Do Watching Eyes Make People Behave
A pair of eyes above an honesty box once made people pay 2.76 times as much for their milk. Pooled studies have not found the same lift in generosity, and the better case for watching eyes is a narrow one, about small acts of rule-breaking.
Answer to the question
Do watching eyes make people behave better?
Only in narrow places. Pooled studies find no effect of artificial watching eyes on generosity, while a review of 15 experiments linked eye images to a 35 per cent lower risk of antisocial behaviour. The famous 2006 coffee-room result comes from one room over ten weeks, and field tests have design limits.
Key takeaways
- Bateson and colleagues (2006): over ten weeks in one coffee room, people paid £0.417 per litre of milk consumed when eyes hung above the honesty box and £0.151 when the poster showed flowers, about 2.76 times as much.
- Northover and colleagues (2017) pooled 26 experiments (about 2,700 participants) and 27 experiments (nearly 20,000) and found no evidence that artificial surveillance cues raise generosity.
- Rotella and colleagues (2021) ran a preregistered dictator game: people felt more observed in public, but gave no more with eye images or in public.
- Dear and colleagues (2019) pooled 15 experiments with 2,035 participants and linked eye images to a 35 per cent lower risk of antisocial behaviour.
- Nettle and colleagues (2012): thefts fell 62 per cent at bicycle racks with eyes and a verbal message, while the rest of the campus saw a 65 per cent rise.
A pair of eyes above an honesty box made people pay 2.76 times as much for milk as flowers did (Bateson and colleagues, 2006).
That result made the watching eyes effect famous, and it is why a drawn pair of eyes turns up in books about hidden motives. Kevin Simler and Robin Hanson cite it in The Elephant in the Brain (2018, p. 60) as evidence that people are built to bend norms when they think nobody is looking. So do pictures of eyes make people behave better? Sometimes, in narrow places. Watching eyes have a respectable record on small acts of rule-breaking and a poor one on giving.
What happened when eyes hung above a coffee-room honesty box
The coffee-room study ran for ten weeks. A notice above the honesty box alternated weekly between a pair of eyes and a picture of flowers, with a different image each week. The eyes varied in sex and head orientation but all looked straight at the viewer. The price list asked 30 pence for tea, 50 pence for coffee and 10 pence for milk. Per litre of milk consumed, people paid £0.417 in the eye weeks and £0.151 in the flower weeks (Bateson and colleagues, 2006, Biology Letters).
The figure is real for that room and those weeks. It comes from one honesty box, with ten swaps of the poster and no one present to see who paid. It showed that a picture could shift behaviour at that box, and it left open how far the shift would travel. The money for the milk was owed, so the room tests rule-keeping more than generosity. How far a single small result can be trusted is the subject of [What Forty-One Participants Buys](https://thewhyweproject.com/articles/what-forty-one-participants-buys).
Simler and Hanson report the effect in their main text and then do something unusual. In the Chapter 4 endnote attached to the sentence, they tell the reader that two recent meta-analyses by Northover and colleagues cast doubt on such effects (The Elephant in the Brain, p. 60 and Chapter 4 endnotes). Most readers never reach the note. The prose presents the watching eyes effect as settled, and the note does not.
Why did generosity stop responding to pictures of eyes?
The two meta-analyses appeared in Evolution and Human Behavior in 2017 and pooled experiments in which artificial cues of being watched were set beside giving. One combined 26 experiments and about 2,700 participants and looked at how much people gave. The other combined 27 experiments and nearly 20,000 participants and looked at whether they gave anything at all. Both found no evidence that the cues raised generosity. Earlier experiments had pointed in different directions, and the authors concluded that scepticism about watching eyes in giving tasks is warranted (Northover and colleagues, 2017, as summarised by the British Psychological Society's Research Digest).
A preregistered dictator game in 2021 pointed the same way. Rotella and colleagues had people divide money in public or private decisions, some with eye images present. Participants said they felt more observed in the public condition, yet they gave no more when decisions were public or when eyes were shown (PLoS One, 2021). An exploratory model, which the authors describe as underpowered, hinted that less prosocial people gave less than prosocial people in private but not in public, and they suggested future work look at individual differences. Why a striking result so often fails to repeat is set out in [The Psychology You Learned Didn't Replicate](https://thewhyweproject.com/articles/the-psychology-you-learned-didnt-replicate).
A person in the room adds what a picture cannot
A person in the room does more than a picture on the wall.
A meta-analysis of 117 papers on observability in general, not on eye pictures, found a small positive link between being observed and behaving prosocially. The link was stronger when a passive observer was present than when participants only perceived that they were being watched, and stronger when decisions had real consequences and in laboratory than in field or online studies (Bradley and colleagues, 2018). It did not test eye pictures, so the comparison is indirect, but it points to a real watcher carrying weight that a drawn pair of eyes may not.
