The Habit Loop

What it is

A habit, on this account, is a circuit rather than a decision. Something in the environment cues it, a wanting follows, an action runs, and something arrives at the end that makes the whole sequence more likely next time. Then the loop waits for its cue again.

Clear's version has four stages and each one is a place to intervene. The cue is whatever the mind has learned to treat as the start of the sequence, usually a time, a place, a preceding action or another person. The craving is the motivational state the cue produces, and it is worth being precise here: the craving is not for the behaviour, it is for the change of state the behaviour promises. Nobody craves brushing their teeth. They crave the feeling of a clean mouth. The response is the behaviour itself, which happens only if it is easy enough relative to how much the person wants it. The reward is what the response delivers, and it does two jobs, satisfying the craving and teaching the mind that this particular cue is worth remembering.

From that structure Clear derives a symmetrical set of instructions. To build a habit: make the cue obvious, the craving attractive, the response easy, the reward satisfying. To break one: invert each of the four. The symmetry is the reason the model has travelled as far as it has, because it converts a vague ambition into four separate levers, any of which can be pulled without the others.

The three-stage predecessor is Charles Duhigg's, in The Power of Habit (2012), and it runs cue, routine, reward. The difference between the two is not cosmetic. Clear's insertion of craving between the cue and the behaviour puts a motivational stage in the middle of what had been a purely associative chain, which is what allows the model to explain why an obvious cue sometimes produces nothing at all. The two versions are routinely quoted as though they were the same model with different labels. They have different stage counts and different claims.

Nir Eyal's Hook Model is a third relative, and reading it against this one is instructive. It takes the same basic loop, aims it at builders rather than at readers trying to fix their own behaviour, and bolts on a fourth stage of its own, Investment, where the user puts effort in. That addition is what makes Eyal's version cyclical rather than merely repeating, because the investment loads the next cue. The popular habit literature is three books aimed at three readers: Duhigg for the general reader wanting to understand habits, Clear for the individual trying to change one, Eyal for the builder trying to engineer one into a product.

The evidence status here is unreviewed, and it should be read exactly as this site defines it. The definition is recorded and nobody here has assessed the research behind it. That is a statement about our sourcing rather than a verdict on the model, and it keeps this page definitional. It sets out what the loop claims and who claims it. It does not certify that habits work this way.

One thing can be said about the model's standing without any assessment at all, because the people involved say it themselves. Clear's contribution is synthesis. He assembled work by Duhigg, B. J. Fogg, Wendy Wood and others into a framing more memorable and more usable than the academic versions, and the vault's own note on him says as much. A useful synthesis is a real contribution. It is not the same kind of object as a finding, and the four-stage loop should be cited as a framework with named authors rather than as a result from psychology.

In effect

Wendy Wood, and where the actual research sits

Wendy Wood, a psychologist at the University of Southern California, is the most-cited habit researcher in the endnotes of the popular books, and the substantive work is hers rather than theirs: "Habits in Everyday Life", the habit-goal interface work with David Neal, and habit-intervention research with Bas Verplanken. Our notes record her finding that roughly 43 per cent of daily behaviour runs on habit and cue rather than on active goal pursuit.

That figure has not been traced to a paper by us. It appears in the vault attributed to Wood without a year or a journal, and it is reported here as what our notes say rather than as a checked number. Anyone wanting to use it should go to Wood's own work first. The general point stands without it: the popular habit literature is a set of framings sitting on top of an academic literature that has its own authors, and citing Wood directly is available to anyone willing to look.

Four o'clock, and the walk to the kitchen

The domestic version is easy to run on yourself. A time of day arrives, the afternoon flattens out, and a person finds themselves at the kitchen counter without any memory of having decided to go. The cue is the hour and the flattening. The craving is not for a biscuit, it is for a break in the afternoon. The response is the walk and the biscuit. The reward is the break, which arrives reliably, and the biscuit, which is incidental to the whole thing.

Clear's four inversions applied here would move the biscuits out of sight, make the walk longer, and supply the break some other way. Most people attack the response, which is the stage that requires the most effort to hold and the one the model suggests is least worth attacking alone.

Strava against Nike Run Club

The commercial reading is that two products in the same category, aimed at the same goal, retain users at very different rates, and that the difference is loop architecture rather than features. Strava's cue is a notification that a friend has run, which is more salient than a generic reminder. Its craving is built on kudos and leaderboards, which produce anticipation rather than mere information. Its response is low-friction, with watch integration doing the recording. Its reward is variable, since a segment leaderboard position is not knowable in advance, which is the property variable rewards get their power from.

The comparison is in our notes as an assertion with no source attached. No retention figures for either product have been checked here, and the claim that the curves differ dramatically is the note's, not a company's or a researcher's. The analysis is a good demonstration of how the four stages are used in practice. It is not evidence about Strava.

What it does not say

It does not say who originated it, and neither does our vault. The note's frontmatter records Charles Duhigg as originator, at medium confidence, with the year unknown, while the body of the same note describes the four-stage model as Clear's. Both are partly right and the field record is layered: Duhigg popularised cue, routine, reward for a general audience, Clear added craving and reframed the set as levers, and the research both are drawing on is Wood's and B. J. Fogg's. A frontmatter attribution held at medium confidence with no year is the exact shape the vault's 2026-09-01 audit flagged as unreliable, and it is why this page carries a shared attribution rather than a name.

It is a description, not a mechanism. The loop names four things in a sequence. It does not explain what makes a cue become a cue, how many repetitions are needed, or why some behaviours automate readily and others never do. Popular writing tends to supply the answer "66 days" here, and the underlying paper reported a median of 66 days with a range of 18 to 254, from a subset of participants whose data fitted a model. The single number is the least informative thing in it.

Popularity is not evidence. Atomic Habits has sold in the millions and is the dominant popular account of habit formation. That says nothing about whether the four stages are the right carve-up, and the tag on this page says nobody here has checked.

The Strava comparison is illustration. No retention data has been verified for either application. Treat that section as a worked example of the model's vocabulary and not as a finding about running apps.

Who developed it

No single person, which is why the frontmatter records a shared attribution. Charles Duhigg, a journalist, popularised the cue, routine, reward loop in The Power of Habit (2012). James Clear added the craving stage and the four inversions in Atomic Habits (2018), and his own account of his contribution is synthesis rather than original research. Wendy Wood, at the University of Southern California, produced much of the academic work all of them draw on, and B. J. Fogg supplied the model of when a behaviour fires at all. Nir Eyal built the product-design descendant. Of that group, only Eyal has a page on this site so far. Duhigg, Clear, Wood and Fogg do not.

---