The Watching Eyes Effect Further Examined Wieteke Vrouwe VU

Running Head: WATCHING EYES EXAMINED
1
Helping a Child: The Watching Eyes Effect Further Examined
Wieteke Vrouwe
VU University Amsterdam
Student number: 2035642
Supervisor: Manesi, Z.
Second Assessor: Balliet, D. P.
Master’s Thesis Social Psychology
December, 2014
2 WATCHING EYES EXAMINED
Helping a Child: The Watching Eyes Effect Further Examined
Wieteke Vrouwe
VU University Amsterdam
3 WATCHING EYES EXAMINED
Abstract
Several studies have shown that the presence of images of watching eyes (either photographs
or drawings) significantly increases prosocial behavior in both laboratory and real-world
settings. Researchers have been calling this phenomenon the watching eyes effect. In this
paper, a novel approach to the watching eyes effect was introduced. Volunteering behavior
was used as a behavioral measure for prosociality. Social Value Orientation (SVO) was used
as a moderating variable in the relationship between watching cues and volunteering.
Furthermore, two conditions of watching cues were specified based on gender to measure the
influence of social information in the effects of watching cues. One hundred and twenty-three
participants joined an automatically running experiment in which they could choose to
voluntarily correct essays of children with a language deficiency, after being primed with
watching cues. The choice to correct an essay and the choice to become a volunteer were
main measurements of volunteering behavior. The results of this experiment show a
significant, positive effect of watching cues on volunteering and a difference in effect
between masculine and feminine watching cues. Future research could consider social
information in watching cues and the costliness of prosocial behavior, as our results firmly
suggest differences in effect of watching cues based on gender, and a different effect of subtle
watching cues on high costly and low costly behavior.
Keywords: watching eyes effect, volunteering, SVO, gender
WATCHING EYES EXAMINED
4 Helping a Child: The Watching Eyes Effect Further Examined
Recently, many researchers have investigated the watching eyes effect on prosocial
behavior (Mifune, Hirofumi, & Yamagishi, 2010; Oda, Niwa, Honma, & Hiraishi, 2011;
Francey & Bergmuller, 2012). Prosocial behavior covers the broad range of actions intended
to benefit one or more people other than oneself—actions such as helping, comforting,
sharing, and cooperating (Batson & Powell, 2003). Several studies have shown that the
presence of images of watching eyes (either photographs or drawings of eyes) significantly
increases prosocial behavior in both laboratory and real-world settings (Bateson, Nettle, &
Roberts, 2006; Ernest-Jones, Nettle, & Bateson, 2011; Powell, Roberts & Nettle, 2012).
Researchers have been calling this effect the watching eyes effect. This research paper
introduces a novel approach to the watching eyes effect by investigating the relationship
between watching cues, volunteering as a behavioral measure of prosociality, and personality.
In previous lab studies, researchers have explored the watching eyes phenomenon
using behavioral economic games. Haley and Fessler, (2005), for example, measured
cooperation in a Dictator Game. Conditions with watching cues included stylized eye-like
shapes on a desktop background. It was found that the presence of eyespots leads to
significantly more generous behavior than the control condition, in which cues of abstract art
were presented. More laboratory studies have since shown that the watching eyes effect can
increase prosocial behavior of various types, including generosity (Mifune et al., 2010; Oda et
al., 2011; Nettle et al., 2013), public goods contributions (Burnham & Hare, 2007), and
condemnation of moral violations (Bourrat, Baumand, & McKay, 2011). However, despite
the large amount of studies showing a significant effect of watching cues on prosocial
behavior, in other lab studies no eyes effect was reported (e.g., Carbon and Hesslinger, 2011;
Fehr and Schneider, 2010; Lamba and Mace, 2010). These differences could possibly be
explained by the different measures used in studies (the probability of donating money versus
WATCHING EYES EXAMINED
5 the mean amount of money donated) (Nettle et al., 2013). Another explanation for dissimilar
results of earlier studies could be the procedural differences in, for example, the length of
exposure to eye images (Sparks & Barclay, 2013).
Other studies have furthermore focused on the effect of watching cues in real-world
settings. Social behavior in different types of situations, such as littering in a cafeteria (ErnestJones, Nettle, & Bateson, 2011), donating to charity in a supermarket (Powell, Roberts &
Nettle, 2012) or removing garbage at a bus stop (Francey & Bergmuller, 2012) was measured.
In all of these studies, watching cues were significantly linked to an increase of prosocial
behavior.
