Decision Making In Prisoner’s Dilemma

12.5 Discussion

Ad Friendliness:
We can conclude that friendliness is positively related to (a) subject’s preliminary willingness to cooperate, as well as to (b) their expectations about their opponent’s willingness to cooperate, also friendlier subjects (c) cooperated more in single-move games and in iterated games against Bully and Tit for Tat. These results are intuitively comprehensible. We suppose, that “friendly” people (tend to) cooperate more because (i) they are more altruistic, and/or because (ii) they expect others to be relatively more altruistic/cooperative (this expectation forms the basis for one’s own cooperation, since mutually beneficial cooperation ensues).
Ad (i): Palfrey & Rosenthal, 1988; and McKelvey & Palfrey, 1992 found that subject’s own altruistic tendencies or attitudes enhance their cooperation.
Ad (ii): Former research confirmed that people in various situations sometimes tend to expect more cooperation from others than merited by facts (for example Camerer & Weigelt, 1988 experimented with game of loan contracts and found that people expected slightly higher repayment than was merited by information available to them).
Majolo et al. (2006) found that people played more cooperatively with their friends then with strangers. The amount of cooperation with strangers was 69% (61% in the second experiment), the amount of cooperation with friends was 86% (87% in the second experiment), results differed statistically in both games (p < 0,05). So, in relation to our results, we can also hypothesize (iii) that the more “friendly” a person is, the bigger proportion of people encountered she considers to be her actual or potential “friends”, and hence she behaves more cooperatively towards a bigger proportion of people. The underlying reason of bigger cooperativeness towards friends can be for example bigger “shadow of future” (new encounters are expected), bigger “shadow of past” (past cooperation that should be reciprocated), higher (and mutual) trust evolved, and/or higher (and mutual) altruism developed.
Previous research confirmed significance of variables at least partly similar to our “friendliness” and “dominance”. Terhune (1968) hypothesized that “affiliation subjects”, when playing together, will be mainly cooperative and that “power subjects” when playing together will be mainly conflictive. Need for affiliation was defined as the need or desire to establish and maintain friendly relations with others; while the need for power was defined as the need or desire to gain and exert control over others. That means that these variables are akin to our “friendliness” and “dominance” variables. In Terhune’s research the motives/personality characteristics of subjects were assessed by trained coders using TAT. Terhune’s rather complex experimentation and analyses can be summarized in the following way: affiliative subjects decreased cooperation when defending themselves against possible defection, while power subjects decreased cooperation when trying to exploit the opponent; so the motivational/personality characteristics studied indeed influenced subjects’ decision making (even though not in the sense of simply increasing or decreasing the overall amount of cooperation, or in the sense of Terhune’s initial hypotheses).
Kuhlman & Marshello (1975) found significant differences among “cooperative” and “competitive” subjects (questionnaires were used to assign subjects to these categories, not all subjects fell in either of these categories). “Cooperative” subjects were significantly (p < 0,1) more cooperative when playing against Tit for Tat and against ALL C than “competitive” subjects. (The categories of competitive, cooperative, and individualist players were studied by Messick & McClintock, 1968; McClintock & McNeel, 1966. Kuhlman et al., 1992 studied expectations of these types of players and discovered that players “projected” their own motives onto their opponents, see section 5.22.)
Note: Psychopaths behaved more competitively and had higher gains in a 40-trials Prisoner’s Dilemma game against Tit for Two Tats strategy (Mokros et al., 2008). Higher rate of competitive choices was significantly linked to the total score of Psychopathy Personality Inventory–Revised and two of its subscales: rebellious nonconformity, and Machiavellian egocentricity. (The subjects in this study were inpatients from two high-security psychiatric hospitals.) The rate of cooperation was 99,5% in the comparison group of non-psychopaths, and 89% in the psychopaths group, the groups differed significantly (p < 0,05). These rates of cooperation are strikingly high: even “psychopaths” were more cooperative than our subjects, who cooperated in 67,5% when playing against the Benevolent strategy (Benevolent is only slightly less punitive than Tit for Two Tats, but practically identical; of course several conditions, such as the experimental setting, the framing of the game situation, or the pay-off matrix differed in our experiment and in that by Mokros et al., 2008; so no precise comparison of the results is possible).
Ad tolerance for ambiguity:
We found that subjects relatively more tolerant for ambiguity cooperated relatively more in single-move games and against Bully. This finding corresponds very well with the fact that (a) single-move games involve relatively higher degree of uncertainty (players have no knowledge about their opponents), and that (b) Bully strategy mixes cooperation and defection in an ambiguous, somewhat unsettling fashion (Bully is both systematically cooperative, and uncooperative, and retaliatory – its description is given section 8.44), and tolerance for ambiguity can help overcome this uncertainty and this ambiguity and maintain cooperation.
Pilisuk et al. (1965) found that tolerance for ambiguity predicted higher cooperation in the iterated Prisoner’s Dilemma game. Tolerance for ambiguity was measured using a six-point Likert-type scale containing items about preference for regularity, clarity, or ambiguity, balance, or asymmetry. Also, Pilisuk sees the ability to reframe old perceptions
“… [as] a part of tolerance for ambiguity. The finding would seem consistent with the conclusion that a cognitive recasting which occurs during the moves and countermoves of an interpersonal conflict is necessary for the self-organization of the two parties into a single cohesive unit” (Pilisuk et al., 1965, p. 505).
Ad dominance:
In our study we found statistically significant relation only between “56” and “47”: β* = 0,44, p < 0,01. This result probably does not carry any real significance, since no statistically significant relation was found between similar measures (“52” and “54”) and “47”.
Ad risk avoidance:
In our study we found statistically significant relation only between “45” a “56”, β* = 0,35, p < 0,05. This relation also might not carry any real significance.
Neither Pilisuk et al. (1965), nor Peter de Heus et al. (2010) found any relation between attitude towards risk and decision making in the Prisoner’s Dilemma game. This might be caused by the fact that, as Peter de Heus et al. (2010) note, it is unclear what constitutes the risky option in the Prisoner’s Dilemma game. In general, however, we can state that defection (D) in an iterated game eliminates the possibility of S outcome in the present move (decreased risk of loss), but possibly lowers the probability of R and also T outcomes in the future (increased risk of loss), and vice versa for cooperation (C). This general “rule” is of course contingent on the kind of opponent/strategy against which we play.