[Editor’s Note: This article was originally published in The International Journal for Philosophy of Religion 17 (1985): 185-191. The page numbers below show the position of the text within that pagination scheme.]
________________________________________
185
The Coherence of the Hypothesis of an
Omnipotent, Omniscient, Free and Perfectly Evil Being (1985)
Michael Martin
Department of Philosophy, Boston University, 745 Commonwealth Avenue, Boston, MA 02215
Richard Swinburne has argued that the hypothesis of theism (h1) is a very simple hypothesis.[1] Because of its simplicity Swinburne maintains that h1 has a higher prior probability than rival hypotheses. This alleged higher prior probability is used by Swinburne in combination with other considerations to argue that h1 has a higher a posteriori probability than its rivals.
One rival hypothesis to h1 that is not explicitly considered by Swinburne is that there is an omniscient, omnipotent, free and perfectly evil Being. Let us call the hypothesis that such a Being exists h2. Although Swinburne does not explicitly consider h2 it is clear enough from what he does say that he would reject h2 as incoherent. He maintains that an omniscient, omnipotent and free Being would have to be morally perfect. In other words, although Swinburne does not put it exactly in this way his thesis is that moral perfection is not an independent attribute of God but is derivable for His other attributes.
Presumably Swinburne’s argument against h2 would then have to proceed as follows: Since h2 is incoherent, its prior probability is 0. Further, since no new evidence could raise the probability of a hypothesis whose prior probability is 0, the probability of h2 will remain 0 no matter what the evidence. Therefore, the a posteriori probability of h2 could never equal h1.
In this paper I will show that Swinburne’s reasons for supposing that h2 is incoherent are mistaken. As a result, there is no reason to suppose, as Swinburne does, that h2 is not as simple as h1. Consequently there is no reason to suppose that h2 has a lower prior probability than h1. If h1
has a higher a posteriori probability than h2, then this would have to be shown by showing that it has greater explanatory power than h2. Swinburne has not done this.
Swinburne’s argument
Swinburne argues that "if one takes a certain view about the status of moral judgements, God’s perfect goodness follows deductively from his omniscience and his perfect freedom".[2] The view Swinburne is referring to is that judgements that
________________________________________
186
some actions are morally good and others are morally bad are propositions which are true or false. He next maintains that if one knows that some moral action is good all things considered, then one has a reason for doing it. He says that in most cases, but not all, this knowledge generates a moral obligation to do A, that is it provides a morally obligatory reason to do A. (The cases in which it does not, Swinburne suggests, are ones in which it does not matter morally what one does.)
Thus God, who knows that some action is morally good all things considered and knows that it does matter morally what is done, has a morally obligatory reason to do that action. Now some human might have a morally obligatory reason to perform an act, according to Swinburne, but not do the action. This is because human beings are not completely free; they are sometimes influenced by non-rational factors. But by hypothesis this could not be a problem for God. God, by definition, is completely free from non-rational influences. Therefore, God must do actions that are morally obligatory since He has a morally obligatory reason for doing them.
