Conceived and designed the experiments: RB DF. Performed the experiments: DF. Analyzed the data: RB DF. Contributed reagents/materials/analysis tools: RB DF. Wrote the paper: RB DF.
The authors have declared that no competing interests exist.
A key issue in cooperation research is to determine the conditions under which individuals invest in a public good. Here, we tested whether cues of being watched increase investments in an anonymous public good situation in real life. We examined whether individuals would invest more by removing experimentally placed garbage (paper and plastic bottles) from bus stop benches in Geneva in the presence of images of eyes compared to controls (images of flowers). We provided separate bins for each of both types of garbage to investigate whether individuals would deposit more items into the appropriate bin in the presence of eyes. The treatment had no effect on the likelihood that individuals present at the bus stop would remove garbage. However, those individuals that engaged in garbage clearing, and were thus likely affected by the treatment, invested more time to do so in the presence of eyes. Images of eyes had a direct effect on behaviour, rather than merely enhancing attention towards a symbolic sign requesting removal of garbage. These findings show that simple images of eyes can trigger reputational effects that significantly enhance on non-monetary investments in anonymous public goods under real life conditions. We discuss our results in the light of previous findings and suggest that human social behaviour may often be shaped by relatively simple and potentially unconscious mechanisms instead of very complex cognitive capacities.
A central issue in evolutionary biology is to understand cooperative behaviour among unrelated and unfamiliar individuals
Research over the last three decades suggests that one important mechanism to explain cooperation in such situations is reputation or ‘indirect reciprocity’
One approach to test whether individuals modify their behaviour in the presence of others is to investigate whether they respond to being watched. For instance, mutual eye gaze has been shown to increase contributions to a public good without any change in anonymity
In the present study, we aimed at going one step further in several aspects to investigate the effects of simple cues of being watched under realistic conditions. We asked whether individuals would invest in a public good (a clean environment) even without receiving any material good (such as money) in exchange. Furthermore, we investigated whether images of pairs of eyes would enhance investments (removal of foreign garbage) among unfamiliar individuals at public locations. We tested 14 different bus stops; hence, in contrast to many studies that only used students at the University, our sample comes closer to a random sample of the population. In order to obtain information about the precise investments of each individual we directly observed people's behaviour. Finally, we combined our behavioural data with a questionnaire to investigate whether images of eyes have a direct effect on cooperative behaviour, or whether the presence of eyes caused an increase in investment merely by enhancing attention towards a sign that requested cooperative behaviour (here: disposal of garbage).
If reputational effects cause an increase in investment in a public good, we predicted that cues of being watched (an image of eyes) enhance the likelihood and the amount of investment involved in removing garbage from a public location (a bus stop bench) compared to controls (an image with flowers). If images of eyes have a direct effect on behaviour that is not mediated by an increase in attention towards a sign requesting garbage removal, we predicted that the sign would not be noticed more often in the presence of eyes, compared to in the control.
DF conducted the experiments at 14 different bus stops in Geneva (Switzerland). The tests were carried out at bus stops that were equipped in a similar way (with a bench and a bus stop shelter,
Three items of garbage (2 PVC bottles (1a) and a newspaper (1b)) were placed on the bus stop bench. Two experimental wastebaskets with signs, one for plastic (2a) and one for paper (2b) were placed, one at each side of the bus stop bench. Above the bus stop bench (in eye height, about 1.5 m) a symbolic sign (3) indicating to throw away garbage was attached. Below the sign either an image of eyes or flowers (4) was placed.
Each test started when a bus left the bus stop, so there was a “clean” experimental situation without any people present. We observed the behaviour of the first subject approaching the bus stop for 2 minutes and recorded the following information: whether or not subject handled (handling refers in all cases to deposition of objects into the bins) garbage, duration of handling items (seconds), location of depositing the items (in which of the two bins the item was deposited), and whether subjects remained standing or sat down on the bench. Furthermore, we recorded the gender of the person. Two minutes after arrival of the subject, we approached the subject to conduct our questionnaire. Treatments (eyes/flowers) were exchanged about every 2 hours and the type of eye and flower was presented in random order.
