Why I Let My Students Cheat On Their Exam

Teaching People Game Theory Is Good. Making Them Live It Is Even Better.

On test day for my Behavioral Ecology class at UCLA, I walked into the classroom bearing an impossibly difficult exam. Rather than being neatly arranged in alternate rows with pen or pencil in hand, my students sat in one tight group, with notes and books and laptops open and available. They were poised to share each other’s thoughts and to copy the best answers. As I distributed the tests, the students began to talk and write. All of this would normally be called cheating. But it was completely OK by me.

Who in their right mind would condone and encourage cheating among UCLA juniors and seniors? Perhaps someone with the idea that concepts in animal behavior can be taught by making their students live those concepts.

Animals and their behavior have been my passions since my Kentucky boyhood, and I strive to nurture this love for nature in my students. Who isn’t amazed and entertained by videos of crafty animals, like Betty the tool-making crow, bending wires into hooks to retrieve baskets containing delicious mealworms? (And then hiding her rewards from a lummox of a mate who never works, but is all too good at purloining the hard-won rewards of others?)

Nevertheless, I’m a realist. Almost none of my students will go on to be “me”—a university professor who makes a living observing animals. The vast majority take my classes as a prelude to medical, dental, pharmacy, or veterinary school. Still, I want my students to walk away with something more than, “Animals are cool.” I want them to leave my class thinking like behavioral ecologists.

Much of evolution and natural selection can be summarized in three short words: “Life is games.” In any game, the object is to win—be that defined as leaving the most genes in the next generation, getting the best grade on a midterm, or successfully inculcating critical thinking into your students. An entire field of study, Game Theory, is devoted to mathematically describing the games that nature plays. Games can determine why ant colonies do what they do, how viruses evolve to exploit hosts, or how human societies organize and function.

So last quarter I had an intriguing thought while preparing my Game Theory lectures. Tests are really just measures of how the Education Game is proceeding. Professors test to measure their success at teaching, and students take tests in order to get a good grade. Might these goals be maximized simultaneously? What if I let the students write their own rules for the test-taking game? Allow them to do everything we would normally call cheating?

A week before the test, I told my class that the Game Theory exam would be insanely hard—far harder than any that had established my rep as a hard prof. But as recompense, for this one time only, students could cheat. They could bring and use anything or anyone they liked, including animal behavior experts. (Richard Dawkins in town? Bring him!) They could surf the Web. They could talk to each other or call friends who’d taken the course before. They could offer me bribes. (I wouldn’t take them, but neither would I report it to the dean.) Only violations of state or federal criminal law such as kidnapping my dog, blackmail, or threats of violence were out of bounds.

Gasps filled the room. The students sputtered. They fretted. This must be a joke. I couldn’t possibly mean it. What, they asked, is the catch?

“None,” I replied. “You are UCLA students. The brightest of the bright. Let’s see what you can accomplish when you have no restrictions and the only thing that matters is getting the best answer possible.”

Once the shock wore off, they got sophisticated. In discussion section, they speculated, organized, and plotted. What would be the test’s payoff matrix? Would cooperation be rewarded or counter-productive? Would a large group work better, or smaller subgroups with specified tasks? What about “scroungers” who didn’t study but were planning to parasitize everyone else’s hard work? How much reciprocity would be demanded in order to share benefits? Was the test going to play out like a dog-eat-dog Hunger Games? In short, the students spent the entire week living Game Theory. It transformed a class where many did not even speak to each other into a coherent whole focused on a single task—beating their crazy professor’s nefarious scheme.

