Original ArticlesHow “ought” exceeds but implies “can”: Description and encouragement in moral judgment
Introduction
Normative claims about what people “ought” to (or, more typically, “should”) do are common and important in human life. It is no surprise, then, that researchers in a variety of disciplines, including philosophy, have studied them extensively. This paper deals with a topic at the intersection of two central theoretical questions regarding “ought” claims.
The first question concerns the relationship between what people ought to do and what they are able to do. One view is that if someone ought to do something, then he is able to do it (e.g. Kant, 2009, Moore, 1959). This view is so common in philosophy that it even has a slogan: “ought implies can.” The contrary view is that a person’s ability does not always constrain what he ought to do. This view does not have a slogan but one possibility is “ought exceeds can.” The second question concerns the linguistic function of “ought” claims. One theory is that “ought” claims (purport to) describe normative facts about an agent’s responsibilities or obligations. A natural corollary of this view is that (sincere) “ought” claims express beliefs or knowledge, which represent those normative facts (e.g. Boyd, 1988, Huemer, 2005, Mackie, 1977). Another theory is that “ought” claims function as encouragement toward certain behavior. A natural corollary of this view is that “ought” claims express an emotion, command, preference, or other mental state that is not truth-evaluable, but which would be satisfied by the agent acting in the indicated way (e.g. Carnap, 1937, Gibbard, 1990, Stevenson, 1937). A third, hybrid theory is that “ought” claims function to both describe responsibilities and encourage their fulfillment (e.g. Barker, 2000, Hare, 1952, Stevenson, 1944).
In contemporary discussions, these two questions are related in an interesting way. More specifically, the hybrid theory of the function of “ought” claims might help explain how “ought implies can” is both wrong and right. (As far as I am aware, Sinnott-Armstrong, 1984 contains the first authoritative statement of this sort of proposal; see also Pigden, 1990.) Briefly, “ought implies can” is false because having a responsibility does not entail being able to fulfill it, but it seems correct because encouraging someone to fulfill a responsibility suggests a commitment to their being able to fulfill it. We can partially unpack that brief statement as follows:
On theoretical and empirical grounds, it has been persuasively argued that “ought implies can” is false when “implies” is understood as conceptual entailment. On theoretical grounds, it is false because it is inconsistent with the logic of excuses and the felicity of some apologies, threatens to trivialize many moral obligations, has sustained fruitless debates, and falsely entails that disabled people lack full moral agency (for a review, see Buckwalter, ms). On empirical grounds, in a wide range of experimental studies, the vast majority of competent adult speakers attribute moral responsibilities in tandem with inability to fulfill them (Buckwalter and Turri, 2014, Buckwalter and Turri, 2015, Chituc et al., 2016, Henne et al., 2016, Mizrahi, 2015, Turri, 2015, Turri, 2016, Turri, 2017, Turri, ms). Some of the clearest and most widely studied cases involve promises (Sinnott-Armstrong, 1984). Promising generates a responsibility that is not preempted or canceled by inability to fulfill it. If you promise to deliver a package by 4 pm, then you have a responsibility to deliver the package by 4 pm. If it turns out that you are unable to deliver the package on time, you still have a responsibility to do so—you ought to do something that you cannot do. Note that this pertains to “ought” in the descriptive sense of having a responsibility. Consider, by contrast, its function to encourage behavior. It would typically be pointless to encourage you to, say, deliver the package by 4 pm if it is acknowledged that you cannot do so—as many commentators have noted, it defies expectations, seems linguistically odd, or is otherwise infelicitous (see Pigden, 1990 for a rewarding and informed review of sources dating back over 500 years).
One hypothesis, then, is that “ought” statements can be interpreted as both descriptions of responsibilities and as encouragement toward their fulfillment. Their descriptive function is compatible with inability, but their encouragement function is not. When I describe you as having a responsibility to fulfill your promise, my remark is perfectly consistent with your being unable to do so. In this respect, “ought” exceeds “can.” But when I encourage you to fulfill your promise, the natural interpretation of my remark requires attributing to me the belief that you can do so. In this respect, “ought” implies “can.”
As noted above, this theory is not original with me. And while many contributors have recognized that empirical evidence is clearly relevant to understanding the function of linguistic expressions (e.g. Huemer, 2005, Pigden, 1990: 18), to date the empirical evidence has consisted principally of introspective and social observations. Introspective and social observation is a good place to start when studying familiar social activities, including the use of normative language. But we should not draw firm conclusions from reported patterns in a single researcher’s experience. The researcher’s experience might be idiosyncratic, or the researcher’s theoretical commitments might cause him or her to selectively keep track of, or misinterpret, observations. Similar remarks apply to a small group of researchers’ experience. One way to avoid these problems is to test observational claims against the judgments of a large sample of competent, theoretically-uncommitted language users experienced in the use of normative language. In this spirit, I conducted two experimental studies to begin rigorously testing the theory, described above, that “ought” exceeds but implies “can.”
More specifically, I conducted two experiments that begin testing two principal hypotheses implicated in the theory. The dual-function hypothesis states that “ought” claims have (at least) two functions: description and encouragement. The interaction hypothesis states that those two functions relate differently to an agent’s ability or inability to perform the relevant action. In the first experiment, participants read a story in which an agent promises to do something, and a speaker tells the agent that he “ought” to do what he promised. I manipulated whether the agent was able or unable to fulfill the promise. Participants recorded judgments about whether the speaker was describing an obligation that the agent had, whether the speaker was encouraging the agent to fulfill the obligation, and whether the agent was able to fulfill it. Three key questions here are whether people actually do interpret the speaker as both describing and encouraging, whether the ability/inability manipulation affects people’s interpretation of those two speech acts, and whether the manipulation differentially affects the interpretation of description compared to encouragement.
