When it comes to mental ability, many of our human talents were shaped by evolutionary forces that arose under the demanding conditions of life on the African savannah 35,000 to 50,000 years ago. Evolutionary psychologists have linked many of our current attributes to these earlier environmental challenges faced by our predecessors (Kanazawa, 2005). This much is noncontroversial. What is less agreed upon, however, is the extent to which present-day cognition is under the control of local conditions—that is, the specific physical, motivational, and psychological conditions under which humans attempt to solve problems. The argument I am making is that it is logically unsafe to claim that our successful performance on cognitive tasks today reflects our evolutionary preparation because the flip side is that our unsuccessful performance reflects our lack of evolutionary preparation—which may be wrong. In fact, a great deal of developmental research demonstrates that even when evolution has prepared us to undertake certain cognitive operations, successful performance depends on local conditions. We may or may not display the feats that evolution has made possible, depending on the local situation. This can be seen anecdotally as well as scientifically, and below I provide examples of both.
Let’s start with the halving rule that stipulates that any entity divided in half will yield two entities. If I ask a 3-year-old child: “If I cut an apple in half, how many pieces will I have?”, the typical child responds “two”. Then if asked “If I cut an orange in half, how many pieces will I have?”, the child will once again reply “two”. However, if the same is asked “If I cut a watermelon (or rug) in half, how many pieces will I have?”, the child often replies “it depends on how big it is” (Ceci, 1996). Clearly, preschoolers possess a halving rule, but their ability to display it is a function of such things as the nature of the entity (small, round edible objects, amorphous shapes, work in the physics sense, horsepower, etc.) Much of what we are capable of demonstrating will not be demonstrated if the local conditions are unfavorable. One important local ingredient is motivation. Depending on how motivationally engaging the task is, performance will vary, sometimes enormously. And when this occurs we are led to erroneous conclusions about what is and is not within the cognitive capacity of Homo sapiens.
Motivational Context. When researchers examine problem-solving ability in different settings that vary the motivation level, the results sometimes fail to replicate. For example, my late colleague, Urie Bronfenbrenner and I asked children to predict where on a monitor geometric shapes would migrate after children pressed the space bar. Each shape began in the center of the monitor, but after the space bar was pressed it would migrate somewhere else on the screen, and that location was completely determined by an algorithm that ran the experiment. In one case, there were three different shapes, a square, circle, and a triangle; they were small or large, dark or light. A curvilinear algorithm determined where each shape of a specific size and hue would migrate:
.8 sin(x) + .6 sin(y) + .4 sin(z) + 5% error1
However, even after 750 trials children were still unable to predict a shape’s migration. The implication is that multiplicative reasoning is beyond their mental capability; the algorithm that controlled the movement of the three shapes was simply too cognitively complex for the children to fathom.
However, when the identical algorithm was used to run a videogame that involved predicting the destination of three vehicles (large or small, dark or light) to avoid a road barrier, children achieved nearly perfect scores by 450 trials (for fuller description see Ceci, 1996). As seen in Figure 1, cognition in situations that are motivating and enjoyable such as the road barrier video game led to a view of the child’s mental ability that was far more generous than the view that emerged from the laboratory shape migration task. This is a common finding in cognitive development research, where slight changes to the setting or the motivational loading of the task can lead to dramatic changes in children’s performance.
Deductive reasoning. Obviously, evolution has prepared today’s children to handle a great deal of complexity but this ability requires the presence of certain situational and motivational conditions in order for it to be expressed. Much research in the areas of problem-solving, memory, and moral reasoning shows that if the testing conditions are void of meaningful associations for the child (e.g., detecting shapes in a sanitized laboratory) the children are less likely to reveal their evolutionarily-prepared underlying abilities. Below are several examples that demonstrate this.
