Twenty-One Days and Other Fictions

AAsk people why they did what they did yesterday and they will give you reasons; ask them how many of yesterday's actions actually required a decision and they will overstate the number. Studies of daily activity carried out since the 1990s have repeatedly found that between a third and a half of waking behaviour is performed in the same place, at the same time of day, and in much the same manner as the day before. Psychologists call such behaviour habitual, and they define it not by how often it occurs but by how it is produced: a habit is a response triggered directly by a situational cue, with little of the deliberation that precedes a decision. The familiar claim that a habit takes twenty-one days to form has no experimental parent. It descends from a surgeon's remark in 1960 about how long his patients took to accept an altered appearance — an observation about adjustment rather than about learning — which self-help authors have repeated so often that it has acquired the authority of a finding.

BThe account of how habits are acquired is by now settled in outline. A behaviour performed repeatedly in a stable context becomes associated with that context, and the association gradually takes over the work of starting it: in time it is the kitchen doorway, not the wish to eat better, that sends the hand towards the fruit bowl. Automaticity — assessed by scales asking whether an action is done without thinking, without planning, and before one notices — rises steeply at first and then flattens, tracing an asymptotic curve rather than a straight line. Two conditions do most of the work. The cue must be met dependably, and the response must follow it closely in time. A behaviour tied to a cue that appears only sometimes, such as a run planned for whenever the weather allows, gathers automaticity far more slowly than one tied to a cue encountered every day.

CHow long the curve takes to flatten is a separate question, and it was to answer it that Ingrid Halloway, a behavioural scientist at the Fenwick Institute, followed 284 volunteers for eighteen weeks between 2011 and 2013. Each volunteer chose one simple action to carry out daily after a cue of their own selection, and rated its automaticity every evening. The median volunteer reached ninety-five per cent of their eventual plateau after 74 days; the quickest arrived there in 19, and the slowest was still climbing when the study closed. Halloway is exact about what the median covers. It belongs to this sample and to this kind of behaviour — single actions, freely chosen, fastened to cues the volunteers already met daily. “It estimates how long repetition needs to do its work,” she has written, “not a date after which effort may be withdrawn.”

DTomas Oyelaran, who studies decision-making at the Bramley Centre, accepts Halloway's curve but not the picture of the mind that is drawn from it. On the orthodox account a mature habit runs off whether or not its outcome is still wanted, and the test of this is devaluation: the outcome is made unappealing — a favourite snack paired with mild nausea, say — and the behaviour is watched to see whether it survives. In rodents it reliably does. In people, Oyelaran argues, the evidence is thinner than textbooks imply; of the eleven human devaluation experiments he reviewed in 2021, four reported persistence, five reported none, and two obtained it only when participants were working under time pressure or holding information in memory. His reading is that human habits stay sensitive to their goals under ordinary conditions, and that automaticity declares itself chiefly when the resources for deliberation are scarce. Halloway's reply is that such scarcity is the ordinary condition of daily life.

EIf habits are anchored to contexts, then disturbing a context should loosen them — a prediction known as the habit discontinuity hypothesis, and one that can be tested wherever people move house. In 2016 a trial in three cities posted a free month-long transport pass to 800 households, half drawn from recent movers and half from residents of long standing. Among those who had moved during the previous three months, bus journeys rose by 21 per cent over the following season; among the long-settled households the same offer changed nothing the researchers could measure. The window, in other words, is not created by the pass but merely used by it. Critics observe that movers differ from stayers in more than their addresses — they are younger, and their journeys to work have changed — and that neither group was followed beyond a single season.

FCommercial interest has, as usual, outrun the evidence. Habit-tracking applications, the largest of which reported 30 million downloads by 2022, are built around unbroken streaks, and treat a missed day as damage to be repaired. Halloway's own figures offer that logic little support: volunteers who missed a single day showed no detectable setback in their automaticity, although those who missed several days within the first fortnight did. What the research does support is unglamorous — take one behaviour, attach it to a cue you already meet, hold the context steady, and expect months rather than weeks. What it does not yet support is the further step taken by many programmes, which is that habits formed in this way will outlast a change of circumstances. The studies that have measured anything have measured formation; on what becomes of a habit afterwards, the confident advice runs well ahead of the data.