In the second experiment, participants read a story about an agent who promises to do something. I manipulated the time at which the agent was no longer able to fulfill the promise. Participants then constructed timeline graphs to represent the time in the story during which several things were true. Among these things were whether the agent has a responsibility to fulfill the promise, whether it is worthwhile to encourage the agent to fulfill the promise, and whether the agent is able to do so. The key questions here are whether people represent responsibilities as outlasting the ability to fulfill them, and whether they think it is worthwhile to encourage someone to do something after he is no longer able to do it.
The results of such studies could potentially support several different conclusions. On one hand, the results could undermine the dual-function hypothesis by demonstrating that people interpret “ought” claims as describing responsibilities but not as encouraging their fulfillment, or vice versa. On another hand, the results could support the dual-function hypothesis but undermine the interaction hypothesis. That is, it could turn out that people interpret “ought” claims as both describing responsibilities and encouraging their fulfillment, but both functions are similarly affected by the manipulation of ability/inability. For instance, people might view inability as equally irrelevant to whether responsibilities exist and whether it is worthwhile to encourage their fulfillment. On yet another hand, the results could support both the dual function hypothesis and the interaction hypothesis. In this case, people would view “ought” claims as having both functions but whereas inability is irrelevant to its descriptive function, it undercuts its encouragement function. It is only this combination of results that would strongly support the theory that “ought” exceeds but implies “can.”
I conducted one final experiment to provide further information regarding related theoretical issues about the function of normative language, in particular whether an “ought” claim has a primary function and how contextual features affect the interpretation of its functions. In this third experiment, participants again read a story about an agent who promises to do something and someone else remarks that this “ought” to be done. I manipulated whether the “ought” claim was made in the present or past tense, and whether it was addressed to the person who made the promise or to a third party. Participants recorded judgments about whether the speaker was describing a responsibility, encouraging its fulfillment, and casting blame. The key question here is whether any of the three functions is consistently present, thereby providing some evidence that it is a good candidate for an “ought” claim’s primary function.
Section snippets
Experiment 1
The purpose of this experiment is to investigate the potential for speakers of “ought” claims to be interpreted as describing moral obligations and as encouraging certain behavior, and whether the interpretation is affected by acknowledged ability or inability to produce the relevant outcome. Broadly speaking, one possible outcome is that there are no meaningful differences in an “ought” claim’s potential to function as a description or as encouragement. Both functions might always be
Experiment 2
The purpose of this experiment was to further investigate people views of ability, responsibility, and encouragement. Whereas Experiment 1 studied these issues in the context of interpreting a third-party’s speech, this experiment had participants assess matters directly and used a measure other than verbal reports. In order to do this, I asked participants to complete a timeline graphing task to represent the period during which certain propositions were true in a story.
Experiment 3
The purpose of this experiment is to investigate whether two additional factors affect an “ought” claim’s potential to function as describing, encouraging, or blaming. The factors are the claim’s tense (present or past) and the party to whom it is addressed (to the person who made the promise, or to a third party). Investigating this could be informative regarding an “ought” claim’s primary function.
General discussion
Three experiments advanced our understanding of the functions of normative language. People view claims about what an agent “ought” to do as having dual potential, namely, the potential to describe moral obligations and the potential to encourage agents to perform actions (Experiment 1). But these two functions relate differently to an agent’s ability or inability to perform. When an agent is unable to perform an action, people are significantly less likely to interpret someone saying he “ought
Acknowledgments
For helpful feedback and discussion, I thank Wesley Buckwalter, Ori Friedman, Walter Sinnott-Armstrong, Angelo Turri, and Sarah Turri. Thanks also to Steven Sloman and three anonymous reviewers for Cognition. This research was supported by the Social Sciences and Humanities Research Council of Canada, the Ontario Ministry of Economic Development and Innovation, and the Canada Research Chairs program.
References (26)
- et al.
Blame, not ability, impacts moral 'ought' judgments for impossible actions: Toward an empirical refutation of 'ought' implies “can.”
Cognition
(2016) Compatibilism can be natural
Consciousness and Cognition
(2017)Is value content a component of conventional implicature?
Analysis
(2000)How to be a moral realist
- Buckwalter, W., & Turri, J. (2014). Inability and obligation: Compelling counterexamples to “ought implies can”....
- Buckwalter, W. (ms). Theoretical refutation of “ought implies can”. University of...
- et al.
Inability and obligation in moral judgment
PLoS One
(2015) Philosophy and logical syntax
(1937)Wise choices, apt feelings
(1990)The language of morals
(1952)
An empirical refutation of ‘ought’ implies ‘can’
Analysis
Ethical intuitionism
Cited by (13)
Revisiting norms of assertion
2018, CognitionCitation Excerpt :This reflects a more general fact that people reliably distinguish what someone should not do from what he is blameless for doing. For instance, people reliably distinguish between broken promises that should have been kept and those that were blamelessly broken (Buckwalter & Turri, 2015; Chituc, Henne, Sinnott-Armstrong, & De Brigard, 2016; Turri, 2016c, 2017b). People also distinguish beliefs that should not be held from blameworthy beliefs (Turri, 2015).
Inability and obligation in intellectual evaluation
2020, EpistemeQualitative methods show that surveys misrepresent “ought implies can” judgments
2023, Philosophical PsychologyDirect and Derivative Moral Responsibility: An Overlooked Distinction in Experimental Philosophy
2022, Advances in Experimental Philosophy of Free Will and ResponsibilityChildren and Adults Donʼt Think They Are Free: A Skeptical Look at Agent Causationism
2022, Advances in Experimental Philosophy of Causation