Before Changing the Habit, Rebuild the Context That Has Become the Return Point

When people try to change a habit, the effort almost always goes into repetition. Keep it up daily. Log it. Never skip. Stack up the count and it will settle, as the reasoning goes. But a habit is not the behaviour itself; it holds together as a link between behaviour and context. The same behaviour, without a settled when, where, and after-what, builds no link however many repetitions accumulate. Conversely, behaviour that has been linked starts without any conscious effort. The state in which you no longer decide each time whether to begin is what a habit is.

And the context being linked to is produced by the shape of daily life. While the shape stays the same, the behaviour that links into it gets chosen from the same narrow set. So what needs checking before increasing the volume of repetition is what you are trying to link the behaviour to.

Seen from the context side, a set of questions becomes answerable. Why did three attempts all fail to continue? Why does what worked during a period of strong resolve fall away in a busy month? Why do good habits come off together with the rest after a move or a change of job? None of these can be explained by the quantity of will, and all of them explain themselves once placed on the context side.

Two things are not settled here. How many days until it settles, and which situation is optimal to place it in. The spread of individual variation in settling time is fourteenfold, and handing over an average is itself what breaks the judgement. Placement cannot be specified from outside either, because the shape of daily life differs by person. What can be handed over is the conditions a usable slot satisfies: whether the situation arrives daily in the same form, and whether what follows it stays clear. Those two hold whatever shape daily life takes.

What is treated here is both daily habits and work you want to keep going in the business. The two cannot be designed the same way. Frequency, the time a single round takes, and the time until results return are all different. Apply a method that succeeded in daily life directly to the business and, when it fails, what gets doubted is not the method but your own attitude.

There is one thing left out. Behaviour at the level of dependence sits outside this article’s range. What is operating there is not context alone, so design of placement does not reach it. The dividing line shows in whether physical changes appear after stopping.

The sections below start by separating the two operations bound together in the phrase changing a habit, and by checking where the standard methods put the effort. It then places how the link is formed, and how period, will, and environment are handled, closing with where to check whether the return point has moved.

“Changing a Habit” Calls Two Separate Operations by One Name

The phrase contains two things: building a new habit and removing an existing one. Their mechanisms differ, so no single prescription works on both.

Building means creating a new link between behaviour and context. Removing means weakening a link that already exists. And links are easier to handle when built than when weakened. When building, you choose which context to link to. When weakening, the link already holds, so the context arrives first. Once the context arrives the behaviour starts on its own, so you are stopping something already in motion, and force is required every single time.

Most of the advice in circulation does not separate the two, treating them as one operation and applying the same prescription. The result is a prescription that works on one and struggles on the other.

There are effective ways to remove, too. Weakening is possible. Still, the weakening operation with the higher reliability is not stopping it by will but erasing the context. A habit is formed not by repeating a behaviour but by its binding to a context.

The word change carries a further problem: it reads as though behaviours were being swapped. Assume a swap and the work takes the form stop A, start B. Yet A and B do not necessarily link to the same context. A is linked to after-breakfast while B suits a different situation — that happens.

Start out intending a swap and you end up forcing B into A’s context. A forced link is placed where it does not fit, so it holds poorly. And when it does not hold, the record did not continue remains. The reason was that the placement did not fit, but the record carries no information about placement. What remains is only the fact of not continuing.

This property tells more the more attempts accumulate. Try again in a different place and the record still files it as did not continue again, in the same column. After three attempts and three failures, what you hold is three failures. What actually happened may be one fact: that it was placed in the same spot three times. Because the record holds no placement, the distinction cannot be made.

So the first thing to add here is neither resolve nor repetitions. It is a field in the record. Alongside what you did, write what you placed it after. One extra field turns three failures into a single piece of information.

This is not a demand for rigorous logging. It is a point about how the resolution of the material in your hands decides the options for the next move. While the resolution stays low, the only move left to choose is resolve harder.

Of the Five Standard Methods, Four Touch Only the Behaviour Side

The methods in circulation reduce to five, and four put the effort into behaviour. Only one touches the context side.

Write out clearly what you want to change. Resolve firmly. Start small. Cut the bad habit first. Use the influence of the people around you.

