A narrow path crosses a meadow. It is a useful image of a repeatedly travelled route, but not a picture of our brain. As an action becomes more familiar, it leaves no visible strip of worn grass. How can researchers nevertheless investigate something becoming less dependent on deliberate thought? One particular study offers both an approach and a lesson in its limits.
An experiment in everyday life
Phillippa Lally and colleagues studied habit formation beyond a single laboratory session. Ninety-six volunteers selected an eating, drinking or activity behaviour that they were asked to repeat daily in the same context. The investigation lasted twelve weeks. That design matters: it concerned self-selected, manageable actions in daily life, rather than every imaginable form of human behaviour. The article first appeared online in 2009 and in the journal’s 2010 volume. These are publication dates, not the beginning of habit research. [1]
What was actually measured?
Among the measures were participants’ reports of how automatic their chosen behaviour felt. Automaticity was therefore a quantity assessed through questions. Nobody observed a path being worn inside a mind. This distinction matters: reports about personal experience can supply useful evidence, while remaining different from a direct record of every action or a measurement of brain activity. Our meadow illustration cannot make the measured quantity visible. It provides an invitation to think, rather than an alternative instrument for measuring habits. [1]
From an intention to an analysable curve.
Four distinct levels of the historical investigation. These are not stages every person must pass through.
- Selection
96 volunteers
They select an eating, drinking or activity behaviour for a twelve-week investigation. [1]
- Data
82 with sufficient records
These participants provide enough data for analysis. [1]
- Fitting
62 fitted patterns
The intended nonlinear curve can be fitted to these individual patterns. [1]
- Model fit
39 with a good fit
The abstract describes the fit as good for this further subset. [1]
Numbers from the Lally et al. abstract; subsets narrow at successive analytical steps. A good model fit is not the same as a successful life change. [1]
One group becomes several subsets
The abstract reports 82 participants providing sufficient data for analysis. The intended curves could be fitted for 62, with a good fit for 39. These numbers are not incidental. An appealing pattern for part of the group does not necessarily describe all 96 volunteers equally well. When retelling a finding, we need to ask which subset supports it. Everyday experience does not always fit equally neatly into the same mathematical shape, even within a carefully framed study. [1]
Gradual change rather than a switch
The model describes an approach to a plateau. Estimated time to reach 95 percent of that plateau ranged from 18 to 254 days. This is not an observed deadline for everyone: 254 days extends well beyond twelve weeks of observation. The curve describes a gradual pattern, rather than a switch suddenly being thrown. The large range makes visible how poorly a single number of days can summarise the different individual patterns. Some estimates necessarily look beyond the recorded period. [1]
Why 21 days is no natural law
The UCL discussion published in 2012 explicitly challenges a universal 21-day rule. It emphasises repeating an action in a recurring context. An occasion within the daily routine can become associated with the behaviour. This is more specific than simply declaring that persistence for a fixed period is enough. Yet the explanation does not provide a replacement date guaranteed to apply to every person and project. An average, wherever one is reported, cannot function as a personal promise. [2]
A gap is not necessarily a restart
The abstract also reports that missing one opportunity did not materially affect the habit-formation process studied. This offers an interesting counterpoint to the idea that one interruption destroys all previous progress. It is a specific finding, rather than a guarantee about any conceivable longer break. A drinking routine and responding to distressing thoughts are not interchangeable activities either. The experiment describes selected actions; it does not judge participants’ character or the value of the other habits in their lives. [1]
The occasion can be the better observation
For a personal note, one precise question can help: what happened immediately before a particular action? Someone might record whether the invented example “taking a glass of water” always follows the same meal. That would be a private observation, rather than a replication of the scientific study. The important distinction is between the occasion, the action and one’s interpretation. Keeping them apart makes it possible to recognise what was actually observed and what was merely inferred from it.
An open question remains interesting
How do actions change when their familiar setting disappears? Our sources provide no individual timetable for that question here. It can nevertheless change the way we look: a familiar action may be less exclusively an expression of an unchanging personality than we assume. The meadow path remains a metaphor. The study examines one limited part of change; a closer look means recognising that boundary without treating the process under investigation as unimportant because it is not universal.
The imagined meadow path is an artistic metaphor, not a depiction of brain processes. The explanation cards organise the study design and analysis; they show neither a measured time series nor an individual success curve.
Lally et al. · How are habits formed: Modelling habit formation in the real world ↗
Original study, first published online in 2009, journal volume dated 2010. The publicly visible abstract was inspected, not the full text. Ninety-six volunteers selected a behaviour; repetition and self-reported automaticity were studied over twelve weeks. Modelling was possible for subsets. Reported times depend on the model and are not deadlines for every person or behaviour.
UCL · Busting the 21 days habit formation myth ↗
UCL research explanation dated 29 June 2012. Discusses repetition in context and challenges a universal 21-day rule. From the same research environment, not an independent replication. It provides neither individual treatment advice nor evidence that every frequently repeated action is a habit.
Sources rechecked in the integrated browser on 24 October 2026. The original study is discussed through its accessible abstract. Its numbers describe a historical investigation, not a universal deadline for change.
