Поиск по сайту




Пишите нам: info@ethology.ru

Follow etholog on Twitter

Система Orphus

Новости
Библиотека
Видео
Разное
Кросс-культурный метод
Старые форумы
Рекомендуем
Не в тему

список статей


Does an altruist-detection cognitive mechanism function independently of a cheater-detection cognitive mechanism? Studies using Wason selection tasks

Ryo Odaa, Kai Hiraishib, Akiko Matsumoto-Odac

1. Introduction

2. Experiments 1 and 2

2.1. Methods

2.1.1. Experiment 1: subjects and procedures

2.1.2. Experiment 2: Subjects and procedures

2.1.3. Data analyses

2.2. Results

2.2.1. Card selection patterns in Experiment 1

2.2.2. Card selection patterns in Experiment 2

2.2.3. Associations between each selection pattern in Experiment 1

2.2.4. Associations between each selection pattern in Experiment 2

2.3. Discussion

3. Experiment 3

3.1. Methods

3.2. Results

3.2.1. Card selection patterns in Experiment 3

3.2.2. Associations between each selection pattern in Experiment 3

3.3. Discussion

4. General discussion

Acknowledgment

Appendix A. Altruist-detection Wason selection task

Appendix B. Social contract Wason selection task

Appendix C. Sharing-rule Wason selection task

Appendix D. Precaution Wason selection task

References

Copyright

1. Introduction

One dominant hypothesis about the social brain proposes that adaptation to a social environment is important for the evolution of human cognition and intelligence (e.g., Alexander, 1974, Byrne & Whiten, 1988, Dunbar, 1998). Gigerenzer (1997) claimed that the challenge for testing social intelligence was “to design the possible mechanisms of social intelligence and to test these by means of experiment, observation and simulation,” proposing that social intelligence has a modular organization and suggesting that a cheater-detection mechanism is one modular candidate.

Reciprocal exchange is an essential and universal feature of human social life (Trivers, 1971). To maintain reciprocal exchange, it is important to detect cheaters who receive benefits without paying any costs and exclude them from social relationships. Cosmides (1989) and Cosmides and Tooby (1992) used the Wason selection task to find evidence of a “Darwinian algorithm” for cheater detection. In a standard descriptive Wason selection task, respondents were asked whether a conditional statement in the form of “if P, then Q” had been violated by any of four instances (P, not-P, Q and not-Q), represented by four cards (Wason, 1968). Typically, fewer than 10% of subjects answer the descriptive selection tasks correctly. It has been argued that when the conditional only describes the factual relationship between P and Q cases, the performances on the task becomes poor regardless of the concreteness of the conditional (Cosmides, 1989, Cosmides & Tooby, 1992, Manktelow & Over, 1990). On the other hand, it has been reported that when the nature of the conditional is deontic rather than descriptive or indicative, respondents change their performance radically. This is called the “thematic content effect.” Cosmides (1989) constructed a social contract version for which many subjects selected the logically correct “benefit” (P) and “no-cost” (not-Q) cards when the rule was “if you take the benefit, then you must pay the cost.”

When individuals engage in reciprocal exchanges, however, complex social interactions make it difficult to detect cheating; this is called the problem of “subtle cheating.” Brown and Moore (2000) proposed altruist detection as a possible mechanism to solve the problem of subtle cheating. They also used the Wason selection task to demonstrate a specialized ability for altruist detection. People were sensitive to information about individuals who help others without seeking credit. While research proponents of the cheater-detection mechanism and that of the altruist-detection mechanism have argued that the two algorithms are domain-specific Darwinian mental modules, it is not clear whether they are functionally independent. Brown and Moore (2000) proposed the possibility that the cheater-detection mechanism solves the altruist-detection problem. If individuals can activate the cheater-detection mechanism, they could also use this mechanism to detect individuals who are definitely not cheaters, i.e., altruists. There may, however, be two independent mechanisms, as Brown and Moore (2000) argued: one for cheater detection and another for altruist detection. Indeed, there is neuroimaging data indicating that detecting others worthy of your trust may be distinct from detecting cheaters (Singer, Kiebel, Winston, Dolan, & Frith, 2004). Researchers, however, have not determined possible commonalities between cheater detection and altruist detection.