On the day of the hour-long test they faced a single question: “If evolution through natural selection is a game, what are the players, teams, rules, objectives, and outcomes?” One student immediately ran to the chalkboard, and she began to organize the outputs for each question section. The class divided tasks. They debated. They worked on hypotheses. Weak ones were rejected, promising ones were developed. Supportive evidence was added. A schedule was established for writing the consensus answers. (I remained in the room, hoping someone would ask me for my answers, because I had several enigmatic clues to divulge. But nobody thought that far afield!) As the test progressed, the majority (whom I shall call the “Mob”) decided to share one set of answers. Individuals within the Mob took turns writing paragraphs, and they all signed an author sheet to share the common grade. Three out of the 27 students opted out (I’ll call them the “Lone Wolves”). Although the Wolves listened and contributed to discussions, they preferred their individual variants over the Mob’s joint answer.

In the end, the students learned what social insects like ants and termites have known for hundreds of millions of years. To win at some games, cooperation is better than competition. Unity that arises through a diversity of opinion is stronger than any solitary competitor.

But did the students themselves realize this? To see, I presented the class with one last evil wrinkle two days later, after the test was graded but not yet returned. They had a choice, I said. Option A: They could get the test back and have it count toward their final grade. Option B: I would—sight unseen—shred the entire test. Poof, the grade would disappear as if it had never happened. But Option B meant they would never see their results; they would never know if their answers were correct.

“Oh, my, can we think about this for a couple of days?” they begged. No, I answered. More heated discussion followed. It was soon apparent that everyone had felt good about the process and their overall answers. The students unanimously chose to keep the test. Once again, the unity that arose through a diversity of opinion was right. The shared grade for the Mob was 20 percent higher than the averages on my previous, more normal, midterms. Among the Lone Wolves, one scored higher than the Mob, one about the same, and one scored lower.

Is the take-home message, then, that cheating is good? Well … no. Although by conventional test-taking rules, the students were cheating, they actually weren’t in this case. Instead, they were changing their goal in the Education Game from “Get a higher grade than my classmates” to “Get to the best answer.” This also required them to make new rules for test-taking. Obviously, when you make the rules there is no reason to cheat. Furthermore, being the rule-makers let students behave in a way that makes us a quintessentially unique species. We recognize when we are in a game, and more so than just playing along, we always try to bend the rules to our advantage.

Morally, of course, games can be tricky. Theory predicts that outcomes are often not to the betterment of the group or society. Nevertheless, this case had an interesting result. When the students got carte blanche to set the rules, altruism and cooperation won the day. How unlike a “normal” test where all students are solitary competitors, and teachers guard against any cheating! What my class showed was a very “human” trait: the ability to align what is “good for me” with what is “good for all” within the evolutionary games of our choosing.

In the end, the students achieved their goal: They earned an excellent grade. I also achieved my goal: I got them to spend a week thinking like behavioral ecologists. As a group they learned Game Theory better than any of my previous classes. In educational lingo, “flipping the classroom” means students are expected to prepare to come to class not for a lecture, but for a question-and-answer discussion. What I did was “flip the test.” Students were given all the intellectual tools beforehand and then, for an hour, they had to use them to generate well-reasoned answers to difficult questions.

The best tests will not only find out what students know but also stimulate thinking in novel ways. This is much more than regurgitating memorized facts. The test itself becomes a learning experience—where the very act of taking it leads to a deeper understanding of the subject.

Peter Nonacs is a professor in the Ecology and Evolutionary Biology Department at UCLA. He studies the evolution of social behavior across species, ranging from viruses, to insects, to mammals and even occasionally humans.
Primary Editor: Kathryn Bowers. Secondary Editor: T.A. Frank.
*Photo courtesy of mrfishersclass.
Explore Related Content
, , ,
  • This is not “cheating” because the students are working within the set rules of the game. The professor set the rules that any lawful act is open for use on this exam and the students abide by them. While this is an interesting read on group dynamics and cooperation, it is a poor choice to label this a study of “cheating.”

    • Fuxy

      No but it makes for a lot of clicks on the article.

  • Nick Thompson

    Dear Thom Hartmann,

    Is this THE Thom Hartmann?

