The Monkey Brain
Psychology research - domestic surveillance and cheating

This was one of my two senior psychology research projects at Swarthmore, which I conducted under the guidance of Prof. Barry Schwartz.

People behave better when they think they are being watched. This effect comes up even with posters of eyes on the wall, and when people think about god. The question is whether this effect extends to the sensation of being watched under a regime of mass domestic surveillance.

Participants on Mechanical Turk were asked to perform a reading comprehension task. They were assigned into one of three conditions: an article excerpting from The Guardian about Edward Snowden and the NSA, an article about a flood in Indonesia, and the same flood article with a pair of watchful eyes in the background image.

Then they were asked to complete a task that the participants were told was for distraction, but in reality was the matrix test developed by Dan Ariely et al. to measure cheating. Participants were incentivized to cheat by being told that those who found the most pairs of numbers that added up to ten would be entered into a lottery to receive an award.

No significant results were found, including for the control group with the glasses. That makes me suspect that the electronic format of the experiment did not test what we expected. I’d like to rerun this experiment in a lab, after doing more thorough pre-testing with the stimulus.


Can being primed to think about the National Security Agency (NSA) and its widespread domestic surveillance programs lead to changes in cheating behavior? The experiment lacked a large enough sample size to tell conclusively, but the data is not suggestive.


Self-consciousness has been shown to cause pro-social behavior. Examples include hanging up a poster with eyes on it (Bateson, Nettle, & Roberts, 2006) or being primed to think about God (Shariff & Norenzayan, 2007). In these experiments, participants were led to donate more money into a collective pot for drinks, act more generously in anonymous economic games, or littering behavior in a cafeteria (Ernest-Jones, Nettle, & Bateson, 2011). The question is whether thinking about widespread, domestic surveillance as being conducted by the NSA causes the same type of self-consciousness as having a poster of eyes, or thoughts of God.


The study was presented as a reading comprehension exercise. Participants were told they would have to read a current events article, participate in a basic arithmetic contest for a cash prize, and then to answer reading comprehension questions about the article. To prime participants with the NSA, an excerpt from a news article was presented as one of the conditions: “The National Security Agency whistleblower, Edward Snowden, has said that the mass surveillance programmes used by the US to tap into phone and internet connections around the world are making people less safe. In short video clips posted by the WikiLeaks website on Friday, Snowden said that the NSA’s mass surveillance, which he disclosed before fleeing to Russia, “puts us at risk of coming into conflict with our own government”. A US court has charged Snowden with violating the Espionage Act, for disclosing the programmes which he described as a “dragnet mass surveillance that puts entire populations under sort of an eye that sees everything even when it’s not needed”. “They hurt our economy. They hurt our country. They limit our ability to speak and think and live and be creative, to have relationships and to associate freely,” Snowden said.”

Participants were randomly assigned into either receiving the NSA article, a control article about a typhoon, or into the “eyes condition”, in which they read the control article and performed the math exercises, with a pair of eyes in the background (Figure 1).

Then participants were assigned into one of two conditions for the “arithmetic challenge”. One condition was setup such that the problems facilitated cheating behavior (the cheating condition), and the other half did not (the honesty condition). This is the “matrix task” created by Mazar, Amir, & Ariely (2008). Participants were shown a page with twenty matrices of numbers, each three by four (Figure 2). Most of the matrices had pairs of numbers that added up to ten. The participant was told that the top ten people who found the most pairs of numbers that added up to ten within a four minute time limit would be entered into a lottery to win a prize. In the cheating condition, participants were told to check a box under each matrix for which they found a pair. In the honesty condition, participants were required to fill in two text boxes underneath each matrix with the numbers that added up to ten.

After the matrix task, participants were not given a reading comprehension test, as they had presumed, and were debriefed about the nature of the experiment, and that the lottery was a fake.


Forty-three participants (M=25, F=18) were recruited on Amazon Mechanical Turk across two different days. They received $0.50 for completing the experiment. A participant in the honesty condition was given “0” for creatively filling in numbers that added up to ten into the text boxes that were not a part of the matrices.

The matrix paradigm was effective in inducing cheating. A paired t-test between the cheating (N=26, mean ± SD, 4.92 ± 3.39) and honesty (N=17, 3 ± 2.45) conditions revealed a significant difference in the number of matrices reported as being answered between groups: t(37)=2.09, p<0.05.

However, the null hypothesis could not be rejected. The NSA-cheating condition (N=4, 4.5 ± 1.91) did not statistically differ from the control-cheating (N=11, 5.18 ± 4.29) condition: t(11.97)=-0.424, p=0.68. Moreover, the eyes were ineffective in reducing cheating. The eyes-cheating condition (N=12, 4.82 ± 3.03) did not differ from the control-cheating condition: t(18)=-0.23, p=0.82.


The failure of this experiment was in allowing for too many conditions. The honesty condition needed to be much smaller than it was in order to verify the validity of the experiment. Data collected for it could not contribute to testing the hypothesis at all, leading to extremely small sample sizes. The NSA-lying condition was central to the experiment, and yet only received four participants. However, the data is not even remotely suggestive that this is an effective way of priming prosocial behavior. Future research should focus on finding a more effective way to prime NSA surveillance, and alternative ways to measure cheating behavior.


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

Ernest-Jones, M., Nettle, D., & Bateson, M. (2011). Effects of eye images on everyday cooperative behavior: a field experiment. Evolution and Human Behavior, 32(3), 172–178. doi:10.1016/j.evolhumbehav.2010.10.006

Mazar, N., Amir, O., & Ariely, D. (2008). The Dishonesty of Honest People: A Theory of Self-Concept Maintenance (SSRN Scholarly Paper No. ID 979648). Rochester, NY: Social Science Research Network. Retrieved from

Shariff, A. F., & Norenzayan, A. (2007). God Is Watching You Priming God Concepts Increases Prosocial Behavior in an Anonymous Economic Game. Psychological Science, 18(9), 803–809. doi:10.1111/j.1467-9280.2007.01983.x


More posts

The Monkey Brain is Sam Zhang's blog of weird datasets and mathematical curiosities.

You can try to read this through an Atom-compatible feed reader, but the posts often rely on embedded HTML and Javascript. For the adventurous, here is the feed.xml.

the zoo