After the observations ended we approached the persons and asked the following questions: Did you notice the sign? Did you notice the eyes/flowers? Did you feel observed by me (the observer)? After the questionnaire, subjects were informed about the underlying reasons for the experiment and the questionnaire. All data were recorded and analysed anonymously.
We only included observations in our analyses in which the conditions concerning our question of interest were appropriate, i.e., subjects were alone at the bus stop for at least 2 minutes and subjects positioned themselves in front of the bus stop so they could see, and potentially be affected by the sign and the experimental garbage. We first analysed the overall effect of the treatment on whether or not individuals engaged in garbage clearing. For the detailed analyses of behavioural differences depending on the treatment we only analysed those cases in which people engaged in garbage removal. This is because individuals that did not clear garbage are neutral to the treatment in the sense that it is not possible to assess the effects of the treatment on their behaviour. In a first step, we tested whether individuals invested more time in handling garbage in the presence of images of eyes compared to the control. In a second step, we investigated whether the amount of time invested was positively correlated to the number of items handled, the number of items place into the correct bin or, the proportion of items placed into the correct bin (precision). Because of the low sample size of people that were willing to respond to our questionnaire, we included all available responses to the questionnaire in the analyses, including those of people that were excluded from the behavioural analyses because of methodological considerations (e.g., when the person was not alone during two minutes after arrival). As some individuals left with the bus before they had answered the full questionnaire we do not have the same amount of answers for each question (between 18–30 answers per question could be analysed). All results reported are two-tailed and were performed with SPSS version 16.0. P-values<0.05 are considered as statistically significant and p-values> = 0.05 but <0.1 are reported as non-significant trends.
For our study we did not obtain a formal ethical permit from the ethical board because we instead avoided all manipulations that could have raised ethical concerns. We adhered to the Ethical standards of the Suisse Society for Psychology (Schweizer Gesellschaft für Psychology) and to the checklist for ethical consideration of psychological studies (“Checkliste für die ethische Beurteilung von Psychologischen Forschungsvorhaben”, see
About one third (28 of 93) of the subjects placed garbage into the bins. When testing all persons present at the bus stop, the treatment had no significant effect on the likelihood that individuals would handle (handling allways refers to depositing objects into the bins) garbage items (NFlowers_no-handling = 32; NFlowers_handling = 12; NEyes_no-handling = 33; NEyes_handling = 16; Chi2 = 0.32; p = 0.57). Individuals who sat down on a bus bench tended to be more likely to handle garbage items compared to individuals who remained standing (Chi2: NStanding_no-handling = 32; NStanding_handling = 8; NSitting_no-handling = 33; NSitting_handling = 20; Chi2 = 3.41; p = 0.065).
Using only the data from people that engaged in handling garbage (and therefore were likely subject to the treatment), people invested about twice the time in handling garbage in the presence of eye images as opposed to flower images (
The amount of time people invested in handling garbage items correlated positively with the number of items handled (Spearmen's rank correlation, N = 28, c = 0.67, p<0.001) and with the number of items people deposited into the correct bin (N = 28, c = 0.50, p = 0.007). However, the proportion of items deposited into the correct bin (i.e. precision) was not correlated with handling duration (N = 27, c = 0.08, p = 0.69). Significant results remain significant when adjusting the critical p-value for multiple testing (critical p = 0.017).