    Anyway, years ago, Craig and Muir did a study of chicken productivity. Instead of breeding chickens for individual productivity and then shoving them in crates of 9 for commercial production, which produces savagely aggressive chickens that have to be debeaked, they “breed” them in crates of 9. Peak individual productivity was reduced but productivity of the crates — which is, after all, what matters to the production poultryman — was increased and aggression was reduced to the point where debeaking was unnecessary. (This is all laid out in Sober and Wilson’s excellent book, Unto Others.)

    What the professor was doing here was performing the chicken experiment on students.: i.e., shifting the level of selection in the class room to the class level from the individual level. Human beings are remarkable for the ease with which they respond to changes in selection from one level to another because of a long evolutionary history which included bottlenecks in which properties of group structure must have been extremely important to individual survival.

    The big educational question is, of course, which sort of selection regimen best prepares students for the world they are actually entering? In the regimen of academic Reaganism under which we are living, we are producing students that will need to be debeaked if their workplaces are going to be productive.

  • David McFarland

    So, which type of exam will you give in future offerings of the course? Other courses?

    • Peter Nonacs

      I will likely do this again, but I’m next slated to teach a much larger class (200+). I’m afraid that in large classes there is just too much opportunity for chaos, and some nasty confrontations across students. So I don’t think my method here will easily scale up.

  • Rajesh Sudhakaran

    Dear Professor,

    The method instituted for the test is certainly laudable and a brilliant way of inculcating the game theory instincts in students. But I would like to know if the results would be same if the stakes were higher and more real life. Like a promotion is at stake in job or you are competing for the same resource etc. It would be a more interesting and successful study if it works.

    • One way to do that would be to create a cap on the total points awarded to all participants. (Assume 10 students, take a semi-average score of let’s say 80) State that there are only 800 points to go around. And you will likely see some above average competition.

  • As a proud Bruin my only regret at UCLA is not taking your class.

  • Prosthetic Lips

    Is this really teaching what humans are good at, or just good within the parameters you set? In nature, being better than the competitor is better, because you get something and s/he does not. If all animals mated, then there is less food, and some die out due to starvation / lack of energy (to get more food). This definitely shows that groups of people do better than single people in most cases (it sounds like the lone wolves still benefited from the discussions). Perhaps a more real-life scenario would be to assign points based on the size of the group (bigger group = smaller average points), and the group got to share the points equally. Then, aligning yourself with a few “brains” would get your group a much higher grade (or sharing the “brains” among all the groups?).

    Anyway, excellent idea on getting the students to think outside their normal comfort zone!

  • I know that at my Law School the professors were required to turn in grades that resembled a Bell Curve distribution. If Berkeley requires similar, then I wouldn’t be terribly shocked by the additional competitiveness. This is of course just a single hypothesis.

    There are definitely some additional competitive pressures to consider in the Law School environment.

  • Brian Bulkowski

    I had an interesting experience with a position we were trying to fill at a company. Instead of the usual interview system (45 minutes each, some practical, some design, etc) we decided to give each interviewee the real world example of a day at the company. We gave them a programming task that could be described simply and completed in half a day, and asked before hand what kind of laptop and tools they wanted preloaded.

    We then left them in a room and told them to complete the task any way they wanted. They could buy the answer on the internet, they could ask friends, they could use the internet. Shockingly, most candidates literally refused to put pen to paper – we got programs with literally just a few lines.

    To make it fair, the group hiring did the same question together, and compared all of our approaches – but all of us were able to get the program done, although some people took longer and did more complete work, some did quicker, shoddier work.

    The guywe hired was excellent,.

  • Jack Everitt

    My one big quibble here is that Prof. Nonacs seems to not understand the meaning of the word “cheating,” over and over again. No cheating occurred here. No reason for the word to be in this article, nor in its title.

    • jkern81

      He addresses this.

      “Although by conventional test-taking rules, the students were cheating, they actually weren’t in this case. Instead, they were changing their goal in the Education Game from “Get a higher grade than my classmates” to “Get to the best answer.””