The first four handle the behaviour side, in the form of what to do, how firmly to commit, what size to begin at, what to stop. The fifth is different in kind. Using the influence of the people around you is the only one that handles the place where behaviour sits.

And that one works. The reason is that the people around you are part of the context. Change who you are with and the behaviour that comes naturally in that situation changes too. It moves the context, so it reaches the link.

The remaining four, while the context stays the same, amount to forcing a different behaviour into the same context. The force required continues for as long as you continue. Work on the behaviour side has its own role. Which behaviour you choose matters, and writing it out clearly is necessary, because a vague object goes nowhere.

What can be stated is that the step of deciding, after the behaviour is chosen, which context to place it in is not among the five. With no such step, placement is settled by drift.

This composition skews the load. Once a link holds, no force is needed to begin: the arrival of the context starts it. Without a link, the judgement shall I or not occurs every time, and force is consumed once per occurrence.

And judgement tips toward omission the more tired you are. So a design that depends on force to keep going falls first in the busy periods. Those periods come around regularly in any business.

The five share a premise as well: that the behaviour to be changed has already been specified. For daily behaviour it holds. Rising early, exercise, reading. They have names, and whether you did them is unambiguous. For business work it can fail. Build a route that reaches me directly can be named, but what the hour today consists of is not settled. What is not settled cannot be placed in a context.

So making business work into a habit requires one prior step. It is the step of making the work concrete down to a unit you can begin in the same form every time. Without that, whatever context you place it in, the judgement what do I do occurs at the placed moment. And that is the judgement omitted in the busy periods.

A Habit Is Not Repetition of Behaviour but a Bond with Context

Repetition does not build the link. Repetition in the same context does.

In 2010 Phillippa Lally and colleagues at University College London had 96 participants each choose one behaviour concerning food, drink, or exercise, and carry it out daily for twelve weeks in the same context (Lally et al., 2010, European Journal of Social Psychology, 40(6), 998–1009). What deserves attention is the design itself. The condition “in the same context” is built in from the start. The situation in which the behaviour is performed is fixed — after breakfast, for instance.

Participants recorded daily whether they had done it and how automatic it felt. The time to reach 95 per cent of the ceiling of automaticity had a median of 66 days. The spread of individual variation ran from 18 days to 254.

That spread is hard to explain by differences in disposition alone. In the same study, the length also differed by kind of behaviour, with exercise tending to take longer than eating or drinking. What was placed, and in which situation, can move the length.

There is another result that gets overlooked. Missing a single day had no large effect on the trajectory of the rise in automaticity. Design on the premise that one lapse ruins everything and a single interruption becomes the end of the whole thing. In fact the trajectory did not break in one day.

The same bond has been set out on the theoretical side as well. In 2007 the psychologists Wendy Wood and David Neal formulated a habit as a bond formed between a repeated context cue and a response (Wood & Neal, 2007, Psychological Review, 114(4), 843–863). The behaviour is not selected afresh from the goal each time; it is called up from the cue side.

The range needs a limit. What Lally and colleagues handled were daily behaviours, and business work — long in a single round, different in its conditions each time — is not directly covered. So the figure of 66 days is not carried over as it stands.

What does carry over is the structure: the link between behaviour and context is what produces automaticity. Take the structure and the place you touch next changes. Not the count, but what it is linked to.

The Time It Takes Is Settled by State, Not by Days

The question “how long does changing a habit take” cannot be answered with an average. A spread from 18 to 254 days is unusable in practice even after averaging.

Design around 66 days as the target and someone on the 254-day side delivers the verdict I have failed on day 66. The ground for the verdict is someone else’s median. A figure that says nothing about your own conditions produces a conclusion about you.

The widely repeated claim that 21 days makes a habit sits even further from that spread. And the shorter the target, the earlier the verdict arrives. The record of not automatic after three weeks gets filed as evidence of weak will. Filed records are carried into the next design.

So a design that sets a period as the target is safer avoided. If something is to be set, it is the state rather than the period. Has the judgement stopped occurring when you begin? That state is observable by you, independent of the day count.