To sum up, according to previous studies, exposing people to images of eyes increases
multiple types of cooperation in various contexts. However, although that in some cases,
images of eyes seem to influence prosocial behavior (Bateson, Nettle, & Roberts, 2006;
Ernest-Jones, Nettle, & Bateson, 2011; Powell, Roberts & Nettle, 2012), in other cases no
effect is found (e.g., Carbon and Hesslinger, 2011; Fehr and Schneider, 2010; Lamba and
Mace, 2010). As these contradictory findings are tremendously interesting, we want to see if
the watching eyes effect occurs whilst working with volunteering behavior, a new behavioral
measure, as the dependent variable. Batson and Powell (2003) noted that volunteering is
mentioned in the range of acts that are called prosocial behavior. Therefore, and with regards
the results of previous studies, in Hypothesis 1 of this study we expect a significant positive
effect of watching cues on volunteering, just as it has on prosocial acts like cooperating,
donating and sharing.
Gender of watching cues
In earlier research on the watching eyes effect, specific social information of the cues
(e.g. emotion, age or gender) was not taken into account (Francey & Bergmuller, 2012;
Rigdon, Ishii, Watabe & Kitayama, 2008). To be able to define the effect of watching cues,
WATCHING EYES EXAMINED
6 we believe that it is necessary to take a closer look at these specific social aspects. A widely
used construct to define differences in research is gender. Bateson, Nettle and Roberts (2006)
found a trend in a watching eyes experiment that indicated masculine eye cues, compared to
feminine eye cues, led people to make higher contributions to an honesty box in a university
cafeteria. This suggests that masculine eye gaze may be particularly powerful in eliciting
prosocial behavior. However, Perrett and colleagues (1998) found that participants preferred
feminized to average shapes of a female face and also preferred feminized to average or
masculinized shapes of a male face. This suggests that female facial features are more
attractive to us, and might thus hold more of our attention. Given these findings, it is difficult
to hypothesize that watching cues of one sex will have more effect on an individuals prosocial
behavior than watching cues of the other sex. We therefore propose two conflicting
statements for Hypothesis 2. First, we indicate that masculine eye gaze could lead to more
volunteering behavior based on findings of Bateson, Nettle and Roberts (2006). Second, we
suggest that female facial features can lead to more volunteering behavior based on findings
of Perrett and colleagues (1998).
Social Value Orientation
Since several studies have not been able to replicate the watching eyes effect, it is
remarkable that knowledge about any moderating factors in this effect is scarce. It has been
suggested that reputational mechanisms have an influence on the watching eyes effect, as they
facilitate prosocial behavior. Reputational models can be explained on the basis of Social
Value Orientation (SVO). SVO is addressed as a concept that theoretically extends the
rational self-interest postulate by assuming that individuals tend to pursue broader goals than
only self-interest (De Cremer & Van Lange, 2011). The concept refers to preferences for
particular patterns of outcomes for the self and others and focuses on a three-category
typology of SVO: a prosocial orientation, an individualistic orientation and a competitive
WATCHING EYES EXAMINED
7 orientation, from which the latter two are referred to as a proself orientation (De Cremer &
Van Lange, 2001; Murphy et al., 2011).
Balliet, Parks and Joireman (2009) stated in their meta-analysis that SVO is related to
cooperation in social dilemmas. De Cremer and Van Lange (2001) indicated both cognitive
and behavioral differences between the prosocial and proself orientation. It was argued that
prosocials experience stronger feelings of social responsibility than proselfs, consequently
influencing their tendency to exhibit greater cooperation. Proselfs, on the other hand, are said
to be more affected by concerns about their own reputation. Since prosocial people have a
default willingness to act prosocially, this makes them intrinsically motivated and directly
affects their reputation, which thus has a high chance to be positive. On the other hand, selforiented people will need explicit incentives in order to be prosocial (Bogaert, Boone &
Declerck, 2008). Self-oriented people seem to be lead by the fear of having a negative
reputation, based on incentives like being in- or excluded as a member of a social group. As
their choices are made based on possible reputational damage, their reputation seems to be
altered by an indirect desire to meet perceived expectations of others (Sparks & Barklay,
2013; Mifune et al, 2010; Oda et al, 2011).
If we follow the suggestion that watching cues elicit reputational concerns, then it
could be that people who are susceptible to reputational concerns (e.g. proselfs) are more
likely to be affected by cues of being watched. As mentioned earlier, there are inconsistencies
in findings on the watching eyes effect. Because of the suggested link between watching cues
and reputational models, we want to discover if including SVO has an influence on the
watching eyes effect. In Hypothesis 3, we expect that SVO has a moderating effect on the
relationship between watching cues and prosocial behavior. As proselfs seem to be more
susceptible to reputational concerns, we expect the effect of watching eyes on prosocial
behavior to be greater for proselfs than for prosocials.
