Reed Richter

This dissertation primarily consists of three related, published articles on decision theory. All three articles are at least in part a response to various problems and issues raised by an approach known as "causal decision theory". But "Rationality, Group Choice and Expected Utility" goes beyond the issue of causal versus non-causal decision theory, and deals in general with the subject of analysing rational choice in terms of expected utility. I develop a new way of handling cases of rational cooperation within the framework of a straight-forward expected utility analysis. This new approach is employed to resolve The Prisoners' Dilemma and other paradoxical cases of group choice. Both "Counterfactuals and Newcomb's Paradox" and "Rationality Revisited" challenge the claim that causal decision theory adequately captures our notion of rational choice. In particular, both articles take issue with the causal theorist's approach to the infamous Newcomb's Paradox, since that is generally regarded as a pivotal case. It is important to note that the two articles individually provide quite independent arguments for the claim that non-causal decision theorists should not be persuaded by the causal theorist's account of Newcomb's Paradox. Both "Rationality, Group Choice and Expected Utility" and "Rationality Revisited" develop the surprising finding that there is a clear sense in which it is not necessarily true that rationality requires an agent to choose the option that he can determine to have the highest payoff. It is the failure to realize this that leads one astray in Newcomb's Paradox.

In discussions surrounding epistemology and rationality, it is often useful to assume an agent is rational or ideally rational. Often, this ideal rationality assumption is spelled out along the following lines: 1. The agent believes everything about a situation which the evidence entitles her to believe and nothing which it does not. 2. The agent believes all the logical consequences of any of her beliefs. 3. The agent knows her own mind: if she believes P, she believes that she believes P; and if she doesn't believe P, she believes that she doesn't believe P. 4. The agent believes nothing of the form 'P and it is not the case that P.' 5. If an agent's background belief-set satisfies 1-4 and if rationality requires the agent to add P to her belief-set, then the resulting belief-set will also satisfy 1-4. While individually plausible, there are cases in which holding on to 1-5 generates paradox. Some resolve such paradoxical cases by granting 1-5 but arguing while ideally rational agents can exist, they can't possibly ever find themselves in such a situation: such case descriptions are epistemically incoherent. Others allow that rational agents can coherently find themselves in such odd circumstances, and argue that it's more reasonable to weaken our concept of ideal rationality and give up premise (2) above. However, this strategy has also been rejected. My aim in this paper is to defend the utility of positing an ideally rational agent in such paradoxical circumstances. I argue in such cases we should give up (2), in particular the assumption that (necessarily) if an ideally rational agent believes both P and the conditional, if P then Q, then she believes Q.. What's important is to hold on to the goal of positing ideal rationality: to maximize the amount of true or probably true information a thinker can justifiably believe in a given circumstance. Normally that will mean holding on to (2), but these unusual paradoxical cases are best handled by giving up (2).

This paper looks at a dispute decision theory about how best to characterize expected utility maximization and express the logic of rational choice. Where A1, … , An are actions open to some particular agent, and S1, … , Sn are mutually exclusive states of the world such that the agent knows at least one of which obtains, does the logic of rational choice require an agent to consider the conditional probability of choice Ai given that some state Si obtains, Prob(Ai/Si). Or, is the logic of choice better represented by considering the probability of the counterfactual if Ai then Si, Prob(Ai ⟥-> Si). Causal decision theory, developed by Allan Gibbard, William Harper, and David Lewis defend the counterfactual analysis; whereas, Richard Jeffrey and others defend the conditional probability analysis, evidential decision theory. I argue that the problems posed by cases of decision instability favor evidential decision theory.

This is an unpublished talk written for a meeting of French philosophers. The paper describes the evolution versus creationism/intelligent design controversy in the U.S. A number of philosophers and scientists try to resolve this issue by sharply distinguishing the realm of science versus any talk of the supernatural. These pro-evolutionists often appeal to science's essential commitment to "methodological naturalism," the view that scientific methodology is essentially committed to naturalism and cannot meaningfully entertain hypotheses concerning the supernatural. I criticize methodological naturalism, suggesting that such an appeal is misguided and counterproductive. I suggest an alternative view of the supernatural consistent with scientific knowledge.