  • doug

    Didn’t they already try this on an episode of Saved by the Bell: The College Years?

  • Fuxy

    This is brilliant.

    The ideal of letting students cooperate and take tests in groups really resonates with my idea of learning.

    They should do more tests like this and in different fields of study. I would actually recommend 50% of all tests be taken like this it encourages students to hunt for the right answer not compete against each other which seems to be a better goal and it also encourages cooperation.

    Let’s face it we’re a social species we work better in social environments.

    The other 50% should give the teachers a good idea who is doing good and who is doing worse in the class.

  • acjohnson55

    A collaborative assignment is a really solid education tool. That’s really all this is.

    What’s missing is an explanation of why the author thinks this led to an
    objective evaluation of what individual students learned *before* test day (if he does indeed think that). That is the point of an exam. Imagine if med students could collaborate on their boards. It would pretty well invalidate the purpose.

    • Peter Nonacs

      All in all, it worked out much better than I expected because the
      students really got into it. They spent a week communicating amongst
      themselves, planning strategy, spinning what if scenarios. Total game
      theory to the max! And game theory was what I was trying to get them to understand. So, one could say that the test was really not about the
      ‘answer’, but instead all about the thinking and planning on how to get
      to the answer before the test and then as the test unfolded. They passed that with flying colors.

      • InvitedGuest

        I’ve embraced take-home “open resource” exams over the past few years on the basis that they are better learning exercises than in-class exams that mostly just evaluate. The students report that they learn more in the class than had been reported in previous years.

        After reading this I’m going to experiment with an exam that allows cooperation.

  • What if the test wasn’t impossibly difficult?
    What if Prof Nonacs had set a mildly difficult test that the brightest ones knew they could do better on than most of the others?
    Since everyone was shunted far onto one side of the possibility of cracking that test, they naturally turned to each other for help. But what what if it was within reach for say 30-40 percent of the class?

  • ABC123DoReMi

    This would be way cooler if he had graded the test on a curve.

  • Thanks for the story. This is a very interesting application of cooperative game theory. There is an inherent assumption in several of the comments that the “real world” is only a competitive place, and that cooperative behavior is just a classroom result.

    Specialization and cooperation are different. Specialization requires each party to recognize the gains from trade. We can trade with those we do not trust by taking every advantage available. Cooperation requires each party to recognize the value of contributing and trusting without a guarantee to receive a greater return.

    Non-cooperative game theory is pretty easy to grasp. Cooperative game theory is more subtle. Biology and economics (my area) both show that cooperative behavior can solve certain types of problems where competitive game theory fails. Sometimes a smaller creature is more valuable as a teeth-cleaner than as a meal. Public goods create great value, but do not arise from competitive behavior.

    Almost everyone knows how to compete, so people place an exaggerated emphasis on competition. Learning when and how to cooperate is undervalued. You’ve taught them something truly important.
    Using this exam structure to teach student how to evaluate a choice between cooperative behavior and competitive behavior is brillliant.

  • This experiment compares the mob median to randomly sampled individual observations, and the individual observations show variance around the center. Statistics still work, even in biology classrooms.

    From the standpoint of educating kids and for future employers to know they’re hiring someone who knows what they’re up to, it’s not obvious that the general application of this testing strategy is desirable. Someone not present at this test would be incapable of discriminating between a mob member who contributed the most to the conversation, and someone who contributed not at all. Certainly they would be able to discriminate between the lone wolf who did the best, and the karmicly challenged lone wolf who tried to organize a “smart kid” cartel, with the cartel organizer not getting hired. Win!

    Considering that higher degrees have the purpose of distinguishing the student from others, isn’t it counterproductive to organize examinations which prohibit individuals from distinguishing themselves?

    [I understand this is not a claim made by the author, but is certainly a conclusion that could be drawn, and might be what the guy talking about de-beacking chickens means to say.]

    As an experiment in the coordination of knowledge, however, it looks like a good one, and the kids sound like they had fun. Which is what college is all about, eh?