Seen as a state, there is another advantage. When it has not yet been reached, a next move follows. If judgement is still occurring, what it links to is not settled, so the placement is what to revisit. Seen as days, the only move that comes out is keep going a bit longer.

The period question has a reverse misreading too. Read it as days are irrelevant and the interval of checking disappears along with them. A wide spread does not mean measurement is unnecessary. It means the object of measurement is state rather than days.

Observing a state requires a minimum period. Over a few days, you cannot separate whether no judgement occurred because of the link or because that week happened to be quiet. No value fixed from outside decides that length. The guideline used here is until the context has come around about twenty times. For a daily context that is three weeks; for a weekly one, several months.

The rereading tells in business work. On work that returns results only after months, setting days as the target exhausts the target first. Whether judgement has stopped occurring can be observed daily, without waiting for the return.

There is a side effect. Seen as a state, no date of achievement gets fixed, which makes progress hard to report outward. In days you can say thirty days running; a state is harder to put into words. Force it back into days here and the ground for the verdict moves to someone else’s median again. If an outward form is needed, counting how many times judgement arose fits your own conditions better than counting days.

Will Works Only Where a Judgement Is Occurring

Will operates in situations where do it or not has been asked. In a settled habit, that question does not arise.

The context arrives and the behaviour starts. No question arises, so nothing is called on to answer it. To intervene by will in a settled habit, you first have to make a judgement occur. Stop what has already begun automatically and turn it into a question. That operation itself takes force.

And the operation is needed every time. The context arrives daily, so at once a day that is thirty repetitions of the same operation in a month.

Even when building a new habit, will works within limits. Will tells at the moment of starting, but it cannot stand in for the work of forming the link. A link forms through repetition in the same context. The strength of will affects the number of repetitions; it barely affects the strength of any single instance of linking.

Three days on firm resolve, then sporadic irregular attempts. In terms of count this accumulates, and the context is not constant, so no link forms. Meanwhile, moderate resolve applied briefly in the same situation daily does form one.

The expectation that strong resolve makes daily repetition in the same situation possible holds for a period. Resolve decays, though, and in busy periods judgement tips toward omission. And busy periods cannot be chosen. So a design premised on strength of resolve gets tested in the worst conditions available. Fall there and the days already accumulated feel reset. The trajectory does not break in a day, but it feels that way to the person who fell.

Put will at the centre and the account of success and failure moves to the person. It continued because will was strong; it stopped because will was weak. That account carries no next move, because nobody can explain how to become stronger. Place the account on the context side and a move appears: change where it is linked.

Will has a practical use as well. Used inside a design that has reduced the number of judgements, a small quantity is enough. Concretely: the judgement that decides the placement, and the judgement that places it again after a collapse. Those two only.

It is not used for daily starting. Used there, it is consumed once per day. The same quantity of will, concentrated on the few design judgements a year, reaches further. This allocation has the same shape as elsewhere in the business: settle the judgements upstream in one batch and let the downstream run as decided.

“Cut the Bad Habit First” Hands the Emptied Context to Drift

Cutting brings two consequences with it: a judgement occurring every time, and the context being left empty afterwards.

The advice that cutting a bad habit should come before building a new one is widespread. The reason offered is that the bad habit is consuming time and attention. A habit that genuinely occupies time does free time up when removed. That part works correctly.

The first consequence is that cutting takes the form that makes a judgement occur every time, so that will can reach it. An existing link moves when its context arrives. You perform the operation of stopping something already in motion, every time. That operation is heavier than forming a new link. Put the heavy work first and force runs out before the light work is begun.

The second consequence is that the context is left empty. That a behaviour was linked to a context means the context was a moment where something happened. Remove only the behaviour and the context remains.

The remaining context does not stay empty. Another behaviour enters, and because it was not chosen, it is settled by drift. Suppose you cut what you used to watch after finishing work at night. The context after finishing work at night remains, so another behaviour of similar character enters. Since you succeeded with the target you cut, the substitution is hard to notice.

So performing the cutting work on its own makes results hard to measure. What should be measured is not the disappearance of the target but what entered that context.