There is another candidate argument about social intelligence; Hiraishi and Hasegawa (2001) argued that humans have specialized mental mechanisms that address problems arising from collective action. Specifically, they constructed a Wason task with a sharing rule: if one receives a resource, one is an “in-group” member. Taking the resource provider's perspective, participants tried to detect “out-group” members (not-Q) taking an undeserved resource (P). Changing the perspective to a resource recipient, participants tried to detect another type of cheating: failure to share (not-P) with in-group members (Q), which was revealed by the selection of all four cards. These results suggested that participants were sensitive to cheaters who exploited or did not contribute to collective action. Again, it is not clear whether cheater detection and the other two types of detection have independent, associated or identical mental mechanisms.

Another thematic content effect can be found in deontic conditionals with precautions. The rule is “if in a hazardous situation, take precautions” and has no direct relation to any social context. This is yet another candidate domain-specific mechanism because harm avoidance is important for individual survival (Cosmides & Tooby, 1992, Fiddick, 2003, Stone et al., 2002). Moreover, insofar as there may be hazards associated with social interactions, precautionary mechanisms may be activated.

In two experiments, we asked subjects to complete four deontic reasoning tasks and examined the associations between each card selection pattern. We also used the descriptive version of the task to monitor logical performance in the absence of any context. First, we used reasoning tasks couched in practical everyday situations. We switched the rules of some tasks because card selection patterns could transfer from one task to another if the logically correct selection pattern was P and not-Q in all tasks. Then, we did a third experiment using less-contextualized deontic reasoning tasks, which had deontic conditionals and a rather simplified description of social situations.

2. Experiments 1 and 2

2.1. Methods

2.1.1. Experiment 1: subjects and procedures

A total of 171 Japanese university students (122 males and 49 females; mean age=20.0±2.5 years) participated in Experiment 1. The study was done in several classes, and students received no monetary reward for their involvement. Each subject received a booklet containing four deontic reasoning Wason selection tasks [altruist-detection, social contract (cheater detection), sharing-rule and precaution tasks] as well as the descriptive Wason selection task. All participants first performed the four deontic reasoning tasks, followed by the descriptive version. The deontic reasoning tasks were presented in random order to avoid possible effects of presentation order.

Table 1 shows the details of all five tasks; the descriptive Wason task version served as a control to determine the frequency of correct choices in the absence of deontic content. In logical terms, the correct answers for an if P, then Q statement were P and not-Q. In our problem, using letters and numbers (K=P, 5=Q, D=not-P and 9=not-Q), the logically correct choices were K and 9.

Table 1.

Details of the rules and interpretations of card selection patterns in three experiments

Rule Selection pattern
P and not-Q Not-P and Q
Experiment 1
Altruist detection
If X helps, then X seeks credit. Altruist-sensitive
Social contract
If X pays cost, then X gets benefit. Cheater-sensitive
Sharing-rule
If X is in-group, then X gets share. In-group equality Out-group exploitation
Precaution
If X takes precaution, then X is in hazardous situation. Precaution-sensitive
Descriptive
If K, then 5. Logically correct
Experiment 2
Altruist detection
If X seeks credit, then X helps. Altruist-sensitive
Social contract
If X gets benefit, then X pays cost. Cheater-sensitive
Sharing-rule
If X gets share, then X is in-group. Out-group exploitation In-group equality
Precaution
If X is in hazardous situation, then X takes precaution. Precaution-sensitive
Descriptive
If K, then 5. Logically correct
Experiment 3
Altruist detection
If X seeks credit, then X helps. Altruist-sensitive
Social contract
If X pays cost, then X gets benefit. Cheater-sensitive
Sharing-rule
If X gets share, then X is in-group. Out-group exploitation In-group equality
Precaution
If X is in hazardous situation, then X takes precaution. Precaution-sensitive
Descriptive
If K, then 5. Logically correct

The altruist-detection version of the Wason task required subjects to imagine needing a friend who would help them adjust to a new career without taking advantage of them. A likely candidate would be a person who volunteered in the community for the sake of helping others rather than for reward, such as a day off work. The rule was that “if they volunteer, then they take a day off during the week.” The four card choices were “seeks a day off during the week,” “does not seek a day off during the week,” “volunteers” and “does not volunteer” (Appendix A).