    • Peter Nonacs

      Thought problem. You have a weird undiagnosed medical condition. You walk into a room of doctors. Who do you want as a physician? The smartest guy in the room? Or maybe the average guy, who is still smart enough to know he needs to ‘cheat’ and ask the smartest guy for his opinion. And maybe also ask the smartest guys in the next 5 rooms, too. What is the “goal” of education? To sift through everyone and rank the smartest? Or teach everyone as best you can to reach their innate capabilities as innovators and problem solvers?

  • Here is my answer to your exam question:

    Natural selection can be regarded as a game. However, we are not the players, but merely the pieces on the board.

    The game is solitaire, and nature is the only player.

    There are no teams, and the only rules are the laws of physics.

    The only objective is the continuation of the game, and the outcome on a personal level is that we are all sacrificial pawns.

    How did I do?

  • Eric Green

    As one of the students in this class I can say it was most certainly a different experience than any other that I’ve had. I was also the “lone wolf” who ended up doing better than the group and did so because I did not feel that the way the group was writing their answer was a sufficient response to the question although all the correct answers were on the board, and I used some of their arguments with some of my own. It was a very fun way to learn about Game Theory and I would highly recommend the class to anyone at UCLA. Thanks Dr. Nonacs!

  • uwestange

    What are the correct answers to the test? I would say, Dawkins in my mind: players: genes; team: organisms; rules: none; objectives: reproduce!; outcome: life on earth as we know it.

    Opinions? 🙂

    • Peter Nonacs

      Since people are asking, here is the entire question and my answer key. Besides being based on several weeks of lectures that have ‘framed’ the question, here are a couple things to note..
      1. The question is actually longer than its summary in the published piece: Article words limits and I wanted the question to be as explicit as possible for the test.
      2. The answer key is my expanded answer. The students nailed large portions of this, but I wanted them to see what else they could have put in. The question is also very open-ended and there are alternative ways to argue the proposition. Therefore, I would not claim that this is THE answer – it is the classes’ and MY best attempt at an answer.

      Write a short but incisive essay applying game theory to the general theory of Natural Selection. Be sure to clearly explain:
      What is (are) the objective(s) of the Natural Selection “game”.
      · What the rules are of this game.
      · Does it matter that some rules arise from behavioral ecology while others are defined by physics or chemistry? If so, why?
      · Who are the players and/or teams?
      · Can there be an optimal or unbeatable strategy that will always win. And if not, why not?
      · Can you ever win at the game of Natural Selection without explicitly
      ‘knowing’ the rules?
      Humans are in the process of identifying, or maybe already have identified all the rules, objectives and what determines winning in Natural Selection. How does knowing all this change the potential future outcomes? Or does it have no effect at all?

      The objective of a Natural Selection game is to survive. Over the history of life on earth, therefore, probably 99.9% of species have lost this game (although some may have been winners for a very long time). Survival, or winning, can take several forms. It can be an immortal individual that never dies. Although such individuals may exist somewhere, they are highly unlikely (i.e., accidents happen!). Alternatively, a species can be successful in its ecological niche to the extent that it excludes all others for extensive periods of time. For example, cockroaches have occupied their niche and excluded all wanna-be cockroach species for 100’s of millions of years. Finally, ancestral species can evolve into new ones, or exhibit an evolutionary radiation into several or many new species. So for example, Homo erectus is not a loser in this game even though there are no more H. erectus alive today. However, one of its descendant species, H. sapiens, is probably the ‘winningest’ primate ever. H. erectus descendant genes happily continue to play the NS game. In
      contrast, Neanderthals have lost – they have left no descendants. (Note – some may disagree here and argue that the 4% of Neanderthal genes found in some human populations means that at least a small part of the Neanderthal genome is a winner and continues in the game.)