A famous method of testing the ability to reason deductively is called the Wason’s task, after its inventor, the British psychologist Peter Wason. An inference of the form:
p or q
should be seen as valid regardless of its propositional content. However, it is clear that content matters. Consider two types of content: in one case four cards are laid out with the instructions that each card has a number on one side and a letter on the other. Adults are given the rule if a card has a vowel on one side then it has an even number on the other side:
E K 4 7
They are asked to determine which cards must be turned over to validate the rule. Even very bright students at select universities correctly answer only 12% of such problems. Yet, if the content is changed from numbers and letters to trains and cars on one side of each card and a familiar city on the other side (every time I go to Boston, I go by train):
Boston Chicago Train Car
nothing has changed in terms of the task’s cognitive complexity, but performance shoots up to 60% (Johnson-Laird, 1983). Once again, we see the role that context plays in the assessment of someone’s mental capacity, and this task has been employed by evolutionary psychologists to make inferences about the impact of natural selection on human reasoning processes (e.g., Cosmides, 1989). But because evolution prepares us to succeed at tasks that result in our genes being added to future generations, then perhaps such demonstrations are beside the point. Perhaps we should focus on tasks that are most obviously associated with survival and mating. Well, it turns out that cognitive operations that foster survival are particularly likely to display this context-dependency. When the context is devoid of any survival value, performance may suggest that we lack the cognitive ability to perform it successfully, whereas when the identical cognitive operations are needed in service of survival, they should be manifest. Here is one such example, a version of the so-called Monte Hall 3-door problem.
You are presented with three boxes and told one of them has a $100 bill hidden inside. If you guess which box, you get to keep the $100. On average, of course, we would guess correctly one third of the time. But suppose I modify the procedure and tell you that after you choose the box you think contains the $100 bill, but before you open it to find out if you guessed correctly, I open one of two unselected boxes—always one that does not contain the $100. Now you are left with two unopened boxes, the one you chose and the other unopened box. I offer you the option of switching boxes: should you stick with your original choice or switch to the other unopened box? Does it matter? Are the probabilities the same or different?
Piattelli-Palmarini (1991) has reported that even distinguished scientists and mathematicians argue about this, with some arguing that it doesn’t matter because the odds will always be 1:3, but others arguing that it matters a great deal. As noted, the odds that your original choice is correct is 1:3 and the odds you are wrong is 2:3. So on the face of it, it would not seem to matter if you switch or stick with your original choice because the odds that the two remaining unopened boxes contain $100 bill are the same, 1:2. But this is actually not the correct way to look at such problems. As Piattelli-Palmarini pointed out, the box you originally chose had and forever will have a 1/3 probability of containing the $100 bill. And the other two boxes combined have a 2/3 probability of containing it. However, after one of these two boxes is opened and thus removed from the equation, the remaining unopened box assumes the full 2/3 probability by itself. Thus, the simple act of switching from your original box (1/3) to the unopened one (2/3) doubles your chances of winning the money! If the box you originally chose contains the $100 bill, then you will be penalized by switching to the other unopened box. But this will occur only 1 in 3 times. Conversely, if the box you chose does not contain the money, then switching will guarantee that you get the money, and this will occur 2 out 3 times.
To see where evolution comes into this, let’s alter the content from boxes with money to a potentially dangerous survival situation to see if people reason any more effectively than they do with the box problem. My colleagues and I posed a similar problem to a group of Brazilian child “bookies” who quoted odds on gambling tickets they sold. Since the type of gambling they do involves a similar conundrum to the 3-door problem above, it is unsurprising that 7 of the 9 bookies rapidly and correctly solved it, shouting to switch almost before we could finish posing the problem to them. And yet when we asked these 9 bookies to solve an analog of this problem that did not involve gambling, only 3 of them correctly solved it. Here is the analog we gave them: “Suppose that it is critically important that you reach another city 200 miles away by midnight, and three different strangers all offer to drive you. However, one of the three is a mass murderer. Obviously, for your survival, it is critical that you not ride with this individual, but you have no idea which of the three it is. Suppose that you “bite the bullet” and select one of the three drivers and after you made this choice, I remove one of the two remaining drivers, always a non-murderer. Should you stick with your original choice or switch to the remaining driver? Only three of the bookies who solved the money problem correctly solved this survival problem (Roazzi & Ceci, 1994). Interestingly, when couched in terms of one of the three drivers being a rapist, women were superior to solving it than men. Schliemann and Carraher (1994), among others, have documented similar findings with Brazilian street children doing everyday math for which an error could be quite costly to their earning a livelihood.
Educational implications. The goal of education is not to drum facts and concepts into children, but to create awareness of how these facts and concepts can be generalized to situations that differ from the ones used to teach them. Thus, the key is transferring knowledge from the contexts used to teach it to ones encountered outside of school. And yet, a great deal of empirical research has documented that young and old, high IQ and low IQ, schooled and unschooled, all fail to transfer learning to new contexts that differ from the context in which they were originally taught (e.g., Ceci, 1996; Leshowitz, 1989). The research described here suggests that context is a constituent of cognition, not something adjunctive or peripheral to it. This view of cognition-in-context has several implications for education.