The social contract (cheater detection) version required subjects to imagine helping with registration at an academic conference. Participants were asked to pay a registration fee if they chose to attend the party. Persons who paid the fee would place a sticker on their name badges. Subjects were asked to determine whether or not individuals were breaking the rule “if they put a sticker on their badges and then joined the party.” The four cards were “joins the party,” “does not join the party,” “puts on a sticker,” and “does not put on a sticker” (Appendix B).

The sharing-rule problem required participants to imagine that they went to a stadium to cheer for their university's baseball team (X university). A team member was providing orange juice to fans saying, “if you are from X university, you get a glass of juice.” Participants were asked to detect individuals who did not follow this rule. The four cards were “gets juice,” “does not get juice,” “belongs to X university,” and “belongs to Y university” (Appendix C).

For the precaution-rule problem, participants were asked to imagine that they had experienced an earthquake, during which their houses collapsed. They want to enter their homes to recover their belongings, but it is dangerous because of the aftershocks. Subjects were asked to detect individuals who did not obey the rule “if one puts on a helmet, then one enters a house.” The four cards were “enters a house,” “does not enter a house,” “puts on a helmet,” and “does not put on a helmet” (Appendix D).

2.1.2. Experiment 2: Subjects and procedures

A total of 255 Japanese university students (205 males and 50 females; mean age=20.4±2.3 years) participated in Experiment 2. The procedure was the same as in Experiment 1, except that we reversed the rules in deontic tasks (Table 1).

2.1.3. Data analyses

Expanding on the work of Brown and Moore (2000), we used McNemar's test to compare card selection patterns in the deontic reasoning tasks and the descriptive task. To distinguish between McNemar's chi-square and regular chi-square tests, we used an “M” prefix to denote McNemar's test. When McNemar's test could not be applied because of small numbers, we used a binominal test. The deontic reasoning tasks were predicted to produce either P and not-Q selection only or not-P and Q selection only (Table 1). In addition, on the sharing-rule problem, there was a possibility that a respondent was sensitive both to the exploitation by an out-group member and to the equality of in-group members. This should have led to the selection of all four cards. Therefore, in order to test associations among selection patterns in the deontic tasks, we classified the card selection patterns into four categories: only P and not-Q, only not-P and Q, all four and other selection patterns. Then, we excluded the subjects who performed “other selection patterns” in all five tasks. Using this procedure, we could prevent a possible bias on the association studies from subjects who could not understand the structure of the questions in reasoning tasks at all. For the remaining subjects, we used Fisher's exact probability test and phi coefficient to test associations between any two deontic tasks. If there was an association between performances on any two tasks, Fisher's test should turn out to be significant, and the phi coefficient would become larger. To examine associations, we divided the selection patterns in the sharing-rule task to in-group equality and out-group exclusion.

2.2. Results

2.2.1. Card selection patterns in Experiment 1

Selection patterns for the four deontic reasoning tasks showed the predicted thematic content effects (Table 2). On the descriptive task, only 7.0% of subjects selected P and not-Q cards (logically correct answer). The percentages were 38.6% on the altruist-detection task and 15.8% on the sharing-rule task, both of which were significantly higher than that on the descriptive task (descriptive vs. altruist-detection, Mχ2(1)=40.13, p<.001; descriptive vs. sharing-rule, Mχ2(1)=6.32, p<.05).

Table 2.

Percentages of selection patterns in each task with statistical comparisons against the “descriptive” (content-free) task

Selection task
Descriptive Altruist-detection Social contract (cheater-detection) Sharing-rule Precaution
Experiment 1
P and not-Q 7.0 38.6⁎⁎ 7.0 15.8 7.6
Not-P and Q 1.8 2.9 43.9⁎⁎ 14.6⁎⁎ 45.6⁎⁎
All four 11.7 1.2 4.1 12.9 2.3
Others 79.5 57.3 45.0 56.7 44.4
Experiment 2
P and not-Q 6.7 6.3 40.4⁎⁎ 23.1⁎⁎ 42.0⁎⁎
Not-P and Q 0.8 23.1⁎⁎ 4.3 9.0⁎⁎ 5.5
All four 11.8 0.0 0.8 2.7⁎⁎ 0.0
Others 80.8 70.6 54.5 65.1 52.5
Experiment 3
P and not-Q 8.6 16.7 4.4 43.9⁎⁎ 61.4⁎⁎
Not-P and Q 2.2 17.2⁎⁎ 42.8⁎⁎ 1.7 2.8
All four 8.9 1.7 2.2 1.9⁎⁎ 0.0
Others 80.3 64.4 50.6 52.5 35.8

<0.05 in McNemar's test and binominal test comparing with the descriptive task.