      The players in this game can be defined as whatever can successfully reproduce. Most often we think of this as being a gene, or at least a strip of DNA such as a transposable element that can make copies of itself. But the player need not be a gene. It can be a cultural norm, an idea, a belief system, a computer virus, etc… Anything that has the potential to replicate and is governed by the rules of Natural Selection. Players can be organized into ever-ascending levels of teams. Genes
      into chromosomes; chromosomes into genomes and individuals; individuals into groups; groups into societies; societies into multi-species communities, ideas into religions or political parties.
      Each of these team levels may help or hinder overall survival by Natural
      Selection. “Teammates” at all levels may help each other to the benefit of all. Or they may increase their own success by sabotaging or exploiting their teammates. Survival, or winning, is however agnostic over how this is accomplished. The ultimate rule is: whatever works – works! Hence the world is full of a wide variety of cooperative, exploitative, nice and nasty winners.

      The rules of Natural Selection are often set by chemistry and physics. DNA, RNA, ATP, proteins, carbs, etc… all have their unique chemical properties and limitations. Whatever evolves must happen within these limitations. Physics limits size, movement rate, body shape, etc… Again, Natural Selection must operate within these hard limitations.
      Thus, fundamental processes of Natural Selection such as what materials you can use to best store information and pass across individuals (i.e., DNA) are set by chemical rules. Behavioral
      ecology rules such as “cooperate if…”, “do X if Y happens…” etc… may be very effective, but they are also completely changeable. They are an evolved consequence, not an evolutionary constraint. Hence, a simple objective such as “maximize inclusive fitness” can and is achieved in millions of different ways by millions of different species. And importantly, behavioral rules can be cheated upon, or changed to your advantage. You really can’t suspend the rules of chemistry or physics!

      Obviously if one can cheat or change the rules, then there is no unbeatable strategy – only what may work the best under the current set of conditions. Again, remember most who have played this game have lost and are extinct. Change an environment’s chemistry and physics and something else is likely to win. Change the behavioral rules, such as alter a Prisoner’s dilemma matrix to a Mutualism matrix, and you get new winning strategies for the Natural Selection Game.
      Equally obviously, success is not dependent on explicitly knowing the
      rules. Whatever works – works. No one needs to understand how it works. It is simply what the winners did.

      Yet if you do know the rules, like humans do (or at least think we do), it can strongly affect behavior and outcomes. To some degree, knowing the constraints established by physics and chemistry allows them to be ameliorated in some novel way. (Wear a coat when it’s cold outside!) Perhaps more importantly, humans understand the pitfalls of behavioral ecology rules. If you know it’s a Prisoner’s Dilemma, you know that you ought to play Tit-for-Tat. If you know that ocean fisheries are a Tragedy of the Commons, you can hopefully behave in ways that prevent that outcome. If you know that too much scrounging is bad for the group, you can enforce punishments to maintain group functioning and
      cohesion. In this test, you knew what the game was and what your objectives were, and you were allowed to write your own rules as to how you answered these questions. Given that your objective was to get a high score, do you think that you achieved this through your manipulation of the game? Or would you have done better if
      you had to play by UCLA’s ‘normal’ set of test-taking game rules that you are not allowed to alter?

  • Very interesting indeed. I wonder what would happen if the exam had a curve. Would the students cooperate the same way? Might be an interesting experiment!

  • The trouble with this is that “natural selection” is not entirely natural. And the newest frontiers of science and information theory are shedding a lot of doubt as to whether “natural selection” by itself could produce what we are and have.

  • Alfredo Louro

    >Instead, they were changing their goal in the Education Game from “Get a higher grade than my classmates” to “Get to the best answer.”

    Huh? Since when is the goal to get a higher grade than your classmates?

    • Alicia

      Many college level courses are graded on curves, so you are graded in comparison to your classmates.
      It was originally intended to account for varying difficulty in tests, but has mostly just given rise to a lot of competition between students.

  • Zoe Morosini

    I absolutely loved this idea. The article was also very well written.

  • Jim

    Which students passed the test?