Swinburne’s argument can be stated more formally as follows:
(1) P is omniscient. (By hypothesis)
(2) Act A is morally good all things considered. (By hypothesis)
(3) Act A is morally important. (By hypothesis)
(4) If P is omniscient then P knows everything that it is possible to know. (By definition)
(5) It is possible to know (2) and (3). (By hypothesis)
(6) P knows that act A is morally good all thing considered and that act A is morally important. [By (1), (2), (3), (4), (5)]
(7) If P knows that act A is morally good all things considered and that act A is morally important, then P has a morally obligatory reason to do A. (By definition)
(8) P has a morally obligatory reason to do A. [By (6) and (7)]
(9) If P has a morally obligatory reason to do A and P is completely free, P will do A if P can. (By definition)
(10) It is possible to do A. (By hypothesis)
(11) P is completely free. (By hypothesis)
(12) P is omnipotent. (By hypothesis)
(13) :. P will do A. [By (8), (9), (10), (11), (12)]
Critique of the argument
In order to evaluate Swinburne’s argument I will first analyze the notion of having a reason and its relation to personal explanations, a type of explanation that Swinburne believes is crucial in theological contexts. In particular it will be shown that having a reason has two aspects (an intentional and a belief aspect corresponding to different parts of a personal explanation). Using this analysis I argue that it is
________________________________________
187
possible to give a personal explanation of why someone does something evil by saying the person did the action knowing it was evil and because it was evil. I argue that although such explanations in ordinary life are implausible, because of our background beliefs and a principle of charity we operate with they are not incoherent as Swinburne thinks. In particular, I argue that they are not impossible in theological contexts. If theism is to be taken seriously, then, I argue that the hypothesis of an omniscient, omnipotent, free and perfectly evil Being must be taken seriously. Prima facie, I suggest that such a hypothesis has an a priori and posteriori probability equal to theism. I show finally exactly where Swinburne’s argument goes wrong and why Swinburne’s argument rests on a fatal ambiguity between the intentional and belief aspect of having a reason. Once this ambiguity is made explicit, the argument is shown to have false premises.
Two aspects of having a reasonand personal explanations
To have a reason for an action could and usually does involve two closely related but analytically distinguishable aspects: the belief aspect and the intentional aspect. If someone asks me why I raised my hand, I might answer by saying "To get the speaker’s attention." This answer assumes two things which are part of my reason. It assumes that I believe that raising my hand will get or at least may get the speaker’s attention and it assumes that I intended to get the speaker’s attention by raising my hand. It would make little sense in normal circumstances to say either "I raised my hand to get the speaker’s attention but I did not believe that raising my hand was a way to get his attention" or "I raised my hand to get the speaker’s attention but I did not intend to get the speaker’s attention."
This same distinction between the belief aspect and the intentional aspect of having a reason occurs in cases in which moral considerations are at issue. Someone who cites as her reason for doing some act that it was morally obligatory for her assumes that doing the act fulfills her moral obligation and that it was her intention to fulfill her moral obligation by doing this act.
In some contexts the belief aspect is obvious and hardly needs mentioning, so only the intentional aspect is called the reason for the action. In other contexts the opposite is true. The intentional aspect is obvious and what is called the reason for the action is the relevant belief. We can distinguish these two aspects of having a reason by using subscripts. ReasonB will refer to the belief aspect; reasonI to the intentional aspect.
In his writing Swinburne distinguishes a type of explanation he calls personal[3] from scientific causal explanation. Although Swinburne does not make this explicit, there is a close connection between his analysis of reasons and his concept of a personal explanation.
A personal explanation of a basic action E, e.g. raising one’s hand, according to Swinburne, has three aspects: a rational agent P, the agent’s intention J, and the agent’s basic power X to bring about E. Thus, to explain E one cites P’s intention
________________________________________
188
J to raise his arm and his power X to do so. When the action is not basic but mediated, the story is a little more complicated. One cites, of course, actor P, his intention J to get the speaker’s attention by raising his arm, and his basic power X to raise his arm. But according to Swinburne’s account one must also cite the connection between the basic action of raising one’s arm and getting the speaker’s attention.
The relation between this account of personal explanation and the giving of reasons is this: The agent’s reason for raising his arm consists of both his intention to get the speaker’s attention (reasonI) and the belief (usually not explicitly stated) that there is a connection between raising one’s arm and getting the speaker’s attention in certain contexts (reasonB). This relation between personal explanation and reason also holds in moral contexts. One explains why P gives to charity by citing the agent’s intention to fulfill his moral obligation, his power to give money, and the connection between giving to charity and fulfilling one’s obligation. The reason for the action was both the intention to fulfill one’s moral obligation (reasonI) and the belief that there is a connection between giving to charity and fulfilling one’s moral obligation (reasonB).