8 WATCHING EYES EXAMINED
To sum up, we believe that some important aspects of the watching eyes effect need
further investigation. Our research contributes to the in depth exploration of the watching eyes
effect in three different ways. In Hypothesis 1, we expect that there is a significant, positive
relationship between watching cues and volunteering behavior. In Hypothesis 2, we
investigate whether there are differences in the effects of male and female watching cues. As
noted earlier, based on the existing evidence, we are not able to make a prediction about the
direction of this difference. Still, since eyes are generally known as ‘the window to the soul’
and are said to reveal social information, we expect a difference between male and female
watching cues. In Hypothesis 3, we expect the relationship between watching cues and
volunteering to be moderated by a personality trait called Social Value Orientation (Murphy,
Ackermann, & Handgraaf, 2011). For this hypothesis, we argue that that the effect of
watching cues on prosocial behavior will be stronger for individuals with a proself orientation
than for individuals with a prosocial orientation.
Method
Participants & Design
One hundred and twenty-three undergraduate students were recruited via
announcements at the VU campus. Participants took part in a laboratory study for course
credit or monetary incentives. The sample consisted of 31 men (25.2%) and 92 women, with
ages ranging from 17 to 29 (M = 21 years, SD = 2.19 years). As we used a between-subjects
design, participants were randomly assigned to three conditions: male eyes (n = 41), female
eyes (n = 42) or abstract art (control) (n = 40).
Materials
Watching cues. As watching cues, 16 neutral images of male eyes and 16 neutral
images of female eyes, collected from the Radboud Faces Database (Lagner et al., 2010), and
16 images of abstract art were projected on a wall (see Appendix A). All images were
WATCHING EYES EXAMINED
9 displayed in gray scale and had an approximate width of 100cm (400x160px). Each
participant got to see the images belonging to their condition one by one for ten seconds.
Since the watching cue itself was manipulated as independent variable, a bogus task was used
to keep the focus on the cues. This task involved viewing the images and rating them on
emotion, familiarity with previous images or eye-color. Dependent on the choices of the
participant, watching cues were shown up to three times. Every time, another question
functioned as a bogus task.
Volunteering behavior. As our dependent variable is prosocial behavior, we used
volunteering as a measure for this variable. Participants were asked to correct a paper,
handwritten by a child with a language deficiency, to help it overcome this issue. This paper
had many syntax- and grammar errors (see Appendix C). Participants were not given any
incentives for their help in correcting essays, however, they were given a show-up fee for the
experiment as a whole. The two main measurements of prosocial behavior included the choice
to correct an extra paper (DV1) and the choice to sign up as a volunteer for the “Voorlees
Express” (DV2), an organization that helps young children with a language deficiency by
reading books with them. Since costly prosocial behavior is defined as any action aimed at
benefiting others that involves some cost to the agent, frequently in the form of a contribution
of time or money (Liu and Aaker 2008), the choice to correct a second essay can be seen as a
short-term choice, far less costly than the choice to sign up as a volunteer (and thus becoming
one on the long-term in real life). Furthermore, we used the amount of time spent correcting
(DV3) and the amount of errors marked (DV4) as extra measurements of volunteering
behavior. When a person chose to correct another essay or signed up to volunteer, their
behavior was marked as prosocial (see Appendix B). Furthermore, the higher the amount of
time spent correcting or errors marked, the more prosocial this behavior was marked to be.
WATCHING EYES EXAMINED
10 Questionnaires. Two questionnaires on social value orientation and a demographic
data sheet were used to get an indication of the personality and social status of participants.
The demographic data sheet asked for information such as gender, age and level of education,
but also dyslexia, political position and socio-economic status. The social orientation
questionnaires focused on SVO and Social Mindfulness.
SVO-Slider Measure. We used the SVO Slider Measure (Murphy et al., 2011) as an
indicator of the participants SVO. This measure has six primary items with nine secondary
(and optional) items. Each item is a resource allocation choice over a continuum of joint
payoffs. An example could be: “There’s a value x between 50 and 100. Your payoff will be x,
while for the person depending on you, the payoff will be 150-x. What value will you
choose?”. Based on the categorization scales of the SVO-Slider Measure, in this particular
sample N = 40 participants were categorized as proselfs, whilst N = 83 participants were
categorized as prosocials.
Social Mindfulness. We used the Social Mindfulness paradigm (SoMi) developed by
van Doesum et al. (2013) as an extra indicator for personality differences. This measure has
twelve social mindfulness items and twelve control items. Each item is a resource choice in
which the participant gets to choose first, while another person is said to choose second. The
paradigm is based on the assumption that social mindful thinking makes an individual choose
in a way that the other person still has a choice between different resources. A proportion of
socially mindful choices can be computed by averaging the scores over all decisions, resulting
in a number between 0 (only unmindful choices) and 1 (only mindful choices). The mean
score of 123 participants in Social Mindfulness was M = .708, with a standard deviation of SD
= .203, indicating that participants have a tendency to show a social mindful mindset.
