Bermúdez describes the extensionality principle – that “preferences, values, and decisions should be unaffected by how outcomes are framed” – as receiving “almost unanimous acceptance” (target article, sect. 2.1), and even as being “almost unquestioned” (target article, sect. 1, para. 2). These claims might come as a surprise to philosophers who work on agency and ethics. In the language of moral philosophy, it sounds as if almost all decision theorists are consequentialists.
Moral theories are conventionally categorised as being either consequentialist, deontological, or virtue ethical. According to consequentialists, what fundamentally make an action morally good or bad are its consequences. According to deontologists, what fundamentally matter are instead the rules or principles which are embodied in an action. According to virtue ethicists, what fundamentally matter are instead the traits of character embodied in the action.
It might seem natural for a consequentialist to accept the extensionality principle. The facts of how an agent frames the outcomes of an action are facts about that agent before they are facts about those outcomes. A perfectly rational consequentialist agent would only be guided by facts about outcomes. In that sense, there is no place for frames in the rational ideal of a consequentialist.
There is a place for frames, however, in the rational ideals of deontologists and virtue ethicists. The principles and the character traits which matter on their views seem able to be thought of as framing outcomes. The same act can be good if it is thought of under the guise of one principle or trait, and bad if it is thought of under another, despite the outcome or the consequences being the same.
The most famous deontological moral philosopher is Kant. Kant believed that we must only act on principles or maxims which we could consent to anyone anywhere following. When we find ourselves unable to give that consent, one of the options open to us is to try to come up with a more nuanced version of that unacceptable principle. For example, I may propose to myself that I pursue a career in forestry just because I want to. The principle I need to be able to consent to is that of anyone pursuing a career in forestry should they want to. This principle is too simple, however: I would not consent to it in a situation in which everyone wanted to pursue a career in forestry. However, I might be able to consent to a principle of pursuing a career in forestry in times when, simultaneously, foresters are needed, I have the skills and means to become one, and I want to become one. In being guided by this more complex principle, I act just as I would have on the simpler one. The outcome is the same: I pursue a career in forestry. It is only an acceptable, preferable outcome, however, when it is chosen as framed by the more complex principle. In this way, Kantian moral theory can violate the extensionality principle.
Bermúdez's (target article, sect. 3.3.1) example of Agamemnon's dilemma might be reimagined in Kantian terms. If Agamemnon can consent to the principle of following Artemis's will in even this case, he can consent it universally. At the same time, he finds himself unable to consent to the principle of murdering his daughter, even though his act would be the same were he following Artemis's will or murdering his daughter.
This dilemma might be phrased in virtue ethical terms, too. Agamemnon is caught between the frames of faithfulness to his gods and of faithfulness to his family. In the Republic, Plato has Socrates imagine a situation in which you have borrowed weapons from a friend, but that friend has since gone mad (Plato, 2012, p. 8/331c). Honouring your debts requires you to return those weapons when your friend demands them. Handing weapons to a madman, though, would be viciously reckless. Honouring your debts is preferable to keeping the weapons from him, but keeping the weapons from him is preferable to recklessly giving them back.
Korsgaard (Reference Korsgaard1996, p. 108, Reference Korsgaard2009, p. 15) uses this example as part of an argument that what agents must assess and choose between are what she terms “actions” rather than “acts.” An action incorporates the principle or end for the sake of which the act is done. Returning the weapons to your friend is an act. Returning the weapons to your friend because they are the property of your friend, and they have asked you to return them, is an action.
Korsgaard reads both Kant and Aristotle, the philosopher most famous as a virtue ethicist, as subscribing to a theory of agency on which actions are our objects of choice and assessment. Bermúdez also appears to subscribe to something like this picture: He suggests that “there is no such thing as making choices over a purely extensional opportunity set, independent of any way of describing or framing the things in it” (target article, sect. 3.1, para. 5). In this respect and in respect of his belief in the coherence of quasi-cyclical preferences, Bermúdez would seem able to claim powerful allies beyond the disciplinary pale of decision theory.
I do not want to overstate the common ground: Bermúdez is working with a different conception of rationality to these philosophers, if one which is different by being less substantive and demanding. There might be empirical reasons to treat moral quasi-cyclical preferences as being special cases, or even not to think of moral reasoning in terms of frames and preferences. In addition, some of these philosophers might be unwilling to admit the possibility of our being unable to resolve conflicts between frames by subsuming them under higher frames. As Aristotle (2012, p. 36/1226b) puts it in the Eudemian Ethics, “those who have no aim before them are not liable to deliberate.” If we are to choose between the urgings of two frames, we must possess some ground on which that choice can be made. The question can only be whether to think of that ground in terms of a higher frame.
