University of South Australia and colleagues links the figure to Maxwell Maltz’s book Psycho-Cybernetics. Of its 20 studies with 2,601 participants, only 4 reported how long habits took, Lally’s among them: medians of 59 and 66 days, means of up to 154 days, and individual times from 4 to 335 days. The authors rated 11 of the 20 studies, Lally’s included, at high risk of bias.¹¹

The habit loop, claim by claim

The three parts of the loop hold up in research, while its stronger claims about craving, habit change and speed do not. The table sets each popular claim against the best evidence we found, with the same evidence labels used elsewhere on WiserHours.

Using the loop to build or break a habit

The research version of the loop points to two levers, according to Wood and Rünger’s review and the 2012 article by Gardner, Lally and Wardle: repeat a new action after the same daily cue until it runs on its own, and weaken an old habit by changing or avoiding the cues that trigger it.¹⁵

For a new habit, Gardner, Lally and Wardle advise choosing one simple action, tying it to a cue you meet every day, doing it every time you meet that cue, and expecting it to get easier over about 10 weeks. Because you cannot build a habit of not doing something, they suggest framing the goal as an action to add rather than one to give up.⁵

For an old habit, the popcorn study’s authors suggest two routes: avoid the setting, or change the way you perform the action so it is no longer automatic, as eating with the nondominant hand appeared to do in their second experiment.² Wood and Rünger also report that in research on how people held back unwanted habits in daily life, the most successful strategy was telling yourself not to do it and watching for slip-ups, which worked by strengthening control rather than weakening the habit.¹

WiserHours

When an old habit needs to go

• Avoid or change the setting that sets it off
• Try doing the action a different way, such as with the other hand (tested in one small popcorn study)
• Tell yourself not to do it, and watch for slip-ups
• Repeat a new action in its place, framed as something to do rather than to stop
• If you move house or change jobs, start the new routine early, while the old cues are gone
• Watch for the old habit when its old cues return, since it can come back

Find this checklist online:
https://wiserhours.com/habits/how-habits-work/#checklist-how-habits-work-break

The bottom line

Once repetition has tied an action to a setting, the setting does most of the work, and the reward that trained the link matters less than the habit loop suggests. Pick a daily cue, repeat one simple action after it for weeks rather than days, and if you want to stop something, change what surrounds it before you count on willpower.