WATCHING EYES EXAMINED
11 Procedure
To develop a complex, electronic measurement instrument for prosocial behavior, we
used open-source experiment builder Open Sesame. For this experiment, it was crucial that
the participant was alone in the room, as presence of other people could influence the
manipulation of watching cues on the participant. Through Open Sesame, we made an
automatically running experiment with multiple paths that also stored specific information
about the time spent and choices made directly in a personal Excel-sheet. Upon arrival at the
laboratory, participants were informed that they would take part in a study on language and
emotions. In the first part of the experiment, participants were asked to answer a question
about the emotion in every watching cue that was being projected on the wall. As the
watching cue itself was the independent variable, this question only functioned as a bogus
task. We used this task to make sure participants focused on the watching cue and were
sufficiently framed with this stimuli. In the second part of the experiment, participants were
informed that they had to correct a paper written by a child with a language deficiency, to
help it overcome this issue. This action was used to measure the construct of our dependent
variable prosocial behavior. After the second part, two more rounds with watching cues and
essays followed. Before each essay, the eye manipulation was shown again. If participants
wished so, they could continue on voluntarily correcting this extra second and third essay.
When a person chose to correct an extra essay, volunteering behavior was measured as
increasing. If they did not choose to do so, participants were directed immediately to the final
task. This final task was to fill in a survey with all questionnaires. Afterwards, participants
were debriefed about the purpose of the study and were also asked to give permission for their
observation.
To sum up, it was mandatory for participants to correct the first essay. This first essay
functioned as a baseline measure for scores on any further essays. However, the correction of
12 WATCHING EYES EXAMINED
two more essays was optional. Before the task of correcting an essay, participants were
always primed with watching cues.
Results
In this section, we will give an overview of the analyses performed on our gathered
data and the results of these analyses. As only 8 of 123 participants chose to correct a third
essay, there was not enough power to do analyses on the choice to correct a third essay.
Because of this, we chose to only analyze the choice to correct a second essay (see Figure 1).
Dependent Variable 1: The Choice to Correct an Extra Essay
A Binary Logistic Regression was performed to assess the impact of different
watching cues on the likelihood that respondents chose to correct a second essay. The full
model was statistically significant, 𝜒2 (2, N = 123) = 9.264, p = .010, indicating that the model
was able to distinguish between respondents who chose and didn’t choose to correct a second
essay. The model as a whole explained between 7.3% (Cox & Snell R Square) and 9.8%
(Nagelkerke R Square) of the variance in choice and correctly classified 63.4% of cases. Two
of the conditions made a unique statistically significant contribution to the model: female
watching cues (Wald = 8.522, p = .014) and abstract art cues (B = -1.428, Wald = 8.513, p =
.004). These results indicate that female eyes enhanced the possibility to correct an extra
essay, whilst, with an odds ratio of .24, abstract art cues decreased the possibility of
correcting another essay with 24%.
A One-Way Between Groups ANOVA with post-hoc tests was performed to evaluate
the differences between conditions on the amount of essays corrected. There was a
statistically significant difference for the three conditions: F(2,120) = 5.5, p = .005, eta
squared = .083, indicating a medium to large effect size. Post-hoc comparisons using the
Bonferroni test indicated that the mean amount of essays corrected for the condition with
female watching cues (M = .74, SD = .767) was significantly higher than for the condition
WATCHING EYES EXAMINED
13 with art (control) (M = .25, SD = .494). However the condition with male watching cues M =
.54, SD = .711) did not differ significantly from either the female- or art watching cues (see
Figure 1).
SVO. A second Binary Logistic Regression was performed to assess the impact of
SVO and watching cues on the likelihood that respondents chose to correct a second essay.
The full model had a significant effect on the choice to correct a second essay, 𝜒2 (5, n = 123)
= 37.542, p < .001, indicating that the model was able to distinguish between respondents
who did and did not chose to correct another essay. The model as a whole explained between
26.3% (Cox & Snell R Square) and 35.6% (Nagelkerke R Square) of the variance in choice.
