For neural dissociation abstract (Stone et al., #3526) <Click here>

    For cross-cultural abstract (Sugiyama et al., #3529) <Click here>

Social exchange is cooperation for mutual benefit.  It is an “I’ll scratch your back if you scratch mine” principle: X provides a benefit to Y conditional on Y doing something that X wants. This very basic form of cooperation is zoologically rare — it occurs in only a handful of species.  But it is as characteristic of human beings as language and tool use.  

Together, these two PNAS companion papers provide new and converging evidence that the human neurocognitive architecture contains programs that are specialized for reasoning about social exchange, with a subroutine designed for cheater detection.  

  • Stone et al reports neuropsychological evidence: the cheater detection subroutine can be selectively impaired, and it dissociates from other, very closely related forms of reasoning.  
  • Sugiyama et al reports cross-cultural evidence from the Shiwiar, a hunter-horticulturalist group in the Amazon: it supports the species-typicality of this reasoning system, and shows that the subroutines relevant to its functioning as an evolutionarily stable strategy (ESS) are developmentally buffered against cultural variation. <Click here for photos>

The results implicate a universal, species-typical neural system that is functionally specialized for reasoning about social exchange. Joint findings of cross-cultural universality and neural dissociability provide a potent combination, one that we hope will set a new and higher standard of evidence for those investigating the architecture of the human mind.


Social exchange — also known in biology as reciprocal altruism, reciprocation, or tit-for-tat —  is an ancient, pervasive, and central part of human social life.  This mutual provisioning of benefits, each conditional on the others’ compliance, is rare in the animal kingdom.  Some species — humans, vampire bats, chimpanzees, baboons — engage in this very useful form of mutual help, whereas others do not.  This is itself telling: Social exchange cannot be generated by a simple general learning mechanism, such as classical or operant conditioning.  All organsisms can be classically and operantly conditioned, yet few engage in exchange.  That strongly suggests that engaging in social exchange  requires specific cognitive machinery, which some species have and others lack.

        That is, there are good reasons to think we humans have cognitive machinery that is functionally specialized for reasoning about social exchange — programs that make thinking about and engaging in social exchange as easy and automatic for humans as building a dam is for a beaver or spinning a web is for a spider.  Such programs are sometimes called “reasoning instincts“.  There are good reasons to suspect that we have evolved cognitive instincts for reasoning about social exchange, and that natural selection has had a long time to shape their design::

  • Social exchange is a human universal.  Moreover, it is richly expressed in all human cultures, taking many different, highly elaborated forms: reciprocal gift giving, food sharing, market pricing, symbolic trades, implicit bonds of mutual obligation, and so on.  
  • It is not a recent cultural invention, like the alphabet, computer programming, or rice cultivation. There is no evidence of a point of origin, of its being spread by contact, or of its being absent in any culture.  In these respects, it is like language (and unlike cultural inventions like writing).
  • Paleoanthropological evidence (e.g., hunter-gatherer archaeology) suggests that this form of cooperation existed in hominids at least 2 million years ago
  • The presence of social exchange in related primates (from whom hominids diverged 5-30 million years ago) suggests that selection pressures involving social exchange have been shaping the minds of our ancestors for a very long time.

Robert Trivers, W. D. Hamilton, Robert Axelrod and other evolutionary researchers used game theory to understand the conditions under which social exchange can and cannot evolve.  For adaptations causing this form of cooperation to evolve and persist (i.e., to embody what evolutionary game theorists call an evolutionarily stable strategy (ESS)), cooperators must have mechanisms that perform certain specific tasks.  For example, reciprocation cannot evolve if the organism lacks reasoning procedures that can effectively detect cheaters (i.e., those who take conditionally offered benefits without providing the promised return).  Such individuals would be open to exploitation, and hence selected out.  

Based on such analyses, Leda Cosmides and John Tooby hypothesized that the human neurocognitive architecture includes social contract algorithms: a set of programs (neural circuits) that were specialized by natural selection for solving the intricate computational problems inherent in adaptively engaging in social exchange behavior, including a subroutine for cheater detection.

