Failure to replicate increasing generosity by eyes (#198)

Return to View Chart

How to Cite this Report

APA Style

Jacob Jolij Tineke de Haan. Failure to replicate increasing generosity by eyes. (2014, July 28). Retrieved 22:44, February 22, 2018 from

MLA Style

"Failure to replicate increasing generosity by eyes" Jacob Jolij Tineke de Haan. 28 Jul 2014 06:01 22 Feb 2018, 22:44 <>

MHRA Style

'Failure to replicate increasing generosity by eyes', Jacob Jolij Tineke de Haan, , 28 July 2014 06:01 <> [accessed 22 February 2018]

Chicago Style

"Failure to replicate increasing generosity by eyes", Jacob Jolij Tineke de Haan, , (accessed February 22, 2018)


Failure to replicate increasing generosity by eyes [Internet]. Jacob Jolij Tineke de Haan; 2014 Jul 28, 06:01 [cited 2018 Feb 22]. Available from:

Reference to Original Report of Finding Haley, K., and Fessler, D. (2005). Nobody's watching? Subtle cues affect generosity in an anonymous economic game. Evolution and Human Behavior, 26, 245-256.
Title Failure to replicate increasing generosity by eyes
If the original article contained multiple experiments, which one did you attempt to replicate? e.g., you might respond 'Study 1' or 'Experiment 4'. Study 1
Link to PDF of Original Report
Brief Statement of Original Result In a dictator game, participants give higher offers to their partner when a pair of eyes is present in the background of a computer screen then when the screen is neutral.
Type of Replication Attempted Conceptual Replication
Result Type Failure to Replicate
Difference? Same Direction, 0.282
Number of Subjects 118
Number of Subjects in Original Study 49
Year in which Replication Attempt was Made 2014
Name of Investigators (Real Names Required) Jacob Jolij Tineke de Haan
Detailed Description of Method/Results Ethics statement
This study was approved by the local Ethics Committee ("Ethische Commissie Psychologie van het Heymans Instituut voor Psychologisch Onderzoek") and carried out according to the Declaration of Helsinki.

118 first-year students (92 women, mean age 21.6 years, SD 3.9 years) gave their written informed consent to participate in this study. Participants were rewarded with study credit for their participation; experiment participation is a required part of the psychology curriculum in Groningen.

Stimuli and procedure
Participants were tested in a computer laboratory with 8 cubicles, i.e. 8 participants per session. After receiving general instructions from the experimenter, participants were shown on the computer screen that they were paired with a participant from another session in a dictator game, and that they could earn up to 5 additional study credits for their participation. These five additional credits would be distributed between the two participants by one of them (the 'dealer'), and this distribution would be final (i.e., the 'dealer' is in full control). After these instructions, participants got to see whether they would be the recipient or the dealer. Obviously, all participants were assigned the role of 'dealer'.
After receiving their role instructions, participants were given 20 seconds to think about their choice. Critically, during these 20 seconds a countdown timer appeared. Centrally, time to decision (in s) was presented, flanked by the prime images, which could be masked or unmasked. Masking was done using a metacontrast mask. We have chosen eyes as prosocial primes, and flowers as a neutral control prime. In the masked conditions, the primes were presented for 10 ms, directly followed by a mask for 150 ms, preceded by a blank interval for 840 ms during which only the counter was presented. The unmasked stimuli were shown for 110 ms, followed by the mask for 150 ms, preceded by a blank interval with only the counter of 740 ms. Masking parameters were chosen on basis of a small pilot experiment (N=6). In this pilot, none of participants reported awareness of the masked stimuli, but yet they reported to have seen all unmasked stimuli. See figure 1 for a schematic overview of a single trial, and an example of the stimuli used. After the countdown interval, participants had to indicate how many credits they wanted to give to the recipient player, ranging from 0 to 5 credits, in whole credits. Finally, an exit questionnaire was administered, asking whether participants had seen the primes, and whether they had seen through the manipulation. Please note that participants only participated in one single condition.

Data were analyzed in SPSS version 20 (IBM, Natick, USA). To assess effect of our manipulations, we used a 2x2 univariate anova with the number of credits given to the recipient as dependent measure, and prime (flowers or eyes) and level of masking (masked or unmasked) as factors. The number of research participation credits students still needed to fulfill the requirements of their was taken into account as covariate. We expect participants to give more credits to the 'recipient' when primed with unmasked eyes than in the other three conditions, and expected a moderating effect of the number of credits still required (with more credits still required resulting in smaller offers to the recipient).
To assess evidence in favor of the null hypothesis that eye primes do not increase generosity compared to flower primes, we computed Bayes factors for the conscious and unconscious conditions, using Zoltan Dienes’ web applet ( Since we had no explicit expectation about the size of an eventual effect, we used a flat distribution ranging from 0 to 5 as a prior.