Variable parameters indicated that SVO has a positive effect on the choice to correct a second
essay (Wald = 7.094, p = .008), whilst condition (male, Wald = 1.404, p = .236; control, Wald
= .000 p = .998) and the interaction of SVO and condition (male, B = 3.436, p = .064; control,
B = .000, p = .998) did not have any effect on the choice to correct an extra essay. The odds
ratio was 19.800 for SVO, indicating that for every raise in SVO, respondents were 19.8 times
more likely to correct a second essay. Furthermore, a striking 53% of the prosocials chose to
correct one or two extra essays whilst on the other hand, 89% of the proselfs stopped after the
first, mandatory essay.
Dependent Variable 2: The Choice to Sign up as a Volunteer
A Binary Logistic Regression was performed to explore the relationship between
watching cues and the likelihood that respondents chose to sign up as a volunteer. The full
model was not statistically significant, 𝜒 ! (2, N = 123) = 1.942, p = .379, indicating that it was
not able to distinguish between those who chose and didn’t choose to become a volunteer.
A second Binary Logistic Regression was conducted to find out whether SVO and watching
cues have an interaction effect on willingness to volunteer. The full model containing two
predictor was not significant, 𝜒2 (5, n = 123) = 3,278, p = .657, indicating that an interaction
WATCHING EYES EXAMINED
14 of SVO and watching cues does not explain willingness to volunteer. Furthermore, no
significant association was found for a main effect between watching cues and willingness to
volunteer, 𝜒2 (2, n = 123) = 1.919, p = .383, Cramer’s V = .125. Also, no significant
relationship was found for a main effect between SVO and willingness to volunteer, 𝜒2 (1, n =
123) = .367, p = .545, Cramer’s V = .055.
Dependent Variable 3: The Amount of Time Spent Correcting
To evaluate whether there is a relationship between watching cues and the amount of
time spent correcting an essay, and to see if SVO has a moderating effect on this relationship,
two two-way Analyses of Covariance (ANCOVA) were conducted. In these analyses,
watching cues (3 conditions) and SVO were used as independent variables. The amount of
time spent correcting was used as the dependent variable. Dyslexia and self-assessed skill in
Dutch language were used as covariates in these analyses.
Preliminary checks were conducted to ensure that there was no violation of the
assumptions of normality, linearity, homogeneity of variances, homogeneity of regression
slopes and reliable measurement of the covariates. In both analyses, no significant effects
were found, Time spent correcting Essay 1: F (2, 115) = .210, p = .811, partial eta squared =
.004; Time spent correcting Essay 2: F (1, 42) = .603, p = .442, partial eta squared = .014,
indicating that there is no relationship between watching cues, SVO and the time spent
correcting an essay.
Dependent Variable 4: The Amount of Errors Marked
For the fourth dependent variable, two two-way Analyses of Covariance (ANCOVA)
were conducted again. In these analyses, watching cues (3 conditions) and SVO were used as
independent variables. The amount of errors marked was used as the dependent variable.
Preliminary checks were conducted to ensure that there was no violation of the assumptions.
In both analyses, no significant effects were found, Errors marked in Essay 1: F (2, 115) =
15 WATCHING EYES EXAMINED
.042, p = .959, partial eta squared = .001; Errors marked in Essay 2: F (1, 42) = 1.705, p =
.199, partial eta squared = .039, indicating that there is no relationship between watching
cues, SVO and the amount of errors marked in an essay.
Discussion
With this experiment, we aimed to create a multidimensional view on a phenomenon
that has researchers in its grip for over ten years: the watching eyes effect. This phenomenon
is characterized by the significant increase in prosocial behavior in both laboratory and realworld settings due to the presence of photographs or drawings of watching eyes (Bateson,
Nettle, & Roberts, 2006; Ernest-Jones, Nettle, & Bateson, 2011; Powell, Roberts & Nettle,
2012). In this paper, we have introduced a new behavioral measure for the watching eyes
effect by focusing on volunteering behavior as the dependent variable.
For our first hypothesis, we expected watching cues to have a positive effect on
volunteering behavior. Indeed, watching cues had a significant effect on the choice to correct
a second essay voluntarily, just as it had on prosocial acts like cooperating (Mifune et al.,
2010 Oda et al., 2011), donating (Burnham & Hare, 2007) and sharing (Nettle et al., 2013).
However, in other behavior associated with volunteering, watching cues did not seem to have
any influence. Watching cues did not affect the amount of errors marked in essays, nor the
choice to sign up as a volunteer. Furthermore, watching cues did not effect the time that was
spent on correcting an essay. A possible explanation for not finding any effects in the
relationship between watching cues and these variables are differences in costliness between
measured volunteering behaviors. Costly prosocial behavior is defined as any action aimed at
benefiting others that involves some cost to the agent, frequently in the form of a contribution
of time or money (Liu and Aaker 2008). The choice to correct a second essay can be seen as a
short-term choice, far less costly than the choice to sign up as a volunteer (and thus becoming
one in the long term in real life). Differences between the costliness of those behaviors could
WATCHING EYES EXAMINED
16 have played a role in the occurrence of an effect, led by the possibility that high costly
behavior is not affected by relatively subtle watching cues.