This led to a series of investigations in cognitive science, the goal of which was to see whether the evolved architecture of the human mind contains computational procedures that are functionally specialized for detecting cheaters in situations of social exchange.  A great deal of data suggesting the existence of such procedures has been amassed (indeed, the first work on this, by Cosmides, was awarded the 1988 Prize for Behavioral Science Research by the American Association for the Advancement of Science).  But until now, this data has been in the form of laboratory experiments conducted on brain intact individuals and in industrialized nations.   

Relevance of “Selective impairment of reasoning about social exchange in a patient with bilateral limbic system damage” (PNAS #3526Stone, Cosmides, Tooby, Kroll & Knight).

Neural dissociation.  A very strong form of evidence that an ability is specialized rather than general is the demonstration that it can be selectively impaired, while leaving intact behavior or judgments that were supposedly the expression of the same, more general capacity.  PNAS #3526 presents evidence from patient R.M., who has focal brain damage, showing that social exchange reasoning can be selectively impaired while reasoning about other domains is left intact.

Indeed, the dissociation discovered is very narrow: patient R.M., whose cheater detection is impaired, is entirely normal on reasoning tasks that are formally identical to the cheater detection problems.  The problems on which he is unimpaired differ from the cheater detection problems (on which he is impaired) only in that they deal with a different domain of human life (taking precautions against hazards).  No other theory proposes reasoning operations that distinguish problems based on this difference in surface content.  Consequently, alternative theories provide no means for predicting that performance on the two problems types could differ.

R.M.’s selective impairment is inconsistent with the hypothesis that cheater detection is simply the expression of more general reasoning abilities.    If human reasoning were carried out by a set of general-purpose procedures, as many believe, then damage to that reasoning system should degrade performance across the board, not solely on social exchange problems.  It is extremely difficult to explain this dissociation except on the hypothesis that there is a specialized ability to detect cheaters in situations of social exchange that is distinct from other reasoning abilities.  It suggests instead that cheater detection is caused by circuitry that is specialized for that function. (See Relevance to debates on human rationality and social intelligence).  

Relevance of “Cross-Cultural Evidence of Cognitive Adaptations for Social Exchange among the Shiwiar of Ecuadorian Amazonia”  (PNAS #3529: Sugiyama, Tooby, & Cosmides)

Universal and species-typical.  Sugiyama et al. tested another prediction that follows from the proposal that cheater detection is caused by a complex, evolved brain specialization.  If, as hypothesized, this specialization evolved as part of our universal, species-typical design, then this leads to the falsifiable prediction that the same pattern of reasoning should be detectable in widely divergent cultural settings.  More precisely, we predicted that those aspects of reasoning performance relevant for social exchange algorithms to perform their evolved function should be buffered against cultural variation, and therefore be uniform, whereas the function-irrelevant aspects should be free to vary with cultural circumstance.

PNAS #3529 reports experiments demonstrating both claims, using an oral and pictorial version of the same kind of task that was given to R.M..  Whereas subjects drawn from foraging and small scale societies usually perform poorly on reasoning tests developed in industrialized nations, these results show that nonliterate, isolated Amazonian hunter-horticulturalists produce the same specialized reasoning patterns as Harvard undergraduates.  Indeed, the results show that, as predicted, it is specifically the adaptively relevant aspects of this reasoning system — the aspects that are necessary for social exchange to be an evolutionarily stable strategy (ESS) — that are developmentally buffered against cultural variability.  

Sugiyama administered reasoning tasks that assay cheater detection to Shiwiar people – members of a hunter-horticulturalist population in an area of the Amazon so remote that it cannot be reached by road or boat.  (As a people, the Shiwiar are among the least exposed on the planet to the influence of globalization.)  On those aspects of cheater detection that are relevant to its functioning as an evolutionarily stable strategy (ESS), the performance of Shiwiar was identical to that of Harvard undergraduates.  Yet it is difficult to imagine two populations that differ more on factors that are usually relevant to reasoning performance (literacy, education, degree of exposure to a market economy, etc). 

