Abstract
When presented with a situation involving an agent’s choice between alternative actions, a moral oracle says what the agent is allowed to do. The oracle bases her advice on some moral theory, but the nature of that theory is not known by us. The moral oracle’s test consists in determining whether a series of questions to the oracle can be so constructed that her answers will reveal which of two given types of theories she adheres to. The test can be applied to moral theories in order to determine if they differ in their recommendations for action. Based on this test, a terminology is developed to specify different forms and degrees of distinguishability between moral theories, or types of theories, in terms of their recommendations for action. In conclusion, the test is applied to consequentialism and utilitarianism.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
The moral oracle’s test has obvious similarities with two well-known thought experiments, the Turing test and John Searle’s Chinese room. It differs from these in focusing on the principles from which the answer was derived rather than on the nature of the entity that provided the answer. In the Turing test we want to know whether the answer was given by a human being and in the Chinese room whether it was given by someone who knows Chinese. In the moral oracle’s test we want to know from what moral theory the answers are derived, not who performed the derivation.
Moral recommendations and moral permissions are interchangeable under the conventional assumption that obligations and permissions are interdefinable so that the question “May I X?” receives the answer “Yes” if and only if the question “Ought I non-X?” receives the answer “No”.
In accordance with the Guardian’s admonition, questions to the oracle should not be about moral theory. Therefore, situations and action alternatives should be specified in non-moral terms, not in terms such as for instance “an action that produces more good than any of its alternatives”.
This formulation assumes that the moral theories applied by the oracle are fully determinate. If we wish to allow for moral theories that leave it open for the Oracle to answer some questions with either “Yes” or “No”, then the following formulation can be used instead: “T1 is potentially distinguishable from T2 if and only if there is a series of questions to the moral oracle and a series of answers that she can make to these questions if she applies a moral theory in T1 but not if she applies a moral theory in T2.”
To my knowledge, the concepts of potential and absolute distinguishability are not available in the previous literature on moral theories. It is hoped that they can be used to provide more precise answers to questions on the relations between moral theories in terms of their practical recommendations.
To make this precise, let A(t,q) be the set of series of answers that a moral oracle applying the theory t can give to the series q of questions. For a set T of moral theories, let A(T,q) = ∪{A(t,q)|t∈T}. Then T 1 is potentially distinguishable from T 2 if and only if (∃q)¬(A(T 1,q) ⊆ A(T 2,q)) and it is absolutely distinguishable from T 2 if and only if (∃q)(A(T 1,q) ∩ A(T 2,q) = ∅).
Moral requirements and permissions can also differ in their strength or stringency. (See Hansson 2001, p. 132 and the references given there.) It is in principle possible for two moral theories to disallow and allow exactly the same actions but differ in the strength (stringency) of some of their prohibitions and permissions. This can be dealt with by introducing question-phrases representing these different degrees, e.g. “Is it morally recommended that I…?” and “Is it morally strongly demanded that I…?”.
Oddie and Milne (1991) showed that from a strictly formal point of view, the normative contents of any moral theory coincide with those of some agent-neutral theory. However, such an agent-neutral theory may have to be unlimitedly complex, and impossible to specify beforehand in a way that covers unforeseen situations. It is therefore doubtful whether it can at all be called a moral theory.—Arguably there are agent-bound consequences that any agent-relative theory will be sensitive to, such as being the person who kills another, innocent person. Under this assumption, agent-relativity is not only potentially but also absolutely distinguishable.
However, it does not seem to be absolutely distinguishable. For that to be the case, we would need to find a pair of examples differing only in terms of action-accompanying mental states, such that any motives-regarding theory will give different recommendations in the two cases. I have not been able to find any such example. Arguably, any sensible such theory will evaluate killing a severely wounded person out of hatred for that person as worse than killing the same person out of misconceived mercy. However, this difference does not necessarily lead to different recommendations for action.
A theory type yields a distinction in the moral oracle’s test if and only if either the theory type or its complement is potentially distinguishable in the test.
Let A be a set of alternatives, and let c(A) be the set of chosen (recommended) elements of A. Then a set V of variables can be so constructed such that for each element a k of c(A) there is a variable v k ∈ V for which v k (a k ) = 1 and v k (a m ) = 0 for all a m ∈ A\{a k }. Then c(A) is the set of elements a of A for which \( {\displaystyle \sum_{v_k\in V}{v}_k(a)} \) is maximal.
References
Caws P (1995) Minimal consequentialism. Philosophy 70:313–339
Dreier J (1993) Structures of normative theories. Monist 76:22–40
Hamlin AP (1989) Rights, indirect utilitarianism, and contractarianism. Econ Philos 5:167–187
Hansson SO (2001) The structure of values and norms. Cambridge University Press, Cambridge
Hansson SO (2006a) Condensed examples in philosophy. Theoria 72(2):97–99
Hansson SO (2006b) The modal status of philosophy. Theoria 72(3):173–176
Kupperman JJ (1980) Vulgar consequentialism. Mind 89:321–337
Lenman J (2000) Consequentialism and cluelessness. Philos Public Aff 29:342–370
Louise J (2004) Relativity of value and the consequentialist umbrella. Philosl Q 54:518–536
Oddie G, Milne P (1991) Act and value: expectation and the representability of moral theories. Theoria 57:42–76
Pettit P (1988) The consequentialist can recognize rights. Philos Q 38:42–55
Portmore DW (2007) Consequentializing moral theories. Pac Philos Q 88:39–73
Scanlon TM (1998) What we owe to each other. Belknap, Cambridge
Schroeder M (2006) Not so promising after all: evaluator-relative teleology and common-sense morality. Pac Philos Q 87:348–356
Sen A (1979) Utilitarianism and welfarism. J Philos 76(9):463–489
Sen A (1983) Evaluator-relativity and consequential evaluation. Philos Public Aff 12:113–132
Sen A (2000) Consequential evaluation and practical reason. J Philos 97:477–502
Skorupski J (1995) Agent-neutrality, consequentialism, utilitarianism… a terminological note. Utilitas 7:49–54
Sosa D (1993) Consequences of consequentialism. Mind 102:101–122
Spielthenner G (2005) Consequentialism or deontology? Philosophia 33(217–235):2005
Zamir T (2001) One consequence of consequentialism: morality and overdetermination. Erkenntnis 55:15–168
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Hansson, S.O. The Moral Oracle’s Test. Ethic Theory Moral Prac 17, 643–651 (2014). https://doi.org/10.1007/s10677-014-9500-5
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10677-014-9500-5