For our second hypothesis, we investigated whether there are differences in effect
between male and female watching cues. Based on our own results, we could say that
feminine watching cues seem to affect prosocial behavior more than abstract art, or nonwatching cues. We found that participants that were in the condition with feminine watching
cues were significantly more likely to correct an extra essay, compared to the grand mean.
Furthermore, participants in the condition with abstract art cues were significantly less likely
to correct an extra essay, compared to the grand mean, suggesting again that eyes in general
increase the likelihood to correct an extra essay. It is possible that, since the effect seems to be
driven by female eyes, feminine watching cues are more effective because female facial
features are said to be more attractive (Perrett et al., 1998) than male facial features. This
could lead to an increased attention to feminine watching cues, which in turn could lead to an
increase in prosocial behavior.
In our third hypothesis, we stated that SVO could have a moderating effect on the
relationship between watching cues and prosocial behavior. It was found that SVO had an
effect on the choice to correct a second essay, however, this was a main effect. A striking
53% of prosocials chose to correct one or two extra essays, whilst 89% of proselfs only
completed the first, mandatory essay. This finding correspond to the construct of SVO; the
higher one’s Social Value Orientation is, the more likely one is to behave prosocially. As
trends do show us that SVO seems to influence the relationship between watching cues and
volunteering, but no moderating effects of SVO have further been found, unfortunately we
cannot draw any firm conclusions here.
To sum up, in this experiment, we have measured different kinds of altruistic
behavior, with one main variable that was high in costliness (the choice to become a volunteer
WATCHING EYES EXAMINED
17 in real life) and with one main variable that was low in costliness (the choice to correct an
extra essay). Besides that, we have expanded the range of stimuli, to define the role of social
information within the watching eyes effect. Lastly, we involved SVO to get a more sound
understanding of the causation of prosocial behavior. Through these choices, we have
examined the watching eyes effect on a new level, which in turn leads to the raise of many
new questions. We are, for example, highly interested in the effect of high costly versus low
costly behavior regarding the effect of watching cues on prosocial behavior, as our results
seem to indicate that subtle watching cues are not powerful enough to influence behavior that
requires personal resources over a longer period of time. Also, since there was not enough
power to test for the moderation of SVO on the relationship between watching cues and
volunteering, we are interested in the effects of this particular construct in future research on
the watching eyes effect.
Limitations, Strengths and Future Directions
Through this experiment, we came across some difficulties regarding the introduction
of volunteering as a behavioral measure in the watching eyes effect. Since the costliness of
the constructs we used to measure volunteering differed, our results suggest that watching
cues do have an effect on volunteering, but only under certain particular conditions. It seems
likely that subtle watching cues only influence behavior that is relatively low in costliness,
compared to behavior that is high in costliness in real life and on the long-term. Future studies
could thus consider the costliness of proposed prosocial behavior, to investigate this
suggestion. In addition, for the short-term choice to sign up as a volunteer for the “Voorlees
Express”, we were not able to check if participants really became a volunteer on the longterm, which could have led to meaningful insights on long-term behavior. Moreover, the
camera that was necessary to check on participants, as they were alone in the room, could
have led to noise in watching cues. This camera was, however, present in all conditions and
WATCHING EYES EXAMINED
18 any effects should thus be ruled out. Lastly, the sample size of the choice to correct a third
essay was too small to draw any conclusions about this variable, which forced us to analyze
only the choice to correct a second essay.
Setting these limitations aside, we have introduced a new level to the watching eyes
effect by expanding earlier research with involving volunteering as a new measure of
prosocial behavior, watching cues specified based on social information and personality trait
SVO. Furthermore, since behavior in social dilemma experiments seems to predict prosocial
behavior outside the laboratory very well (Van Lange, Bakkers, Schuyt & Van Vugt, 2007),
the external validity of this study needs to be mentioned. In addition to this, we are pleased to
have connected 69 participants (56.1% of the total pool) to the Voorlees Express volunteering
organization. Moreover, the trends in our study indicate that SVO has an influence on the
watching eyes effect. Since it is suggested that watching cues elicit reputational concerns, it
could be that people who are susceptible to reputational concerns (e.g. proselfs) are more
likely to be affected by cues of being watched. Future research could contribute to a better
implementation of watching cues in different kinds of real life situations by further examining
the influence of SVO on the relationship between watching cues and prosocial behavior.