If cognitive structures are initially tied to the contexts in which they are taught, then attempts to assess students’ understanding of these cognitive structures should reinstate the original contexts because structures may not be instantiated by testing contexts that differ from the learning ones. Reinstating original learning contexts will demonstrate if the student acquired the knowledge in the first place and can recall it. However, this is not enough; no one would be satisfied if students who got A’s in math were unable to calculate their bowling averages or figure change at the store just because these were not the contexts in which they originally learned math. Teaching for transfer of knowledge requires the provision of myriad contexts during learning to enable students to see the connections across contexts, such as applying an insight from algebra to physics (Bassok & Holyoak, 1989). The typical developmental trajectory is one in which a child initially acquires a concept in a specific context and then gradually learns the same concept in different contexts. When this happens, some (but not all) students begin to realize the generality of what they learned, and they make connections across domains (e.g., anything cut in half yields two parts). Elsewhere, I provide examples of how this occurs (Ceci, 1996). But for educators the challenge is to design learning situations that allow cross-domain connections to be formed.
In conclusion, ignoring motivation and meaning—as much of the research in the social sciences routinely does—can lead to findings that do not transfer to important real-world contexts. It can result in ungenerous depictions of the cognitive abilities of people when they appear to be unable to carry out the underlying cognitive operations to solve a task that poses no important ecological challenges for them, when in actuality they often are able to carry out these cognitive operations under motivationally rich circumstances that represent important challenges for them. In work with my colleagues, Urie Bronfenbrenner and Jeff Liker, we have demonstrated that context plays a critical role in cognitive performance; expert gamblers (Ceci & Liker, 1986) and children baking cupcakes (Ceci & Bronfenbrenner, 1985) all exhibit modes of cognizing that are tied, at times surprisingly tightly, to the social and physical context in which the task is presented. Whatever talents were shaped by the demands of living in small, closely-knit hunter-gatherer groups on the African savannah 35,000-50,000 years ago, are most likely to be exhibited when we face modern circumstances that activate the same motivations under which these talents were formed.
Stephen J. Ceci is a developmental psychologist at Cornell University. He studies the accuracy of children’s courtroom testimony (as it applies to allegations of physical abuse, sexual abuse, and neglect), and he is an expert in the development of intelligence and memory. He has been the recipient of numerous awards, including the prestigious Lifetime Contribution Awards from the American Psychological Association (APA) and the Association for Psychological Science (APS) as well as many divisional and smaller society awards.
Bassok, M. & Holyoak, K. (1989). Interdomain transfer between isomorphic topics in algebra and physics. Journal of Experimental Psychology: Learning, Memory, Cognition, 15, 153-166.
Ceci, S. J. (1996). On Intelligence: A bio-ecological treatise on intellectual development. 2nd ed. Cambridge, MA: Harvard University Press.
Ceci, S. J. & Roazzi, A. (1994). Context and cognition: Postcards from Brazil. (pp. 26-49). In R. J Sternberg & R. K. Wagner (Eds.), Intellectual development. New York: Cambridge University Press.
Ceci, S.J., and Bronfenbrenner, U. (1985). Don’t forget to take the cupcakes out of the oven: Prospective memory, time-monitoring and context. Child Development, 56, 152-164.
Ceci, S. J. and Liker, J. (1986). A day at the races: IQ, expertise, and cognitive complexity. Journal of Experimental Psychology: General, 115, 255-266.
Cosmides, L. (1989). The logic of social exchange: Has natural selection shaped how humans reason? Cognition, 37, 187-276.
Johnson-Laird, P.N. (1983). Mental models. Cambridge, MA: Harvard University Press.
Kanazawa, S. (2005). Why Beautiful People Have More Daughters.
Leshowitz, B. (1989). It is time we did something about scientific illiteracy. American Psychologist, 44, 1159-1160.
Piattelli-Palmarini, M. (1991, March/April). Probability: neither rational nor capricious. Bostonia, pp. 28-35.
Schliemann, A. D. & Carraher, D. W. (1994). Proportional reasoning in and out of school. In P. Light & G. Butterworth (Eds.), Context and cognition: Ways of learning and knowing. Hemel Hamstead, UK: Harvester Wheatsheaf.