⁎⁎

<0.01 in McNemar's test and binominal test comparing with the descriptive task.

The percentage of subjects who chose not-P and Q cards was only 1.8% on the descriptive task. In contrast, the percentages were 43.9% on the social contract task, 14.6% on the sharing-rule task and 45.6% on the precaution task. All of them were significantly higher than that on the descriptive task [descriptive vs. social contract, Mχ2(1)=66.33, p<.001; descriptive vs. sharing-rule, Mχ2(1)=15.75, p<.001; descriptive vs. precaution, Mχ2(1)=71.12, p<.001]. There was no significant change in the frequency of subjects selecting all four cards between the descriptive task and the sharing-rule task (11.7% and 12.9%, respectively).

2.2.2. Card selection patterns in Experiment 2

Generally, the predicted thematic content effect was observed in Experiment 2 as well. On the descriptive task, the percentage of subjects who selected logically correct P and not-Q cards was 6.7% (Table 2). The percentages were 40.4% on the social contract task, 23.1% on the sharing-rule task and 42.0% on the precaution task [descriptive vs. social contract, Mχ2(1)=72.25, p<.001; descriptive vs. sharing-rule, Mχ2(1)=30.02, p<.001; descriptive vs. precaution, Mχ2(1)=77.66, p<.001].

The percentage of subjects who selected not-P and Q cards on the descriptive task was 0.8%. There was a significant increase in not-P and Q selection on the altruist-detection task [23.1%, Mχ2(1)=51.41, p<.001] and on the sharing-rule task [9.0%, Mχ2(1)=16.00, p<.001]. Contrary to the prediction, however, there was a significant decrease in the number of people who selected all four cards in the sharing-rule task [descriptive, 11.8% vs. sharing-rule, 2.7%, Mχ2(1)=13.83, p<.001].

2.2.3. Associations between each selection pattern in Experiment 1

After excluding the subjects who performed “other selection patterns” in all the tasks, 138 subjects (102 males and 36 females; mean age=19.8±2.2 years) remained.

We analyzed associations among the selection patterns for the different deontic tasks; for example, we tested whether those who responded in the predicted selection pattern on the altruist-detection task (P and not-Q) were more likely to respond according to the predicted selection pattern during the social contract task (not-P and Q).

We found no significant associations between performance on altruist detection and that of the other tasks, except for out-group exclusion (Table 3). The phi coefficient indicated that subjects who could detect altruists tended to show low performance in finding exploitation by out-group members. Performances on the social contract task were associated with both out-group exclusion and precaution. Subjects who could detect cheaters tended to select out-group members who received a share. They also showed high performance on the precaution task. Performances on the precaution task were associated with out-group exclusion, but the phi coefficient was relatively low.

Table 3.

Phi coefficients between each selection pattern in the deontic reasoning tasks

Altruist Cheater In-group Out-group
Experiment 1
Altruist detection
Cheater detection −0.08
In-group equality −0.07 −0.10
Out-group exclusion −0.26⁎⁎ 0.24⁎⁎
Precaution-sensitive −0.10 0.37⁎⁎ −0.05 0.19
Experiment 2
Altruist detection
Cheater detection −0.14
In-group equality −0.08 −0.05
Out-group exclusion −0.16 0.13
Precaution-sensitive −0.24⁎⁎ 0.47⁎⁎⁎ −0.05 0.43⁎⁎⁎
Experiment 3
Altruist detection
Cheater detection 0.08
In-group equality −0.02 −0.02
Out-group exclusion −0.01 0.32⁎⁎⁎
Precaution-sensitive −0.05 0.04 −0.10 0.09

p<.05.

⁎⁎

p<.01.

⁎⁎⁎

p<.001.

2.2.4. Associations between each selection pattern in Experiment 2

After excluding the subjects who performed “other selection patterns” in all the tasks, 189 subjects (153 males and 36 females; mean age=20.4±2.4 years) remained.