Using the data from the questionnaire we tested whether eyes could have caused the effect of higher investment in the treatment by raising the attention towards the sign that instructed subjects to throw away garbage. We analysed whether individuals noticed the sign more often when eyes were present compared to the control. Individuals noticed the sign more often when flowers were present compared to in the presence of eyes (NFlowers_not noticed = 5; NFlowers_noticed = 8; NEyes_not noticed = 13; NEyes_noticed = 4; Chi2 = 4.43; p = 0.042), suggesting that the effect of the eyes cannot be explained by increased attention towards the sign requesting garbage removal. There was no difference concerning how often people reported to have noticed the experimental eyes or flowers (NFlowers_not noticed = 8; NFlowers_noticed = 5; NEyes_not noticed = 12; NEyes_noticed = 5; Chi2 = 0.27; p = 0.60). None of 18 individuals responded to have felt observed by the experimenter in the opposite bus stop (NFlowers = 8, NEyes = 10).
A central issue in the study of human cooperation is the question about which mechanisms promote investments into public goods. While theoretical and lab studies showed that reputation can enhance cooperation under certain conditions
Reputation has been shown to foster cooperation also in public good games
Based on this reasoning, we suggest that the most parsimonious explanation for our results is that images of eyes triggered a reputational response and that the personal benefits individuals derived from maintaining a good reputations produced investments in the public good
A second potential explanation for our findings is that images of eyes have been perceived as a direct threat of punishment (instead of the more indirect threat of obtaining a bad reputation). Punishment has been suggested to enhance cooperation across taxa
About one third of the subjects cooperated by disposing garbage. This proportion seems relatively high, given that individuals did not obtain any material benefits from their contribution. Overall (including all individuals that were present at the bus stop) people were not more likely to engage in garbage clearing in the presence of images of eyes. Hence, it appears that cues of being watched were not sufficient in our setting to trigger a behavioural response. However, this interpretation should be treated with caution. Three reasons might have contributed to the lack of a treatment effect on whether or not people engaged in garbage clearing.
First, an implicit rule might have been acting only for people who wished to sit down. This is the rule: “If you produce garbage, then you must dispose of it properly.” As we placed the garbage ourselves at the bus stop, this rule likely did not affect people who remained in some distance from the bus stop bench. However for people who wished to sit down this rule likely became relevant. These people were in a way “forced” to “take possession” of the garbage, as a potential passer-by could assume that the passenger had generated the garbage and was littering. The finding that the motivation to sit down appeared to trigger removal of garbage corroborates this, as persons who sat down tended to be more likely to engage in garbage removal. However, if individuals only wanted to sit down, they could have done so by simply pushing aside a garbage item without even touching it with the hands (personal observation). Moreover, in order to sit down, it would have been sufficient to remove only one of the three items. Instead, half of the persons removed and deposited more than one garbage item into the bins. Hence, individuals invested more than necessary to sit down and this amount of surplus investment appears to have been modulated by the presence of eyes.
Secondly, not all people who were present at the bus stop might have been affected by the treatment. In contrast to the situation in a University kitchen, in which individuals prepared their drink in front of the experimental cues of being watched
Third, handling of foreign garbage creates a high threshold for actually engaging in behaviour as many people might sicken at the idea to handle potentially dirty foreign rubbish. The threshold involved in engaging in removal of foreign garbage (e.g., potential risks of infection) may be too high to be surpassed by subtle cues of being watched.
In our field study we encountered several additional sources of variation compared to other studies, partly resulting from our attempt to enhance realism. These may have increased the variance in our data and thereby might have decreased the size of the effect. By placing a sign to dispose garbage we may have raised the attention to an injunctive norm (disapproval of littering). Studies suggest that injunctive norms coupled with a conflicting descriptive norm (in our case the presence of litter indicating that people litter at this bus stop) might reduce the likelihood to engage in cooperative behaviour (here: antilittering)
Overall, a number of studies (including this study) found effects of cues of being watched on behaviour
Many studies that investigated contributions to public goods have been conducted under highly controlled lab conditions that allow for explicit testing of particular parameters of interest
Theory and reality about human behaviour have repeatedly been shown to deviate considerably, one of the latest examples being the financial crisis
We thank Redouan Bshary, Katharina Förster and the members of the Eco-Ethology group at the University of Neuchâtel for discussion, and Nichola Raihani and two anonymous referees for very thoughtful comments that helped to improve the manuscript.