To wrap up, we would like to recommend future researchers to consider the costliness
of prosocial behavior in regards to the watching eyes effect, since our results indicate that
high costly behavior might not be influenced by subtle watching cues. Furthermore, research
could focus on the effects of gender in watching cues, as we found that the watching eyes
effect seems to be driven by female watching cues. We are also interested in the examination
of other social information in watching cues, as we believe that more personal traits could
play a role in the effect of these cues. Lastly, since this experiment gave us information about
a trend in SVO regarding the watching eyes effect, we would recommend other researchers to
WATCHING EYES EXAMINED
19 further explore the effects of SVO and reputational mechanisms in the interesting
phenomenon that is called the watching eyes effect.
Acknowledgments
We express our warm thanks to Zoi Manesi at VU University for her support and
guidance in improving the manuscript. We also thank software developer Jarik den Hartog at
VU University for his help with the creation of our automatically running experiment.
20 WATCHING EYES EXAMINED
References
Balliet, D., Parks, C. & Joireman, J. (2009). Social Value Orientation and Cooperation in
Social Dilemmas: A Meta-Analysis, Group Processes & Intergroup Relations, 12(4),
533-547.
Balliet, D., Li, N. P., Macfarlan, S. J. & Van Vugt, M. (2011). Sex Differences in
Cooperation: A Meta-Analytic Review of Social Dilemmas, Psychological Bulletin,
137(6), 881-909.
Batson, C. D. & Powell, A. A. (2003). Altruism and Prosocial Behavior. Handbook of
Psychology, 3, 463–484.
Bateson, M., Nettle, D. & Roberts, G. (2006). Cues of being watched enhance cooperation
in a real-world setting, Biology Letters, 2, 412-414.
Bogaert, S., Boone, C. & Declerck, C. (2008). Social value orientation and cooperation in
social dilemmas: A review and conceptual model, British Journal of Social
Psychology, 47, 453-480.
Bourrat, P., Baumard, N. & McKay, R. (2011). Surveillance cues enhance moral
condemnation, Evolutionary Psychology, 9(2), 193-199.
Burnham, T. C. & Hare, B. (2007). Engineering Human Cooperation, Human Nature, 18(2),
88-108.
Carbon, C. C. & Hesslinger, V.M. (2011). Bateson et al.’s (2006) cues-of-being-watched
paradigm revisited, Swiss Journal of Psychology, 70, 203-210.
De Cremer, D. & Van Lange, P. A. M. (2001). Prosocials Exhibit Greater Cooperation than
Proselfs: The Roles of Social Responsibility and Reciprocity, European Journal of
Personality, 15, 5-18.
Eagly, A. H. (2009). The his and hers of prosocial behavior: An examination of the social
psychology of gender, American Psychologist, 64, 644–658.
WATCHING EYES EXAMINED
21 Ernest-Jones, M., Nettle., D. & Bateson, M. (2011). Effects of eye images on everyday
cooperative behavior: a field experiment, Evolution and Human Behavior, 32, 172178.
Fehr, E. & Schneider, F. (2010). Eyes are on us, but nobody cares: Are eye cues relevant for
strong reciprocity? Biological Sciences, 277, 1315-1323.
Francey, D. & Bergmuller, R. (2012). Images of Eyes Enhance Investments in a Real-Life
Public Good, PLoS ONE, 7 (5), 1-7.
Haley, K. J. & Fessler, D. M. T. (2005). Nobody’s watching? Subtle cues affect generosity in
an anonymous economic game, Evolution and Human Behavior, 26, 245-256.
Lamba, S. & Mace, R. (2010). People recognise when they are really anonymous in an
economic game, Evolution and Human Behavior, 31, 271-278.
Langner, O., Dotsch, R., Bijlstra, G., Wigboldus, D. H. J., Hawk, S. T., & van Knippenberg,
A. (2010). Presentation and validation of the Radboud Faces Database, Cognition &
Emotion, 24(8), 1377—1388.
Liu, W. & Aaker, J. (2008). The happiness of giving: The time-ask effect, Journal of
Consumer Research, 35(3), 543–557.
Mifune, N., Hirofumi, H. & Yamagishi, T. (2010). Altruism toward in-group members as a
reputation mechanism, Evolution and Human Behavior, 31, 109-117.
Murphy, R. O., Ackermann, K. A. & Handgraaf, M. J. J. (2011). Measuring Social Value
Orientation, Judgment and Decision Making, 6 (8), 772-781.
Nettle, D., Harper, Z., Kidson, A., Stone, R., Penton-Voak, I. S. & Bateson, M. (2013). The
watching eyes effect in the Dictator Game: it’s not how much you give, it’s being seen
to give something, Evolution and Human Behavior, 34(1), 35-40.