Bicycle racks tell a different story from charity tins
Where the behaviour is antisocial rather than generous, the record is better. A systematic review and meta-analysis by Dear and colleagues (2019) combined 15 experiments from 13 papers, with 2,035 participants. The authors had expected the effect to be steadier for antisocial behaviour than for giving, given how sensitive people are to threat. Photographs or stylised images of watching eyes were associated with a 35 per cent reduction in the risk of antisocial behaviour. The authors set that against the 16 per cent reduction that systematic reviews estimate for CCTV cameras, and argued that cheap eye images deserved a place in crime-reduction policy.
The best-known field test is a bicycle study at a university. Daniel Nettle and colleagues put signs with watching eyes and a related verbal message above bicycle racks at three sites with high rates of bicycle stealing and followed reported thefts for twelve months before and after. Thefts fell 62 per cent at the signed sites, while the rest of the campus saw a 65 per cent increase (Nettle and colleagues, 2012, PLoS One).
The authors allow that the signs may have displaced offending to places without them. Tom Stafford, writing in Mind Hacks in 2013, added three problems. No sign with only the message separated the eyes from the words. Owners who cared about security may have chosen the signed racks. And a thief warned that some bikes are watched can pick an unwatched rack, which is a rational choice and not a nudge from a pair of eyes.
Our reading is that watching eyes are a weak, local and cheap deterrent for small rule-breaking, and not a lever on goodness. Charities that print eyes on collection tins should not expect more giving, and the case that real audiences move generosity is made in [Public Charity and Who Is Watching](https://thewhyweproject.com/articles/public-charity-and-who-is-watching). A council weighing eyes for a theft hotspot has a better case, and a fair trial there would add a comparison sign carrying the message without the eyes.
For small rule-breaking the direction of the evidence is supported, at a size that fifteen experiments can only roughly estimate. For generosity it has not held up.
The coffee room is better read as rule-keeping than as generosity, since the money for the milk was owed.
Common questions
Do watching eyes make people behave better?
Only in narrow places. Meta-analyses find no effect of artificial watching eyes on generosity, while a review of 15 experiments linked eye images to a 35 per cent lower risk of antisocial behaviour. The famous 2006 coffee-room result comes from one room over ten weeks, and the field tests have design limits.
Do watching eyes make people give more to charity?
The evidence says no. Two meta-analyses covering 26 and 27 experiments found no evidence that artificial surveillance cues raise generosity, and a preregistered dictator game in 2021 found no extra giving with eye images, although people felt more observed.
Was the coffee-room study wrong?
It is a real result from one room. Per litre of milk, people paid £0.417 in the eye weeks and £0.151 in the flower weeks. Later pooled evidence on giving did not match it, so it is better read as a result about paying what was owed.
Does a real observer work better than a picture of eyes?
A meta-analysis of 117 papers on observability found a small positive effect, stronger with a passive observer present than when people merely perceived being watched. It did not test eye pictures, so the comparison is indirect.
Sources
Bateson, M., Nettle, D., & Roberts, G. (2006). 'Cues of being watched enhance cooperation in a real-world setting.' Biology Letters, 2(3), 412-414. doi:10.1098/rsbl.2006.0509
Northover, S. B., Pedersen, W. C., Cohen, A. B., & Andrews, P. W. (2017). 'Artificial surveillance cues do not increase generosity: two meta-analyses.' Evolution and Human Behavior, 38(1), 144-153. doi:10.1016/j.evolhumbehav.2016.07.001
Rotella, A., Sparks, A. M., Mishra, S., & Barclay, P. (2021). 'No effect of watching eyes: An attempted replication and extension investigating individual differences.' PLoS ONE, 16(10), e0255531. doi:10.1371/journal.pone.0255531
Bradley, A., Lawrence, C., & Ferguson, E. (2018). 'Does observability affect prosociality?' Proceedings of the Royal Society B, 285(1875), 20180116. doi:10.1098/rspb.2018.0116
Dear, K., Dutton, K., & Fox, E. (2019). 'Do watching eyes influence antisocial behavior? A systematic review & meta-analysis.' Evolution and Human Behavior, 40(3), 269-280. doi:10.1016/j.evolhumbehav.2019.01.006
Nettle, D., Nott, K., & Bateson, M. (2012). 'Cycle thieves, we are watching you: Impact of a simple signage intervention against bicycle theft.' PLoS ONE, 7(12), e51738. doi:10.1371/journal.pone.0051738
Stafford, T. (2013). 'Did the eyes really stare down bicycle crime in Newcastle?' Mind Hacks, 16 May 2013. https://mindhacks.com/2013/05/16/did-the-eyes-really-stare-down-bicycle-crime-in-newcastle/
Simler, K., & Hanson, R. (2018). The Elephant in the Brain: Hidden Motives in Everyday Life. Oxford University Press, p. 60 and Chapter 4 endnotes