Hard measurement skews judgement too. Cut an hour a day and an hour is freed on paper. Look back at the end of the month and the results of that hour may be nowhere to be found. The freed time has been absorbed by unchosen behaviour. The cutting itself succeeded, so this does not register as a failure.

Unregistered, the work proceeds to searching for the next target. Targets for cutting can be found without limit, so the work can continue indefinitely. A sense of progress is available, and the design of placement is never reached.

Cutting has one further awkward property. Its results appear as things that did not happen. What did not happen is hard to observe, so no sense of progress arrives. Work that yields no such sense is hard to sustain in itself.

So the criterion for taking up the cutting work is not how bad the target is. It is whether what you want to put in that context instead has already been decided. If it has not, that context is not yet at the stage of being touched.

Change the Context and the Habits Come Off Together

A move, a change of job, a shift in daily hours. After changes like these, habits that had been running come off in a batch. Good habits and the rest come off alike.

The phenomenon cannot be explained from the side of will. Moving house does not change the quantity of will. From the context side it explains itself. You did not stop the behaviour; the cue that used to call it up stopped arriving.

This form has been measured. In 2005 Wendy Wood and colleagues tracked how the exercise, newspaper-reading, and television habits of students changed after relocation (Wood, Tam & Witt, 2005, Journal of Personality and Social Psychology, 88(6), 918–933). Once the environment changed, behaviour that had been running automatically returned to being selected from the side of intention.

The phenomenon carries two implications. The first is that, in removing a habit, erasing the context is the more reliable way. Building a situation where the context does not arise consumes less force than stopping it by will. The second is that just after the environment changes is the easiest time to build new links. Existing links have come off temporarily, so the contexts are vacant.

Coming off is sometimes temporary. After a while a similar context is reconstructed. Even in a different place, if the structure of daily life is the same, an isomorphic context returns. So what you place during the interval before it returns is the fork. Place nothing and the old behaviour returns.

The same property works in reverse. Habits you did not want to lose also come off when the environment changes. Periods of changing the form of the business often change the structure of daily life at the same time. Good links that had held come off along with everything else. Read that as a failure of the change and the pull is toward restoring the old environment.

The reading runs like this. Coming off is evidence that the habit was supported by context rather than by will. Rebuild the support and the same habit returns. What it is rebuilt onto is a different situation inside the new environment.

And the property is usable without changing the environment, because contexts can be made deliberately. Choose one existing context that arrives reliably every day and place the behaviour there. It is less powerful than a move, and it has the advantage of being chosen.

Placements Are Limited to Slots That Arrive Daily and Stay Clear Afterwards

Two conditions govern the choice of placement: that it arrives reliably every day, and that other commitments rarely fall immediately after it. Slots satisfying both are limited to a few in a day.

Satisfying only the first is not enough: if what follows is occupied, execution does not happen. The thirty minutes before the working day arrive daily, and if meetings land right afterwards three days a week, that slot is unusable. What decides usability is not the clock time but how clear the slot immediately after is.

Searching with these two narrows the candidates considerably. The narrowing itself means something. The use of the remaining hours is decided by the jobs in hand. The range over which your own discretion actually extends is those few slots. So what you put there tells in the result.

A narrow range does not mean the choice is light. If anything the reverse: the fewer slots available, the more weight each one carries. The judgement here occurs only a few times a year, which makes it the place to concentrate will.

The unit placed has conditions too. It must be concrete enough that, when the context arrives, no judgement about what to do occurs. Work on building the route produces a judgement at the placed moment. Reread one enquiry from last week and save the reply I sent does not.

The level of concreteness is harder to reach for business work, because conditions differ each time. What works is fixing not the behaviour itself but the entrance to it. With the same entrance, the content afterwards can differ every time and the judgement at the start still does not occur.

The claim that only a few candidates exist invites an objection: there is more time than that. There genuinely is more free time. Free time and time that stays clear afterwards are different things, though. Gaps in the schedule appear in a different place each week according to the jobs. A place that differs weekly does not become a context. What works as a cue is time whose position is fixed.