Oda, R., Niwa, Y., Honma, A. & Hiraishi, K. (2011). An eye-like painting enhances the
expectation of a good reputation, Evolution and Human Behavior, 32, 166-171.
WATCHING EYES EXAMINED
22 Ohlsen, G., Van Zoest, W. & Van Vugt, M. (2013). Gender and Facial Dominance in Gaze
Cuing: Emotional Context Matters in the Eyes That We Follow, PLoS ONE, 8(4),
e59471.
Perrett, D. I., Lee, K. J., Penton-Voak, I., Rowland, D., Yoshikawa, S., Burt, D. M., Henzi, S.
P., Castles, D. L., Akamatsu, S. (1998). Effects of sexual dimorphism on facial
attractiveness, Nature, 394(6696), 884-887.
Powell, K. L., Roberts, G. & Nettle, D. (2012). Eye Images Increase Charitable Donations:
Evidence From an Opportunistic Field Experiment in a Supermarket, Ethology,
118, 1-6.
Rigdon, M., Ishii, K., Watabe, M. & Kitayama, S. (2009). Minimal Social Cues in the
Dictator Game, Journal of Economic Psychology, 30(3), 358-367.
Sparks, A. & Barklay, P. (2013). Eye images increase generosity, but not for long: the
limited effect of a false cue, Evolution and Human Behavior, 34 (5), 317-322.
Sylwester, K. & Roberts, G. (2010). Cooperators benefit through reputation-based partner
choice in economic games, Biology Letters, 6(5), 659-662.
23 WATCHING EYES EXAMINED
Figures Figure 1. Frequencies of Essays Corrected by Different Watching Cue Conditions
24 WATCHING EYES EXAMINED
Appendix A Examples of Faces from the Radboud Faces Database (Lagner et al., 2010) Figure 1. Feminine Watching Cues Figure 2. Masculine Watching Cues 25 WATCHING EYES EXAMINED
Appendix B Example of Scores on Dependent Variables E1 Error Score % Time E1 s E2 Error Score % Time E2 s E3 Error Score % Time E3 s 0,84 413,291 0,67 153,389 -­‐ -­‐ 0,72 363,036 -­‐ -­‐ -­‐ -­‐ 0,72 486,308 -­‐ -­‐ -­‐ -­‐ 0,56 431,985 0,458333 301,368 -­‐ -­‐ 0,72 407,803 0,708333 374,605 -­‐ -­‐ 0,8 657,951 -­‐ -­‐ -­‐ -­‐ 0,84 638,343 0,791667 310,513 0,761905 307,123 0,88 806,051 -­‐ -­‐ -­‐ -­‐ 0,72 34,934 -­‐ -­‐ -­‐ -­‐ Volunteer Amount_Essays SoMi_Average SVO SelfSocial 1 1 0,583333 3 2 1 0 0,916667 3 2 1 0 1 3 2 0 1 0,666667 3 2 1 1 1 3 2 0 0 0,833333 2 1 1 2 0,833333 3 2 1 0 0,916667 3 2 0 0 0,833333 3 2 26 WATCHING EYES EXAMINED
Appendix C Example of a Handwritten Essay with Syntax-­‐ and Grammar Errors in Dutch 27 WATCHING EYES EXAMINED
Example of the Essay with Highlighted Syntax-­‐ and Grammar Errors Mijn hobby Ik hou heel erg van Voetbal. Uit school ga ik meteen naar het velt. Bij de club hebben we jongen en meisjesteams. Gelukkig ook aparte douchjes. Ik train 4x in de week en dan ook nog wedstrijd in de weekend. Dat is wel veel maar ik vindt toch andere dingen niet zo leuk als dit. Soms ben ik wel kapot hoor vooral als we krachtraining doen. maar dat is natuurlijk ook goed. Deze zeisoen staan we al best lang 2de, maar ik weet zeker dat we kampioen kunnen wordde!! Mijn grote voorbeeld is Zlatan ibrahimovic. Ook hij heeft bij Ajax gespeelt. Zlatan is beter als alle andere spelers omdat hij ook heel ardig is. Hij heeft ook een boek gemaakt. Ajax is de beste club. Jammer dat ik niet altijd de wedstrijden kan zien. Maar zelf voetballen is nog meer belangrijker. Ik hoop dat ik later ook nog voor Ajax kan spelen. Dan zou me leven echt compleet zijn. Maar ik weet niet of dat ooit gebeurt. Ik vind voetbal ook leuk omdat veel die meisjes het cool vinden. Soms doe ik daarom express me shirt uit als ze komen kijken. Maar eigelijk mag dat niet van de coach. Dit was me verhaal over voetbal.