HOMEWORK ASSIGNMENT #6: DUE IN CLASS ON FRIDAY, FEB. 20, AND ASSOCIATED DISCUSSION QUESTIONS

DISCUSSION QUESTIONS FOR WED., FEB. 18:

1. In Frank's theory what is the role of the moral emotions? Frank speaks of the emotions as "incentives". On Frank's theory, are emotions analogous to external incentives (e.g., punishment and other sanctions)? (To answer this question, you must consider his discussion of "the self-interest theory" on pp. 67-70, in addition to his presentation of his theory of the emotions earlier in the chapter.)

2. On Frank's account, is there any significant difference between overeating, due to an evolutionarily favored urge for heavy food intake, when food is abundant [p. 52], and being moral (for morality's sake)? If so, what is it? If not, why not?

3. Frank claims that there are two commitment problems solved by the moral sentiments [pp. 93-94]. What are the two problems and how are they solved by the moral sentiments?

DISCUSSION QUESTIONS FOR FRIDAY, FEB. 20:

1. The solution to one of Frank's problems depends on the reliability of clues to emotion. Which one? Why? Frank reports some preliminary experimental evidence which he claims supports his commitment model [pp. 139-143]. The question we will discuss in class is: How strongly does Frank's evidence support his commitment model.

2. To forestall objections based on calculation of expected utility of Cooperating and Defecting in his experiments, Frank states: "I should stress that the point of our experiments was not to see whether cooperators do better in prisoner's dilemma games. Indeed, the structure of the game is such that they necessarily do worse."(p. 143) Is this true of his experiment? Explain how his theory is at least compatible with cooperators doing better than defectors in his own experiment.

HOMEWORK ASSIGNMENT #6[Due in class on Friday, Feb. 20]: [This question continues from the preceding discussion question]: Frank continues: "The benefit to being a cooperator, if there is one, lies in being able to recognize other cooperators and interact selectively with them." Your assignment is to determine whether, on Frank's data and other reasonable assumptions, this benefit of being able to recognize other cooperators and to interact selectively with them would make the Expected Utility of being a Frankian Conditional Cooperator (FC) agent at least as great as the Expected Utility of an Individualistically Rational agent (IR). Calculate the expected utilities of FC and IR on the following assumptions: (1) They are involved in a two-stage n-person game. At the first stage, everyone tries to find an FC partner for the second stage of the game. (a) The probability that an FC agent will be correctly identified as an FC agent is 73/83. The probability that an FC agent will be incorrectly identified as an IR agent is 10/83. (b) The probability that an IR agent will be correctly identified as an IR agent is 15/39. The probability that an IR agent will be incorrectly identified as an FC agent is 24/39. [These probabilities are based on data taken from Frank's Table 7.2 on page 140.]

(2) At the second stage of the game, those identified as FC (whether correctly or not) interact in a one-shot PD with each other, and those identified as IR (whether correctly or not) interact in a one-shot PD with each other. Obviously, the pay-offs depend on whether or not one has been identified as FC or IR. (a) Agents Identified as IR. All agents identified as IR (whether correctly or not) are left to interact with each other, so they will all expect defection from the other, and will themselves defect (even if they are in fact FC agents). (b) Agents Identified as FC. (i) FC agents identified as FC. All FC agents identified as FC will Cooperate. Because of misidentifications, the probability that their partner is also FC and will also Cooperate is 73/97. The probability that an FC agent will be taken advantage of by an IR agent (mistakenly thought to be FC) who Defects is 24/97. (ii) IR agents identified as FC. The probability that an IR agent identified as FC will interact with a true FC agent (whom she will be able to take advantage of) is 73/97. The probability that an IR agent identifed as FC will interact with another mistakenly identified IR (in which case both Defect) is 24/97.

Using the above probabilities and the pay-offs from our standard PD matrix, calculate the Expected Utility of FC and IR. Do the calculations support Frank's claim that FC agents would do better than IR agents in the two-stage n-person game?