 Are the results novel and unexpected? Yes.  Most psychologists and anthropologists believe that high level cognitive competences emerge from general-purpose cognitive abilities trained by culturally specific activities, rather than as part of our evolved, reliably developing, species-typical design. (see Human Rationality) That cheater detection should be well-developed among the Shiwiar is a falsifiable prediction of the evolutionary account, which posits that this competence should be distributed in a species-typical, human universal fashion.  Alternative accounts positing that such high level, domain-specific competences are the product of culture predict a more culturally contingent distribution of reasoning abilities, rather than human universality: one should see “cultural dissociations”, such that abilities found in some societies (e.g., developed nations) are absent in others. For example, many mathematical abilities are found in only some cultures, can be shown to have spread by contact between cultures, and are known to be acquired by explicit instruction.  Cheater detection reasoning has been found in every developed and developing country that has been investigated, but a remaining test was to find a culture as different from industrialized societies as possible, and as isolated as possible, to see if it was present there as well.  Finding that it is adds substantial weight to the claim of species-typicality for the social exchange specialization, and shows that this ability cannot easily be explained by economic development, exposure to European derived cultures, and so on. 

Cultural variation and the ESS concept in biology.  Methodologically, no cognitive neuroscientists have previously proposed looking for a “cultural dissociation” in reasoning (analogous to a neural dissociation), especially one that should differentially impact ESS-relevant versus ESS-irrelevant aspects of an adaptation. This is what we found.  The only difference between the populations was that Shiwiar showed more interest than Harvard students in investigating possible acts of generosity.  Shiwiar and Harvard undergraduates were identical when it came to investigating potential cheaters. It is the only work we know of to show identical performance across very different cultural groups (Shiwiar hunter-horticulturalists and Harvard undergraduates) on those aspects of a reasoning problem that are relevant to the adaptation functioning as an evolutionarily stable strategy, yet different performance on those aspects that are irrelevant to the adaptation functioning as an ESS.  

Finally, the previous history of cross-cultural research on human reasoning has identified few universals: on most tests of reasoning, subjects drawn from small scale foraging populations do poorly.  But most past tests have investigated the kind of logical and mathematical thinking that requires schooling to learn. They have not asked which reasoning abilities might be part of an evolved system – i.e., part of human nature.  In contrast to most prior work in cross-cultural psychology, these experiments were looking for evidence of a reasoning instinct.

Relevance of these joint results to various debates in the natural and social sciences.

The discovery that social exchange reasoning is a neurally dissociable, species-typical cognitive adaptation is significant for a number of reasons: 

Human rationality   First, it challenges a major assumption in the behavioral sciences: that human reasoning and rationality are produced by a small set of computational procedures that are general purpose (rather than functionally specialized for evolutionarily important domains) and domain-general (i.e., that apply the same operations to all domains of human life, whether they involve social exchange, foraging, predator avoidance, choosing a mate, managing hazards, or, indeed, doing schoolwork involving logic, mathematics, and probability).  Research indicating that the mind has a system functionally specialized for cheater detection cuts to the heart of debates on the nature of human reasoning and rationality.  It raises the possibility that the power of human intelligence comes from bundling together a collection of diverse mechanisms, each of which is specialized for reasoning about a different adaptive domain.  This has stirred a great deal of interest and debate, not only in the cognitive sciences, psychology, and neuroscience, but in economics, law, anthropology, and philosophy – indeed, in every field that concerns itself with human rationality. 

Selection pressures  There is active research in evolutionary biology on what information-processing steps are involved in the evolution of reciprocation.  There has been widespread interest in whether selection pressures identified by evolutionary biologists can be shown to have shaped higher mental functions in humans.  These findings directly address this question.  The cross-cultural data are particularly relevant.  Game theoretic results show that, for social exchange to be an evolutionarily stable strategy (ESS), individuals must be able to detect cheaters.  In other words, adaptations for social exchange cannot evolve in a species without programs that cause cheater detection; other properties are more free to vary.  The data from the Amazon are striking from this point of view: When it comes to identifying potential cheaters, the performance of Shiwiar hunter-horticulturalists is identical to that of Harvard undergraduates.  In other words, parts of the adaptation that are relevant to its functioning as an ESS were buffered against cultural variation.  Other parts of the adaptation (such as curiosity about acts of generosity) were freer to vary with cultural situation.  