Any Known Methodological Differences
(between original and present study)?
We have tried to replicate and extend the finding that priming participants with a cue of being watched, in this case a pair of eyes, makes them behave more prosocially, and whether this priming effect is contingent on awareness of the primes. However, in this study we have failed to find any evidence of priming at all – rather, we find substantial evidence for the null hypothesis. How can we explain these results? Of course, there are some notable differences between our study and earlier studies on priming prosocial behavior using eye primes [21, 22]. Bateson, Nettle and Roberts [22] present a quasi-experimental design in which they report increased deposits in a department honesty box in weeks in which they put up a picture of eyes compared to weeks in which they put up a picture of flowers. Although there is no reason at all to doubt the validity of this dataset, the quasi-experimental design of authors makes it virtually impossible to evaluate the size and magnitude of the priming effect. We do not know how many 'participants' there were, nor do we know the characteristics of these participants. Moreover, although Bateson, Nettle and Roberts have quite cleverly controlled for consumption by looking at overall milk consumption (tea is typically consumed with milk in the UK), there may have been many factors uncontrolled for that may have contributed to the observed effects, such as participants seeing through the manipulation, participants paying for a full semester in one week or even participants not taking milk with their consumption. We therefore acknowledge the work by Bateson, Nettle, and Roberts as an interesting real-life demonstration of the eye-priming-effect, but given the many caveats we will not consider it further here. Our study shares more features with the laboratory study of Haley and Fessler [21], who compared the effect of several prime types (cues of being watched versus neutral primes in auditory and visual modalities) on offers made in a dictator game. Since we modelled our experiments after this study, we tried to keep most features in line with those of Haley and Fessler - sample size (N ~ 30 per group), paradigm (dictator game) and decision time (20 s) were comparable. Notable deviations, though, were 'currency' (real money in [21], study credits in our study), setting (in a large computer testing room in [21], in cubicles in our study), and, most importantly, prime presentation (on the background of the computer desktop in [21], centrally flashed during countdown in our study). First, we would like to note that our sample size is equivalent to that of Haley and Fessler [21], which does not comply with Simonsohn's advice to use a sample size of 2.5 times that of the original study in direct replications [31]. As a result, our failure to find an effect cannot be interpreted as evidence that the effect we sought to replicate does not exist. However, it should be noted that our study is not a direct replication attempt, but rather a conceptual replication. Moreover, if the lack of an effect in our study could have been attributed to lack of power, our Bayes factor analysis would have revealed ‘inconclusive’ evidence for the null, rather than the substantial evidence for the null hypothesis we find in our dataset. What we may conclude, therefore, is that the effect of eye primes on prosocial behavior does not generalize from the settings used by Haley and Fessler [21] and Bateson, Nettles, and Roberts [22]. Our failure to find an effect is informative about the actual overall effect size, and the conditions in which it occurs [32]. One explanation for the observed null result is that in our experiment there was a moderating variable, eradicating the effect specifically in the present study. Could any of the factors in which our study deviated from Haley and Fessler’s [21] have caused the lack of an effect we observe? First, there is little reason to assume that the deviation in 'currency' might have resulted in a total lack of effect: even with completely virtual 'currencies' (that is, tokens that hold no real-world value) strong effects have been reported with social interaction paradigms as we used here [33]. Our 'currency', despite having no direct economic value, did have real world value to the participants, and indeed we observed that participants choose to keep most credits to themselves, rather than offering an equal split. Second, the presentation of the primes differed. In our unmasked condition, primes were presented centrally and were clearly visible to the participants. In the original study by Haley and Fessler [21], the primes were part of a logo and 'hidden' (but easily visible) on the computer screen. In both studies, participants have seen the primes, but is seems quite obvious that in our study the primes have drawn more attention than the primes used by Haley and Fessler, even though none of our participants was explicitly aware of the function of the eye primes.
Email of Investigator
Name of individuals who
actually carried out the project
Tineke de Haan ran subjects; Jacob Jolij and Tineke de Haan analyzed data
Location of ProjectMunting building, University of Groningen
Characteristics of Subjects
(subject pool, paid, etc.)
University students from subject pool
Where did these subjects reside?Netherlands
Was this a Class Project?No
Further Details of Results as pdf
Additional Comments
Email of Original Investigator
Quantitive Information
I have complied with ethical standards for experimentation on human beings and, if necessary, have obtained appropriate permission from an Institutional Review Board or other oversight group.
TAG: Attention TAG: JDM TAG: Language TAG: Learning TAG: Memory TAG: Perception TAG: Performance TAG: Problem Solving TAG: Social Cognition TAG: Social Psychology TAG: Thinking

Are you posting an unpublished replication attempt that you conducted yourself, or noting a published replication attempt?

Post Unpublished
Post Published