We found no significant associations between performance during altruist detection, cheater detection and in-group equality (Table 3). However, there were significant associations between performances during altruist detection and out-group exclusion, as well as the precaution task. The phi coefficients indicated that subjects who could detect altruists tended to show low performance in detecting exploitation by out-group members, and they were not good at finding people who did not take precautions in hazardous situations. Subjects who could detect cheaters tended to show high performance in the precaution task. Performances in the precaution task were also positively associated with out-group exclusion.

2.3. Discussion

The four deontic reasoning tasks we employed were appropriate to test sensitivity to another party being an altruist, a cheater, willing to share a resource and following a precaution rule in the context of danger because they showed the predicted thematic content effects. The thematic content effects observed in Experiments 1 and 2 were not very strong and on some deontic reasoning tasks, the percentages of respondents who did not respond in the predicted manner approached 50%. However, given that we did not use the word must nor may in the conditional statements and that we did not stress the importance of finding violation to the rules in the context, the percentages were understandable (Platt & Griggs, 1993).

In both experiments, we found no significant associations between performances during Wason selection tasks that required cheater detection and during those that required altruist detection. This result is not sufficient to conclude that these two mechanisms work independently but counters the hypothesis that altruist detection is achieved through a cheater-detection mechanism. We also found negative associations between performances during altruist detection and out-group exclusion, positive associations between performances during cheater detection and precaution tasks as well as positive associations between performances during out-group exclusion and precaution tasks. In these two experiments, altruist-detection card patterns did not correspond to out-group exclusion, while card patterns of cheater detection did correspond to out-group exclusion as well as to precaution. These facts suggest that correspondences of card patterns could have affected the associations between performances in each task.

The precaution task was not expected to reflect a social context. On adaptationist grounds, it seems to be inexplicable that the precaution task was found to be associated with tasks dealing with social contexts. Indeed, a recent neuroimaging study suggested that reasoning about social contracts and reasoning about precaution were neurologically dissociated (Fiddick, Spampinato, & Grafman, 2005).

A possible reason for this association is that the precaution task employed in this study activated the cheater-detection algorithm; the precaution task rule we employed was “if one puts on a helmet, then one may enter a house.” Although entering a collapsed house is a hazardous situation, it also could be interpreted as a benefit because an individual could collect his or her belongings. Moreover, there is a possibility that individuals interpreted putting on a helmet as a contract, so participants might have interpreted the precaution task as a type of social contract task. To avoid this confusion, we performed another experiment using less-contextualized versions of deontic reasoning tasks. Moreover, we changed correspondences of rules from those in Experiments 1 and 2 to consider the possible effects of rule correspondence on associations (Table 1).

3. Experiment 3

3.1. Methods

A total of 360 Japanese university students (148 males and 212 females; mean age=19.9±1.7 years) participated in Experiment 3. The procedure was the same as in Experiments 1 and 2, except that we used less-contextualized deontic selection tasks that had deontic conditionals and a rather simplified description of social situations. Fiddick (2003) had used the same kind of tasks as “abstract deontic versions of the Wason selection task.” The altruist-detection task featured the rule, “if one attains a profit V, then one performs an act A that benefits others,” and the subject was asked “to find a friend” who would be the beneficiary. The social contract version featured the rule, “if one pays a cost C, then one takes a benefit B,” and the subject was asked to find a person who is breaking the rule.

In the sharing-rule task, subjects were asked to find the persons who did not follow the rule, “if one receives a share B, then one is in group X.” The precaution task featured the rule, “if one performs a dangerous act D, then one must take precaution P,” and the subject was asked to find a person who was placed in a hazardous situation by inadvertently ignoring the rule.

Data analyses also followed Experiments 1 and 2.

3.2. Results

3.2.1. Card selection patterns in Experiment 3

The predicted thematic content effect was observed in Experiment 3 as well (Table 2). The percentage of subjects who selected logically correct P and not-Q cards on the descriptive task was 8.6%. The percentages were 43.9% on the sharing-rule task and 61.4% on the precaution task, both of which were significantly higher than on the descriptive task [descriptive vs. sharing-rule, Mχ2(1)=99.9, p<.001; descriptive vs. precaution, Mχ2(1)=176.84, p<.001].

