How distance from food automatically reduces consumption.
What the evidence says
- Placing food farther away reduces consumption — SMD −0.60 (95% CI −0.84 to −0.36) in the 2019 Cochrane review, a moderate effect the review itself rates LOW-CERTAINTY, with risk of bias among its GRADE reasons (corrected 2026-09-10 from ”mainly for inadequate blinding”, which is not in the paper’s text)[1],[2]
- The proximity effect may not depend on conscious engagement — the authors’ own hedge, twice: they report ”the first evidence that the proximity effect may not rely on conscious engagement to affect behaviour” and that it ”may operate non-consciously”, crediting the concept to Hollands et al. 2016. The evidence is that few participants reported noticing the manipulation — which is about attention, not about being unable to override it[3]
- Even small distances matter (20cm vs 70cm showed effects)[2]
- The effect was not moderated by cognitive resource in either study, across participants from higher and lower educational backgrounds. The authors conclude ”an effect unlikely to be moderated by cognitive resource. This indicates potential for interventions altering food proximity to contribute to addressing health inequalities, but requires testing in real-world settings”. A hopeful inference, hedged three times, and narrower than “works regardless of education”: Study 2 tested the distance-by-education interaction and found none (β = -0.38, p = 0.488), a null in one sample rather than a showing that the effect holds equally across education (corrected 2026-09-19, following `cpi03`: the earlier quotation, ”suggesting it operates via nonconscious processes” and ”supports the potential”, was the quarantined restatement’s wording, not the paper’s, which never says nonconscious; and ”education was a sampling frame, not a tested moderator” was wrong, because the paper tested it)[3]
- The Cochrane meta-regression names FOUR conditions under which the effect was greater, and this file carried one. The review’s own sentence: the effect was greater ”the farther away the product was placed; when only the targeted product(s) was available; when participants were of low deprivation status; and when the study was at high risk of bias”. The fourth belongs beside the SMD above wherever that number is quoted — an effect that grows with a study’s risk of bias is an effect whose size depends partly on how the studies were run.[1]
- The equity moderator, which this file already had: `cpi01` found larger proximity effects among low-deprivation samples than in samples mixed in deprivation — a meta-regression the authors say to read with caution and not as causal, and they write that their evidence limits any assessment of equity (corrected 2026-09-10: this read ”populations — i.e. the intervention may benefit already-advantaged groups more”, an inference that is in `cpi01`’s `pmc-excerpts.txt` and not in the paper). Both findings are real and they are at different levels: `cpi03` tested moderation within its own samples; `cpi01` compared across studies. See [[kpi01]], which carries the equity concern in full[1]
- Effect operates through increased perceived effort, not reduced salience — measured twice by different groups: `cpi07` (n = 54) and `cpi03` (n = 159 and 246) each rated effort and salience separately and each found effort moved while salience did not[2],[3]