Context Cues

What it is

The useful thing to know first is that setting is not an add-on to the habit construct. In the paper that produced the field's famous statistic, Wood, Quinn and Kashy's 2002 diary studies, a habit was defined in the analyses as a behaviour the participant reported performing just about every day and usually in the same location. Location stability is one of the two things measured. A reader who treats context as a practical tip has the architecture backwards: in the measurement that made the literature famous, it is half the definition.

Wood's own cue list, from her 2019 book, is the practical version: locations, electronics, people, times of day, and other actions. Phones are included, because a device carried everywhere is a context never left. The last item on the list is what habit stacking exploits, by attaching a new behaviour to the end of an existing one.

The mechanism claim is stronger than a correlation, and the best evidence for it is a null. Wood reports a study of 475 actively quitting smokers in Washington, DC, with daily craving reports and geotracked phones matched against a geocoded map of tobacco retailers. The expected pattern held, in that high craving plus an opportunity to buy produced relapse. The result that matters is that quitters also relapsed after entering tobacco-selling stores having reported craving of exactly zero. Exposure to purchase cues alone did it. In a complementary field test on sixty-five pub smokers under a smoking ban, almost half unintentionally started to smoke in the pub, and heavier smokers did no worse than lighter ones, so nicotine dose predicted nothing and habit strength predicted everything.

The counterpart is what cues are not. Mandatory calorie labelling in New York City from 2008 was studied across six years with more than seven thousand fast-food customers, with noticing falling from fifty-one to thirty-seven per cent by 2014. There was no effect: customers at labelled and unlabelled locations alike increased the calories purchased per meal, and the labels did not change how often people ate out. Information was supplied at the point of decision and the cue structure was untouched.

The field study behind the standard advice about moving house deserves to be reported with its own numbers. Wood, Tam and Guerrero Witt, in 2005, followed 115 students transferring into Texas A&M, assessed one month before and one month after the transfer, with forty-seven of 162 lost between sessions, so attrition runs at about twenty-nine per cent. For roughly seven in ten, the transfer meant moving to a new community. Three behaviours were tracked, exercise, television watching and newspaper reading, with habit strength measured as self-reported frequency multiplied by self-reported location stability and behaviour self-reported. Counted honestly, four indicators of context change crossed with three behaviours gives twelve regression models, and the predicted habit-by-context interaction appeared in seven of them. The abstract's own summary is that the habits survived the transfer only when aspects of the performance context did not change, and that in some cases the disruption placed behaviour back under intentional control. The result is real, and the base is one sample of transfer students with self-reported outcomes.

In effect

Relapse at zero reported craving

The Washington, DC, quitting study is the strongest research case because the rival explanation was measured and came back null. Participants who said, in the moment, that they did not want to smoke at all still relapsed after entering a shop that sells tobacco. Craving was not doing the work. That is a harder result to produce than a positive correlation between wanting and doing, and it is the clearest evidence that cue and desire are separable.

The kettle, the kitchen doorway and the biscuit

The everyday case is the one that makes the construct legible. A biscuit eaten standing at the counter while the kettle boils is not a decision renewed each evening; it is a sequence attached to a place, a time and a preceding action. The practical lever follows directly, and so does its cost. Disturb the setting enough to shake loose the thing you want gone, and you will also lose the habits you built on purpose and no longer have to think about. Context change is indiscriminate: it does not know which of your routines you approve of.

Supermarket layout and the end of the aisle

Retailers have worked this logic for a century without the vocabulary. The same product in the same shop sells differently from an end-of-aisle display because the cue structure of the route has changed, not because the shopper has been persuaded of anything. The honest version of the commercial claim is narrow: changing where a thing sits changes how often a repeated behaviour fires, and it tells you nothing about whether anybody wants it more. The New York labelling result is the mirror image, an information intervention at the point of decision that left the cue structure alone and moved nothing.

What it does not say

It does not say cues override intention. The 1998 review that founded this literature found intention still predicting frequent behaviour in stable settings with its weight more than halved rather than erased.

It does not say cue removal is selective. The same disruption that breaks an unwanted habit breaks the wanted ones built in the same setting, and no study in this literature has shown otherwise.

It does not support the strongest version of the transfer-student result. Seven of twelve regression models in one sample of 115 students, with self-reported behaviour and twenty-nine per cent attrition, is evidence of promise rather than a general law about moving house.

It does not license large correlational datasets as mechanism. Wood's own gym-proximity dataset of 7.5 million devices is correlational with self-selection into gym proximity unaddressed, and should not be used to support the claim this page makes.

Who developed it

The cue-and-response definition and the cue taxonomy are Wendy Wood's. The context-change field study is hers with Leona Tam and Melissa Guerrero Witt. The context-versus-goal dissociation evidence is David Neal's first-authored work, and the moderator that makes context stability load-bearing is Judith Ouellette's first-authored 1998 review with Wood.

---