Social intelligenceEvidence that the human mind may contain a cognitive adaptation specialized for cheater detection also speaks to a current debate about social intelligence: Does the human brain contain neural specializations designed to deal with the social world, as distinct from other types of information?  The claim that humans possess a distinct social intelligence has been supported by work over the last decade on the “theory of mind” mechanism and its selective impairment in autism by (among others) Alan Leslie at Rutgers and Simon Baron-Cohen at Cambridge, as well as by recent work on selective impairment of judgments of emotion recognition, moral reasoning, and trustworthiness in patients with limbic system damage (e.g., by Valerie Stone, Antonio Damasio, Ralph Adolphs, and others).  These companion papers are the first to provide support at the neural level for a third component of social intelligence (social exchange reasoning).

Frequently Asked Questions (FAQ)

Q:  How do you test people’s reasoning

A: We used a standard test called the Wason selection task. For examples of this task, <Click here>. The Wason task is an information search task.  When it presents a rule involving social exchange, it tests whether people know what counts as cheating, and whether they understand what information they would need to get to see if someone had cheated.  It was developed originally as a test of logical reasoning, to see if people are good at looking for violations of logical rules. For background on this task and why it was used, <Click here>.

Q:  Does this mean we are good at catching liars?

A:  Lying is a form of deception.  Cheating and deceiving are not one and the same thing. A cheater is someone who has taken a benefit from you without doing what you asked in exchange.  It is surely true that many people try to hide the fact that they have cheated through lies and deception.  But that is a separate matter (in fact, sometimes people cheat us without bothering to hide the fact!).  Our test is an information search task (see question above).  It is not a test of people’s ability to “read” nonverbal communication.  It does not assess whether people are good at “reading” faces and voices to tell whether someone is lying. (For interesting work on that topic, contact Paul Ekman (UCSF) and Bella DePaulo (UCSB)).

Q:  If we are so good at detecting cheaters, they why do so many people get away with cheating? How does an Enron thing happen?

A:  First of all, exchange usually goes very well.  Every time you buy something at a store, restaurant, website (etc), you have engaged in a successful social exchange: you got a carton of milk, the grocery store got your money.  In fact, in market economies like our own, every day we trade with total strangers…and, even so, most social exchanges are successful.  This is, in part, because we are interested in and good at cheater detection; so good, that market economies have developed many safeguards to deter cheating (cash register records, security mirrors, accounting firms, etc) and to make sure that, if cheating does occur, it is punished (collection agencies, credit card companies, police).  

        That said, it is important to realize that our cheater detection mechanisms evolved in face-to-face societies, small hunter-gatherer groups where people lived together, knew each other well, could more easily observe what the people around them were doing and, therefore, where it was more difficult (though not impossible) to keep secrets or hide what you were doing.  Cheater detection mechanisms did not evolve for a world with complex ledgers filled with numbers, fancy accounting practices, and so on.  Indeed, in cases like Enron, it (allegedly) took a legion of trained “professionals”, complex accounting practices, and lots of deception to obscure the real value of the company from those who bought or held their stock believing it was worth more than it actually was.  

        People are now suspicious of anything that looks like it might be “creative” bookeeping, even when it is on the up and up — see Time Magazine’s article, Under the Microscope (2/4/02).        

Q:  What parts of the brain are involved in cheater detection?

A:  R.M., who had a selective deficit in cheater detection, had bilateral oribital frontal damage, bilateral anterior temporal damage, and a total disconnection of both amygdala.  Another patient who had massive oribital frontal damage, but nothing else, was normal on both tasks.  So was a patient who had anterior temporal damage, with partial (but not complete) disconnect of his amygdala.  This suggests that the amygdala, or some circuit involving several areas including the amygdala, might be involved.