And after placing it, it collapses. Aim for a design that does not collapse and the collapse gets recorded as failure. Place it on the premise of collapsing and where it returns to afterwards becomes observational material. Collapse itself is not a record of failure.

The judgement of changing placement needs one guideline as well. If judgement still occurs every time after the context has come around some twenty times, that slot does not fit. Move before the count is reached and the candidates run out without any slot having been tested through.

What Decides the Return Point Is Where the Hours Sit Daily

The return point is not on the side of resolve. Where the hours sit from day to day is what decides it.

Contexts are concrete situations inside daily life. After breakfast, at the end of a work block, in transit. These are decided by the shape of daily life. And the shape of daily life is built by where a person feels it natural to be. How many morning hours go to what, whom you meet, what gets priority. These are not re-decided daily; they simply are so.

That state of simply being so is the return point. It is the position you return to when left alone, even after moving away.

So changing a habit has two layers. The surface layer is the work of tying a behaviour to a particular context — concrete and reducible to procedure. The deep layer is the work of moving the return point itself. What it moves is not the individual links but the side that produces the set of contexts.

Handle only the surface and the links you built break each time the shape of daily life returns. That shape collapses once during busy periods or environmental changes, then returns to its original form. If the return point is the same, the form that comes back is the same. Suppose you built the habit of placing production time in the morning. The link held. Then, after two crowded weeks, the morning hours revert to preparing for meetings. They reverted because that is the original shape.

Moving the deep layer means resetting the original shape. What gets moved is neither behaviour nor context. It is the assumption about what form of work you are someone who does. Change that assumption and the order of priority changes. Change the order and what survives a busy period changes. Change what survives and the shape that comes back changes.

That said, it does not follow that behaviour changes automatically once the assumption changes. The surface work is still required. Links do not hold unless built. What can be stated is that what decides where you land when the surface work breaks is the deep layer.

The deep work has the property of invisible progress, because it does not appear in behavioural records. Records carry only the surface, so a period in which the deep layer is moving looks like a period of doing nothing. Rush to pile up surface work here and the placements get chosen by the old order of priority.

In sequence the deep layer comes first. The reason is simple: what decides the placements chosen at the surface is the deep layer. With an outdated assumption, placements get chosen by outdated criteria. Choosing again is possible only after the assumption has moved. In practice this shows up as decide first, then accumulate.

Business Work Has Different Conditions, and What Changed Is Read from the Return Point

Business work differs from daily habits in three conditions: frequency, the time a single round takes, and the time until results return.

First, frequency differs. Important business work can come round once a month. Low frequency means more time before a link forms. The response is to link not that work itself but the behaviour of gathering the material it needs into a daily context.

Second, the time per round differs. Production and design do not finish in short blocks, so they do not fit into free gaps. Without securing the slot in advance, the arrival of the context does not produce execution. The response is to secure the context as a time band.

Third, the time until results return differs. Exercise changes how you feel within weeks; changing the form of a business can take months to return anything. During a period with nothing coming back, the reason to continue is not supplied from outside. What supports that period is not results but the return point.

Bind them together under the single word habit-forming and the fact that daily life continues while business work does not looks like a problem with your own consistency. What can be carried across is the structure only; the procedures cannot.

What to check follows from this. First, can you say what the behaviour is linked to? Not “I do it daily” but “I do it after this”. If you cannot, the link has not been built yet.

Second, is a judgement occurring when you begin? If the question shall I arises, the link is still weak. If it starts without the question, it is working.

Third, where do you land after a collapse? This is the indicator that separates surface from deep. Landing in the old shape means the deep layer has not moved; landing nearer the new shape means it has. The judgement can only be made after one collapse. Guard against collapsing and it stays unverifiable.

Does stacking up repetitions make it settle? Stacking is required. Without a settled place to stack into, though, the repetitions accumulate nowhere. What decides that place is where the hours sit from day to day — the return point.

The handling of the restoring function itself is in a mental block is not a defect, the order of resetting the return point is in self-transformation does not last because the order of changing is reversed, and the case where the reset destination never comes up as a candidate is in not acting is not a failure of will. Where the link sits within recognition as a whole is in what cognitive science shows is that behaviour is generated out of the world being felt.