Personal explanations in terms of evil intentions and correct beliefs
An actor can act out of a variety of moral, immoral and non moral reasons. Thus an actor can give to charity not in order to fulfill his moral obligation but because of a non moral reason. He may intend to raise his prestige in the community (reasonI) and believe that giving to charity would do so (reasonB). He might even give to charity from completely immoral reasons. He might intend to undermine the self reliance in the community (reasonI) and believe (perhaps wrongly) that giving to charity will do this (reasonB).
Conversely, it would seem prima facie that someone could do an immoral act with moral, non moral or immoral reasonsI. One could immorally kill one’s brother in order to save one’s own life (reasonI), believing that killing him would do this (reasonB) and yet take into account no moral considerations.
The crucial question is whether the following type of personal explanation is possible and consequently whether giving the following reasons are possible. Suppose Jones kills his brother. He believes that he has a moral obligation not to kill his brother (reasonB); he has the power to kill his brother; he has the intention to do what is not his moral obligation to do; that is to do what is morally wrong (reasonI). We also stipulate that Jones is not under any duress or under the influence of any irrational impulses. His choice to kill his brother is completely free. I believe that although such an explanation is unusual it is not incoherent. In such a case (and if this is typical of Jones’ behavior) one might say that Jones is an evil person, not out of ignorance but in a basic and fundamental way. He intends to do evil knowing that it is evil because it is evil.
In ordinary contexts the above explanation of Jones’ behavior would not be the explanation that one would tend to give. Indeed, there is initially an implausibility
________________________________________
189
to such an explanation for two reasons. First, there are inductive grounds to suppose in general that people, who do immoral acts do them out of ignorance or out of irrational impulses. Second, people operate with a principle of charity.[4] One attempts initially to explain the immoral action of others in ways that preserves the purity of their intentions (reasonI) at the expense of their beliefs (reasonB). This charity goes hand in hand with our moral attitude of respect for persons. It is difficult to respect someone who acts out of a basically immoral intention–the intention to do evil knowing it is evil, because it is evil.
Nevertheless, I believe that although this principle does guide our thoughts it at best provides an initial orientation to the formulation of personal explanations and that evidence could be imagined that would induce us to say that someone was basically evil; that he or she did an evil act, knowing it was evil because it was evil. Such a person would not be irrational; rather he or she would simply have different reasons from ours. If there were not such a basically evil person, this would be a contingent fact, not a logical or conceptual necessity.
One might argue that if confronted with such a basically evil person, one would be confronting an essentially alien creature. Perhaps, but such a creature would not be so alien that its action would be mysterious and unintelligible. (Compare a creature that was non logical, i.e. did not operate with the law of identity or excluded middle.) On the other hand, others might contend that such an essentially evil creature is not so alien after all. It may be argued that most of us have had basically evil intentions, i.e. intentions to do something evil, knowing it is evil because it is evil. It is only external factors, e.g. fear of punishment or conditioning, that prevent these intentions from being fulfilled. However, a being that is perfectly free would have no such external factors preventing it from doing evil.
Considerations so far suggest the possibility of a Being that is omnipotent, omniscient, free and perfectly evil; that is, they suggest the possibility of hypothesis h2 mentioned above. Such a Being would know what is morally obligatory (his reasonB would be correct); he would have the power to do what is morally obligatory, he would be completely free to act but because of his reasonI he would do what is morally wrong. Such a Being would not be irrational but would operate with different reasonsI.
We will not call such a Being “the Devil”, since this term has different connotations from those I have in mind. Let us call this Being “The Absolute Evil One”. Just as God is morally perfect in the sense that He always does what is morally correct, the Absolute Evil One is perfectly evil in the sense that he always does what is morally wrong.
It is possible that no one has in fact ever believed in the Absolute Evil One. Nevertheless, it would have to be shown that h2 would be less probable than h1, the hypothesis of theism. It would have to be shown, in particular, that h2 explains less than h1 or that h2 has a lower prior probability than h1.