You have
Access
Bermúdez describes the extensionality principle – that “preferences, values, and decisions should be unaffected by how outcomes are framed” – as receiving “almost unanimous acceptance” (target article, sect. 2.1), and even as being “almost unquestioned” (target article, sect. 1, para. 2). These claims might come as a surprise to philosophers who work on agency and ethics. In the language of moral philosophy, it sounds as if almost all decision theorists are consequentialists.
Moral theories are conventionally categorised as being either consequentialist, deontological, or virtue ethical. According to consequentialists, what fundamentally make an action morally good or bad are its consequences. According to deontologists, what fundamentally matter are instead the rules or principles which are embodied in an action. According to virtue ethicists, what fundamentally matter are instead the traits of character embodied in the action.
It might seem natural for a consequentialist to accept the extensionality principle. The facts of how an agent frames the outcomes of an action are facts about that agent before they are facts about those outcomes. A perfectly rational consequentialist agent would only be guided by facts about outcomes. In that sense, there is no place for frames in the rational ideal of a consequentialist.
There is a place for frames, however, in the rational ideals of deontologists and virtue ethicists. The principles and the character traits which matter on their views seem able to be thought of as framing outcomes. The same act can be good if it is thought of under the guise of one principle or trait, and bad if it is thought of under another, despite the outcome or the consequences being the same.
The most famous deontological moral philosopher is Kant. Kant believed that we must only act on principles or maxims which we could consent to anyone anywhere following. When we find ourselves unable to give that consent, one of the options open to us is to try to come up with a more nuanced version of that unacceptable principle. For example, I may propose to myself that I pursue a career in forestry just because I want to. The principle I need to be able to consent to is that of anyone pursuing a career in forestry should they want to. This principle is too simple, however: I would not consent to it in a situation in which everyone wanted to pursue a career in forestry. However, I might be able to consent to a principle of pursuing a career in forestry in times when, simultaneously, foresters are needed, I have the skills and means to become one, and I want to become one. In being guided by this more complex principle, I act just as I would have on the simpler one. The outcome is the same: I pursue a career in forestry. It is only an acceptable, preferable outcome, however, when it is chosen as framed by the more complex principle. In this way, Kantian moral theory can violate the extensionality principle.
Bermúdez's (target article, sect. 3.3.1) example of Agamemnon's dilemma might be reimagined in Kantian terms. If Agamemnon can consent to the principle of following Artemis's will in even this case, he can consent it universally. At the same time, he finds himself unable to consent to the principle of murdering his daughter, even though his act would be the same were he following Artemis's will or murdering his daughter.
This dilemma might be phrased in virtue ethical terms, too. Agamemnon is caught between the frames of faithfulness to his gods and of faithfulness to his family. In the Republic, Plato has Socrates imagine a situation in which you have borrowed weapons from a friend, but that friend has since gone mad (Plato, 2012, p. 8/331c). Honouring your debts requires you to return those weapons when your friend demands them. Handing weapons to a madman, though, would be viciously reckless. Honouring your debts is preferable to keeping the weapons from him, but keeping the weapons from him is preferable to recklessly giving them back.
Korsgaard (Reference Korsgaard1996, p. 108, Reference Korsgaard2009, p. 15) uses this example as part of an argument that what agents must assess and choose between are what she terms “actions” rather than “acts.” An action incorporates the principle or end for the sake of which the act is done. Returning the weapons to your friend is an act. Returning the weapons to your friend because they are the property of your friend, and they have asked you to return them, is an action.
Korsgaard reads both Kant and Aristotle, the philosopher most famous as a virtue ethicist, as subscribing to a theory of agency on which actions are our objects of choice and assessment. Bermúdez also appears to subscribe to something like this picture: He suggests that “there is no such thing as making choices over a purely extensional opportunity set, independent of any way of describing or framing the things in it” (target article, sect. 3.1, para. 5). In this respect and in respect of his belief in the coherence of quasi-cyclical preferences, Bermúdez would seem able to claim powerful allies beyond the disciplinary pale of decision theory.
I do not want to overstate the common ground: Bermúdez is working with a different conception of rationality to these philosophers, if one which is different by being less substantive and demanding. There might be empirical reasons to treat moral quasi-cyclical preferences as being special cases, or even not to think of moral reasoning in terms of frames and preferences. In addition, some of these philosophers might be unwilling to admit the possibility of our being unable to resolve conflicts between frames by subsuming them under higher frames. As Aristotle (2012, p. 36/1226b) puts it in the Eudemian Ethics, “those who have no aim before them are not liable to deliberate.” If we are to choose between the urgings of two frames, we must possess some ground on which that choice can be made. The question can only be whether to think of that ground in terms of a higher frame.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Conflict of interest
None.