For FAQ from Valerie Stone, <click here>

Press account links: NatureNew ScientistNPRBBCCox NewsReutersLA TimesYahooThe Independent (UK), Daily Telegraph (UK), Ananova (UK), The Globe and Mail (Canada), The Strait Times (Singapore), VOA news (Denver), SB News-Press.

Contact Information for Prof. Lawrence S., tel: 541-346-5142 or 541-683-7823.  Department of Anthropology, University of Oregon, Eugene, OR 97403

Note 1.  The cognitive programs that cause social exchange in different species do not need to be the same; for example, vampire bats only share foraged blood, whereas humans share a wide variety of goods and services.  But the point is that, whatever programs allow the behavior, they need to solve certain problems, such as detecting cheaters.

Note 2.  Functionally specialized: that is, the design of the relevant programs makes them very good at solving problems involving a particular adaptive problem (social exchange, in this case).  To make an analogy to home computers: Word is a program that is functionally specialized for editing prose; Excel is functionally specialized for number crunching.  We are saying that the mind contains programs that are functionally specialized, in just that sense, for reasoning about social exchange.

Note 3.  Reasoning instincts.  In The Language Instinct, Steve Pinker pointed out that language has the following properties:  (1) It is complexly specialized for solving an adaptive problem, (2) it reliably develops in all normal human beings, (3) it develops without formal instruction, (4) it develops effortlessly, and without conscious understanding of its underlying logic, and (5) it is distinct from more general forms of intelligence.  But these are the characteristics people usually have in mind when  they call something an “instinct”.  We mean instinct in the same way: reasoning about social exchange seems to have all of these 5 properties.

Note 4.  Laboratory evidence using the Wason selection task.  Because conditionally delivered behavior requires conditional reasoning for its regulation, Cosmides & Tooby used the Wason selection task, an experimental protocol developed to study conditional reasoning, in order to test for the presence of social contract algorithms and their predicted properties.  The Wason selection task asks subjects to look for violations of a conditional rule (If P then Q), such as “If a person eats hot chili peppers, then that person will drink cold beer”. 

        A conditional rule is violated whenever P happens but Q does not happen (in this case, whenever someone ate hot chili peppers but did not drink cold beer).  In the Wason task, the subject is given incomplete information about four people [in this case, one ate hot chili peppers (P), one ate broccoli (not-P), one drank cold beer (Q), one drank hot tea (not-Q)]. To respond correctly, the subject would need to check the person who ate hot peppers and the person who drank hot tea (i.e., P & not-Q).  Yet studies in many nations have shown that reasoning performance on descriptive rules like this is low: only 5-30% of people give the logically correct answer, even when the rule involves familiar terms drawn from everyday life. <For sample Wason tasks, and how the mind “sees” them, click here>

        To show that people who ordinarily cannot detect violations of conditional rules can do so easily when the rule expresses a social contract and a violation represents cheating would be (initial) evidence that the mind has reasoning procedures specialized for detecting cheaters. If this succeeds, then one looks for futher evidence of special design 

        That is just the pattern found: People who ordinarily cannot detect violations of if-then rules can do so easily and accurately when that violation represents cheating in a situation of social exchange.  Given a rule of the general form, “If you take benefit B, then you must satisfy requirement R” (e.g., “If you borrow my car, then fill up the tank with gas”), people will check the person who accepted the benefit (borrowed the car) and the person who did not satisfy the requirement (did not fill the tank) – the individuals that represent potential cheaters. The adaptively correct answer is immediately obvious to almost all subjects, who commonly experience a pop-out effect.  No formal training is needed.  Whenever the content of a problem asks one to look for cheaters in a social exchange, subjects experience the problem as simple to solve, and their performance jumps dramatically.  In general, 65-80% of subjects get it right, the highest performance found for a task of this kind.  