The percentage of subjects who selected not-P and Q cards on the descriptive task was 2.2%. There were significant increases in not-P and Q selection on the altruist-detection task [17.2%, Mχ2(1)=41.3, p<.001] and on the social contract task [42.8%, Mχ2(1)=156.5, p<.001]. The percentage of not-P and Q subjects on the sharing-rule task was not significantly different from that on the descriptive task (1.7%, binominal test: p=.77). As in Experiment 2, the percentage of subjects who selected all four cards for the descriptive task was significantly higher than for the sharing-rule task [descriptive, 8.9% vs. sharing-rule, 1.9%, Mχ2(1)=16.5, p<.001].

3.2.2. Associations between each selection pattern in Experiment 3

After excluding the subjects who performed “other selection patterns” in all tasks, 284 subjects (121 males and 163 females; mean age=19.9±1.6 years) remained. We found no significant associations between performance during altruist detection and the other tasks (Table 3). There were significant associations between performances during cheater detection and out-group exclusion. The phi coefficient indicated that subjects who could detect cheaters were good at detecting exploitation by out-group members. Performances in the precaution task were not associated with any other tasks.

3.3. Discussion

Using a less-contextualized deontic version of the selection task, we found no significant association between performance on the precaution task and on tasks featuring social contexts. This supports our hypothesis that some subjects interpreted the precaution task as a social contract task in Experiments 1 and 2. Our results indicate a content sensitivity in human reasoning at the performance level.

Even though altruist-detection card patterns corresponded to cheater-detection card patterns in Experiment 3 (Table 1), there was no significant association between the two performances, suggesting an altruist-detection mechanism independent of the cheater-detection mechanism. We found a significant positive association between cheater detection and out-group exclusion. In Experiment 3, the cheater-detection card pattern did not correspond to the out-group exclusion card pattern. Moreover, a significant association was also found in Experiment 1, in which the two card selection patterns corresponded to each other. In addition, we observed a weak association in Experiment 2. These results indicate that subjects who performed well during the social contract task were also sensitive to exploitation by out-group members during the sharing-rule task.

4. General discussion

In all three experiments, there was no significant association between altruist detection and cheater detection. An altruist-detection mechanism may have evolved as an adaptation to a different social domain than the one in which cheater-detection evolved. Theoretical analysis supports this conclusion: while both mechanisms function in social contexts, the altruist-detection mechanism would be useful when an individual is forming a relationship of social exchange with another person with whom a relationship has yet to be established, while the cheater-detection mechanism could help maintain an already established relationship of social exchange. These two mechanisms seem to work during separate periods of social exchange, which may explain why there was no significant association between them. Chang and Wilson (2004) studied the effects of affective recollection on performance of reasoning tasks; their results also supported independence of these mechanisms. They reported that individuals who recalled being cheated performed better during a cheater-detection reasoning task than those who recalled happier experiences. On an altruist-detection reasoning task, however, they found that emotion had no effect.

Another common result to our experiments was that subjects who performed well during the cheater-detection task were also sensitive to exploitation by out-group members. In order for the cheater-detection algorithm to work effectively, individuals need to know whether others are paying a cost or receiving benefits. This requires relatively close relationships among individuals; open social relationships and exchange of resources with unspecified persons could easily allow the required information to be concealed or distorted. Researchers, however, have suggested that individuals who value close social relationships are more likely to be sensitive to exploitation by out-group members (Hiraishi, Ando, Ono, & Hasegawa, 2004). If this is correct, then individuals who live in close social relationships may be more sensitive both to cheaters and to exploitation by out-group members, which may explain the associations of performance during cheater-detection and sharing-rule tasks.

A simple alternative explanation might also be possible. Fiddick (2004) indicated that people associate different emotions with violations of social contracts and precautions: people tend to associate anger with violation of social contracts and fear with violation of precaution rules. Chang and Wilson (2004) also demonstrated that negative emotions facilitate cheater detection. Unfortunately, there are no studies that examined emotion and sharing-rule tasks. It is plausible, however, that people feel anger or other similar emotion when they find exploitation of resources by out-group members. If this is the case, the common emotion of anger might explain the association between cheater detection and out-group exploitation.