However, h2 would seem to explain the existence of design in the world and the existence of the universe as well as h1 does. Given h2 there would, of course, be a problem of explaining the existence of good, but there is a problem of explaining
________________________________________
190
evil if one assumes h1. However, traditional defenses such as the free will defense are open to believers in the Absolute Evil One. For example, it might be argued that a world in which humans have free will would be worse over all that a world without it.
Further, the prior probability of h2 would seem equal to h1, since, according to Swinburne, the prior probability of h1 is based primarily on the simplicity of h1 which in turn is based on purely logical and tautological considerations. The simplicity of h1 and h2 are identical with respect to all crucial factors. They both postulate an omnipotent, omniscient and completely free Being. These three attributes are, according to Swinburne, simpler than any finite attributes of power, knowledge or freedom. Further, there does not seem to be a difference between the simplicity of a Being that is perfectly evil and a Being that is perfectly good other things being equal.
Swinburne’s argument reconsidered
So far I have argued that h2, the hypothesis of the Absolute Evil One, is not incoherent despite what Swinburne claims. I have admitted that in ordinary contexts a hypothesis that postulates that someone freely does something evil knowing it is evil because it is evil should not be the hypothesis of choice. This in part explains why Swinburne’s argument has a certain plausibility. But it does not explain it completely. There is another reason why Swinburne’s argument seems plausible: It rests on an ambiguity that becomes clear once one tries to pinpoint exactly where Swinburne’s argument goes wrong in the formal reconstruction of his argument presented above. Let us reconsider premise (7).
(7) If P knows that Act A is morally good all things considered and that Act A is morally important, then P has a morally obligatory reason to do A.
This premise, according to Swinburne, is necessarily true. But in view of the above analysis of having a reason, it can be seen that this premise is ambiguous. It could mean:
(7′) If P knows that Act A is morally good all things considered and that act A is morally important, then P has a morally obligatory reasonI to do A.
Interpreted in this way (7′) is not necessarily true; it is in fact false. P could know that Act A is morally good all things considered and that Act A is morally important yet not have the intention to do what is morally obligatory. Indeed, the Absolute Evil One would not have this intention. On the other hand, (7) could be interpreted in the following way:
(7″) If P knows that act A is morally good all things considered
________________________________________
191
and that Act A is morally important, then P has a morally obligatory reasonB to do A.
This premise is true given what Swinburne seems to mean by “morally obligatory”. But although (7″) is true on this interpretation, this interpretation affects the truth of (9). Now (9) becomes:
(9′) If P has morally obligatory reasonsB to do A and P is completely free, P will do A if P can.
(9′) is false if P is the Absolute Evil One. For the Absolute Evil One has an obligatory reasonB to do A but does not have a reasonI to do A.
Conclusion
I have argued that the hypothesis h1, the existence of The Absolute Evil One, a Being that is omnipotent, omniscient free and perfectly evil is coherent if theism is. So if Swinburne’s argument is accepted, then the simplicity of h1 and h2 are the same and consequently the prior probability is the same. Whether h1 and h2 have the same a posteriori probability will depend on their respective explanatory power. Prima facie they would seem to be equally explanatory. However, this would have to be shown by more detailed argument. The upshot is that Swinburne’s implicit a priori dismissal of the possibility of The Absolute Evil One must be reconsidered. Consequently, his probabilistic argument for the existence of God must be reconsidered since his probabilistic argument implicitly assumes that The Absolute Evil One is impossible.
Notes
[1] Richard Swinburne, The Existence of God (Oxford: Oxford University Press, 1979), Chapter 5: See also Richard Swinburne, The Coherence of Theism (Oxford: Oxford University Press, 1977), Chapter 11.
[2] Swinburne, The Existence of God, p. 97.
[3] Ibid., Chapter 2.
[4] This principle of charity should not be confused with another principle of charity mentioned by Swinburne (see Ibid., p. 56).