        It’s not familiarty.  This good performance has nothing to do with the familiarity of the rule tested.  First, familiarity does not enhance performance for descriptive rules.  Second — and most surprising — people are just as good at detecting cheaters on culturally unfamiliar or imaginary social contracts as they are for ones that are completely familiar, providing a challenge for any counterhypothesis resting on a general-learning skill acquisition account. A totally unfamiliar rule — e.g., If a man eats cassava root, then he must have a tattoo on his face — can elicit excellent cheater detection.  All one needs to do is embed it in a scenario that says that the people involved consider eating cassava root to be a benefit, and having a tattoo the requirement one must satifsy to be eligible for that benefit.  

          It’s not logic.  Further experiments showed that subjects do not choose “P & not-Q” on social contract problems because they activate logical reasoning.  Instead, they activate a differently patterned, specialized, logic of social exchange.  For social exchange problems when formal logic (i.e., the propositional calculus) and social exchange logic predict different answers, subjects overwhelmingly follow the evolved logic of social exchange. They investigate anyone who has taken the benefit and anyone who has not satisfied the requirement that it was contingent upon, even if this results in a logically incorrect answer, such as “not-P and Q”.  (See Sugiyama et al on standard versus switched social contracts).It’s not a general ability to reason about permission rules (not a deontic logic).  Rules that are very similar to social contracts, but that lack important features, do not elicit the cheater detection effect.  For example, rules expressing that you are permitted to take an action only if a precondition has been met do not elicit good detection of violations unless the acion to be taken is a benefit to the potential violator. (But see below, on precautionary rules.)

Special design for social exchange.  Many cognitive scientists have now investigated social contract reasoning, and many of the predicted design features have been tested for and found.  For example, the mind’s automatically deployed definition of cheating is tied to the perspective one is taking; for the reasoning enhancement to occur, the violations must potentially reveal cheaters; when detecting violations of social contracts would reveal only innocent mistakes, enhancement does not occur.  For citations to reviews of evidence relevant to the claim that there are cognitive mechanisms specialized for cheater detection, click here:

        Perhaps the strongest evidence that there is a neural specialization designed for cheater detection is the discovery that cheater detection can be selectively impaired by brain damage, without impairing other reasoning abilities.  If social contract reasoning were a byproduct of a more general ability to reason, one could not lose the specific ability without also suffering damage to the general ability supposedly responsible for it.  This is the result reported in Stone et al. (see above)

        Consistent with social exchange reasoning being a species-typical ability, social contract reasoning effects are found across cultures, from industrial democracies to hunter-horticulturalist groups in the Ecuadorian Amazon.  This is the result reported in Sugiyama et al. (see above)  Indeed, performance is uniform specifically on those components of the adaptation that are essential to its functioning as an evolutionarily stable strategy.  Other aspects were freer to vary.

Managing hazards: Another evolutionarily important problem: .  Social exchange is one of many evolutionarily important problems for which there should be functionally specialized reasoning systems.  Another is managing hazards, which requires precautionary reasoning.  Precautionary rules have the form, “If one engages in hazardous activity X, then one must take precaution Y.”  People are good at detecting violations of precautionary rules, though performance on precautionary rules does not vary with conditions that affect cheater detection (e.g., intentionality is not relevant to performance — nor should it be. Violating a precaution puts you in danger, whether you did it on purpose or not; see Fiddick, 1998).  Social contracts and precautions are both permission rules, but not all permission rules are social contracts or precautions.  People are not good at detecting violations of permission rules that do not fall into one of these two categories.

 Wason tasks with precautions versus social contracts.  Due to focal brain damage, the patient suffered a selective impairment in his ability to detect cheaters on the Wason selection task (a task originally designed to test logical reasoning). The advantage of the selection task is it allows the experimenter to vary the content one is asked to reason about, while holding all else constant.  This allows one to compare performance on reasoning tasks that are formally identical, and that tax performance factors (e.g., memory load) to exactly the same extent.  Prior selection task results showing that (brain-intact) people are good at reasoning about a different domain – precautionary rules – provided a control task that is ideal for the purpose of demonstrating that social exchange is narrowly dissociable: This control task is so parallel in every respect to the cheater detection task that no other theory in the cognitive sciences views it as a different task at all.  The patient R.M.’s ability to reason well on this control task – to detect violations of precautionary rules – is completely unimpaired (he also performs well on other tests of general cognitive function).  Yet his ability to detect cheaters is markedly impaired, compared both to normal controls and to other individuals with brain damage in closely related areas. 