Writers, philosophers, and theologians have oft made the comparison between being a mature human being and a masterpiece work of art or design. Employing the analogy between the creation of artistic value and the creation of full-fledged human value, this paper stakes out a middle ground between pro-choice and pro-life by considering a more general account of value and the relationship between being a potential X and a mature implementation of X's potential. I argue that the value of a potential X is a function of a number of factors, most importantly, what I call the "accessibility relation" between a potential X and a full-fledged instantiation of this potential. The value is as much intrinsic to the “seed” as to some future implementation of the seed’s potential. This approach inclines even a secular humanist to reasonably confer a significant degree of moral value to a human conceptus, and even more to an early term fetus.

Michael Philips’ ‘Racist Acts and Racist Humor’ attempts to analyze the ethics of racism. At the heart of his discussion is the view that… “racist” is used in its logically primary sense when it is attributed to actions. All other uses of “racist” … must be understood directly or indirectly in relation to this one. Accordingly, racist beliefs are beliefs about an ethnic group used to “justify” racist acts, racist feelings are feelings about an ethnic group that typically give rise to such acts, and racist epithets are the stings and arrows by means of which certain such acts are carried out. Books and films are said to be racist … if they perpetuate and stimulate racist beliefs or feelings.

This paper rejects a view of science called "methodological naturalism." According to many defenders of mainstream science and Darwinian evolution, anti-evolution critics--creationists and intelligent design proponents--are conceptually and epistemologically confusing science and religion, a supernatural view of world. These defenders of evolution contend that doing science requires adhering to a methodology that is strictly and essentially naturalistic: science is essentially committed to "methodological naturalism" and assumes that all the phenomena it investigates are entirely natural and consistent with the laws of physics. Thus encountering any unexplained phenomenon, science assumes a priori that there is some natural cause and will only test a natural hypothesis. Since by definition supernatural causes are assumed to be not subject to the constraints of physical or natural law as understood by science, supernatural hypotheses and explanations must be banned from proper science. Science simply can't say that God did, or did not do it. I argue that the success of science is directly relevant to rational belief in supernatural causes, and that in fact science can and does say in particular cases that "God didn't do it." I suggest that pro-evolution proponents can better defend science and the theory of evolution by rejecting methodological naturalism.

The Hasting Center's, "Guidelines on the Termination of Life-Sustaining Treatment and the Care of the Dying" (1987), outlines a position on assisted suicide that I argue is contradictory. On one hand the guidelines offers a position on human dignity and autonomy that accords competent patients the right to intentionally kill themselves by requesting doctors to terminate life-support. Yet, on the other hand, the guidelines argue that terminating life-support upon request is not ever the moral equivalent of doctored-assisted suicide, and granting the right to the former does not grant a right to the latter. In this paper I argue that in some circumstances a right to obligate your doctor to pull the plug on life-support is morally equivalent to the right to obligate your doctor to assist you in dying by more direct means, e.g. administering a lethal dose of medication. If the fundamental values of autonomy, self-determination, and human dignity justify this right in the one case, they eo ipso justify this right in the other. Both circumstances are subject to the same potential abuse, so citing such potential abuse does not defeat granting an autonomy right in either case.

This paper proposes a view uniformly extending expected utility calculations to both individual and group choice contexts. Three related cases illustrate the problems inherent in applying expected utility to group choices. However, these problems do not essentially depend upon the tact that more than one agent is involved. I devise a modified strategy allowing the application of expected utility calculations to these otherwise problematic cases. One case, however, apparently leads to contradiction. But recognizing the falsity of proposition (1) below allows the resolution of the contradiction, and also allows my modified strategy to resolve otherwise paradoxical cases of group choice such as the Prisoners' Dilemma: -/- (1) lf an agent x knows options A and B are both available, and x knows that were he to do A he would be better off (in every respect) than were he to do B, then doing A is more rational for x than doing B.

In their development of causal decision theory, Allan Gibbard and William Harper advocate a particular method for calculating the expected utility of an action, a method based upon the probabilities of certain counterfactuals. Gibbard and Harper then employ their method to support a two-box solution to Newcomb’s paradox. This paper argues against some of Gibbard and Harper’s key claims concerning the truth-values and probabilities of counterfactuals involved in expected utility calculations, thereby disputing their analysis of Newcomb’s Paradox. If we are right, then Gibbard and Harper’s method of calculating expected utility does not adequately represent rational choice.