Acknowledgments

We thank Drs. Laurence Fiddick and William Michael Brown for their meaningful comments. This study was supported by a research activation fund from Nagoya Institute of Technology.

Appendix A. Altruist-detection Wason selection task

Imagine that you take a job with a company. Although you are excited about future career opportunities, you worry about finding friends who will help you in business. You would like to have close friends who would not take advantage of you in the workplace or in your personal life.

The company that hired you encourages employees to volunteer. For example, on weekends, some employees participate in activities such as community cleanup or helping handicapped persons; however, since few employees are willing to volunteer, the company institutes a rule: if an employee volunteers on the weekend, then he or she can take a day off.

So, you consider coworkers who do not follow the rules below acceptable as friends:


“If they volunteer, then they take a day off during the week.” (Experiment 1)

or

“If they take a day off during the week, then they volunteer.” (Experiment 2)

The four cards below display information about four coworkers. One side of each card displays information about whether or not they volunteer, and the other side displays information about whether or not they take a day off.

Indicate only the card(s) that you definitely need to turn over to determine if any of your coworkers are potential friends. Although you can choose more than one card, all cards should meet the minimum requirements.


1.Seeks a day off during the week

2.Does not seek a day off during the week

3.Volunteers

4.Does not volunteer

Appendix B. Social contract Wason selection task

Imagine yourself helping with registration at an academic conference. A party is held during the conference and participants are asked to pay if they want to attend; however, there are many conference participants, and some of them are sneaking into the party without paying. So, the conference organizer requests that participants who have already paid put a sticker on their badges, and makes a rule:


“If they put a sticker on their badges, then they join the party.” (Experiment 1)

or

“If they join the party, then they put a sticker on their badges.” (Experiment 2)

You are asked to check whether participants follow the rule. The four cards below display information about four participants. One side of each card displays information about whether or not they put a sticker on their badges, and the other side of each card displays information about whether or not they join the party.

Indicate only the card(s) that you definitely need to turn over to determine if the participants follow the rule. Although you can choose more than one card, all cards should meet minimum requirements.


1.Joins the party

2.Does not join the party

3.Puts on a sticker

4.Does not put on a sticker

Appendix C. Sharing-rule Wason selection task

Imagine that you go to a stadium to cheer for your university's baseball team (X university) playing a match against Y university. A member of the X university team has brought orange juice and is providing it to fans from X university. He asks you to check who receives the juice. That is, there is a rule:


“If they are from X university, they get a glass of juice.” (Experiment 1)

or

“If they get a glass of juice, then they are from X university.” (Experiment 2)

You are asked to check whether fans follow the rule. The four cards below display information about four persons. One side of each card displays information about whether or not they get a glass of orange juice, and the other side of each card displays information about their university.

Indicate only the card(s) that you definitely need to turn over to determine if the individuals follow the rule. Although you can choose more than one card, all cards should meet minimum requirements.


1.Gets juice

2.Does not get juice

3.Belongs to X university

4.Belongs to Y university

Appendix D. Precaution Wason selection task

Imagine that you experience an earthquake and take shelter. Most of the houses in your area have collapsed due to the strong quake. You and other victims decide to return to your homes to recover your belongings; however, entering houses is dangerous because of aftershocks. Therefore, authorities decide that everyone who enters a house must obey the rule:


“If they put on a helmet, then they enter a house.” (Experiment 1)

or

“If they enter a house, then they put on a helmet.” (Experiment 2)

You are asked to check whether victims obey the rule. The four cards below display information about four persons. One side of each card displays information about whether or not they put on a helmet, and the other side of each card displays information about whether or not they enter a house.

Indicate only the card(s) that you definitely need to turn over to determine if the persons follow the rule. Although you can choose more than one card, all cards should meet the minimum requirements.


1.Enters a house

2.Does not enter a house

3.Puts on a helmet

4.Does not put on a helmet

References

Alexander, 1974 1.Alexander RD. The evolution of social behavior. Annual Review of Ecology and Systematics. 1974;5:326–383.

Brown & Moore, 2000 2.Brown WM, Moore C. Is prospective altruist-detection an evolved solution to the adaptive problem of subtle cheating in cooperative ventures? Supportive evidence using the Wason selection task. Evolution and Human Behavior. 2000;21:25–37.