Why dissociations are interesting:  Precautionary rules and social contracts are not considered to be psychologically different by other theories. That is, every other reasoning theory says that one and the same cognitive mechanism causes reasoning about these two types of rule.  If this were true, damage to this one mechanism would necessarily knock out reasoning about both precautions and social contracts.  Only the hypothesis that there are two adaptive specializations — one for reasoning about hazards, the other for reasoning about social exchange — predicts that it should be possible, in principle, for one system to be damaged while the other remains intact.  And that is what the Stone et al. study found — a patient whose precautionary reasoning is entirely normal, but whose ability to detect cheaters on social contracts was very impaired.  

This is called a single dissociation (ability A impaired, ability B intact).  In this case, a single dissociation is particularly probative because the tasks are, from a logical point of view, identical.  They all use the Wason task, they were all about the same number of words long (same memory load), normal, brain intact subjects performed equivalently on both sets of problems, and there was no ceiling effect in the normal subjects (a ceiling effect could mask real differences in difficulty between the stimulus sets).  Thus the two problem sets appear to be perfectly matched; they differ only in their content (social contracts or precautions). This matching in task demands controls for the possibility that some unknown and extraneous feature of the stimulus set made one set of problems more difficult than another for theoretically irrelevant reasons (that is, reasons not related to social contract or precautionary reasoning per se).  For a dissociation in which cheater detection is intact, and more general logical reasoning abilities impaired, see Maljkovic, 1987 (cited in Stone et al.).

Note 5.  Evidence of special design.  To show that an aspect of the phenotype is an adaptation, one needs to show that it is well-designed for solving the adaptive problem in question (e.g., cheater detection, efficient foraging, color vision), and that it is not better explained as a byproduct of some other adaptation or some more general process.  In the case of reasoning about social exchange, one needs to show that performance shows the specific patterning one would expect of a program designed for solving problems involving social exchange, and that these patterns are not better explained as a byproduct of more general reasoning processes (such as propositional logic, deontic logic, permission schemas, or any other proposals).  Much of our research has involved testing against such alternative hypotheses. The neural dissociation, for example, is inconsistent with every alternative hypothesis that has been proposed so far.   For more on standards of evidence for calling something an adaptation, see the classic work by George C. Williams, Adaptation and Natural Selection: A critique of some current evolutionary thought.  (1966, Princeton University Press)

Note 6.  On why the genetic basis of adaptations coded for by suites of different genes should be universal in humans, see:

Tooby, J. & Cosmides, L.  (1990).  On the universality of human nature and the uniqueness of the individual: The role of genetics and adaptation.  Journal of Personality, 58, 17-67.

<Click here> to download the paper.  (large, slow, PDF file)

Reviews of relevant evidence:

Cosmides, L. & Tooby, J.  (1992).  Cognitive adaptations for social exchange.  In J. Barkow, L. Cosmides, & J. Tooby (Eds.), The adapted mind: Evolutionary psychology and the generation of culture.  New York: Oxford University Press.

Cosmides, L. & Tooby, J.  (2000).  The cognitive neuroscience of social reasoning.  In M. S. Gazzaniga (Ed.), The New Cognitive Neurosciences, Second Edition.  Cambridge, MA: MIT Press.  (Chapter 87, pp. 1259-1270.)

Fiddick, L., Cosmides, L., & Tooby, J.  (2000).  No interpretation without representation: The role of domain-specific representations and inferences in the Wason selection task.  Cognition, 77, 1-79.

Cosmides, L.  (1989).  The logic of social exchange: Has natural selection shaped how humans reason?  Studies with the Wason selection task.  Cognition, 31, 187-276.

Fiddick, L.  (1998).  The deal and the danger: An evolutionary analysis of deontic reasoning.  Doctoral dissertation, Dept. of Psychology, University of California, Santa Barbara.