Byrne & Whiten, 1988 3.Byrne R, Whiten A. Machiavellian intelligence: Social expertise and the evolution of intellect in monkeys, apes and humans. Oxford: Clarendon Press; 1988;.

Chang & Wilson, 2004 4.Chang A, Wilson M. Recalling emotional experiences affects performance on reasoning problems. Evolution and Human Behavior. 2004;25:267–276.

Cosmides, 1989 5.Cosmides L. The logic of social exchange: Has natural selection shaped how humans reason? Studies with the Wason selection task. Cognition. 1989;31:187–276. MEDLINE | CrossRef

Cosmides & Tooby, 1992 6.Cosmides L, Tooby J. Cognitive adaptations for social exchange. In:  Barkow JH,  Cosmides L,  Tooby J editor. The adapted mind: Evolutionary psychology and the generation of culture. Oxford: Oxford University Press; 1992;p. 163–228.

Dunbar, 1998 7.Dunbar RIM. The social brain hypothesis. Evolutionary Anthropology. 1998;6:178–190.

Fiddick, 2003 8.Fiddick L. Is there a faculty of deontic reasoning? A critical re-evaluation of abstract deontic versions of the Wason selection task. In:  Over DE editors. Evolution and the psychology of thinking: The debate. Sussex: Psychology Press; 2003;p. 33–60.

Fiddick, 2004 9.Fiddick L. Domains of deontic reasoning: Resolving the discrepancy between the cognitive and moral reasoning literatures. The Quarterly Journal of Experimental Psychology. 2004;57:447–474.

Fiddick et al., 2005 10.Fiddick L, Spampinato MV, Grafman J. Social contracts and precautions activate different neurological systems: An fMRI investigation of deontic reasoning. Neuroimage. 2005;28:778–786. MEDLINE

Gigerenzer, 1997 11.Gigerenzer G. The modularity of social intelligence. In:  Whiten A,  Byrne RW editor. Machiavellian intelligence II: Extensions and evaluations. Cambridge: Cambridge University Press; 1997;p. 264–288.

Hiraishi et al., 2004 12.Hiraishi K, Ando J, Ono Y, Hasegawa T. General trust and individual differences on the Wason selection task with sharing-rule. Psychologia. 2004;47:226–237.

Hiraishi & Hasegawa, 2001 13.Hiraishi K, Hasegawa T. Sharing-rule and detection of free-riders in cooperative groups: Evolutionarily important deontic reasoning in the Wason selection task. Thinking and Reasoning. 2001;7:255–294.

Manktelow & Over, 1990 14.Manktelow K, Over D. Deontic thought and the selection task. In:  Gilhooly K,  Keane M,  Logie R,  Erdos G editor. Lines of thinking: Reflections of the psychology of thought. London: Wiley; 1990;p. 153–164.

Platt & Griggs, 1993 15.Platt RD, Griggs RA. Darwinian algorithms and the Wason selection task: A factorial analysis of social contract selection task problems. Cognition. 1993;48:163–192. MEDLINE | CrossRef

Singer et al., 2004 16.Singer T, Kiebel SJ, Winston JS, Dolan RJ, Frith CD. Brain responses to the acquired moral status of faces. Neuron. 2004;41:653–662. MEDLINE | CrossRef

Stone et al., 2002 17.Stone VE, Cosmides L, Tooby J, Kroll N, Knight RT. Selective impairment of reasoning about social exchange in a patient with bilateral limbic system damage. Proceedings of the National Academy of Science. 2002;99:11531–11536.

Trivers, 1971 18.Trivers RL. The evolution of reciprocal altruism. Quarterly Review of Biology. 1971;46:35–57. CrossRef

Wason, 1968 19.Wason PC. Reasoning about a rule. Quarterly Journal of Experimental Psychology. 1968;20:273–281. CrossRef

a Graduate School of Engineering, Nagoya Institute of Technology, Nagoya 466-8555, Japan

b Graduate School of Arts and Sciences, The University of Tokyo, Tokyo 153-8902, Japan

c Department of Welfare and Culture, Okinawa University, Naha 902-8521, Japan

Corresponding author. Tel.: +81 52 735 5112; fax: +81 52 735 5112.

PII: S1090-5138(06)00018-3

doi:10.1016/j.evolhumbehav.2006.03.002



2007:12:08