In what sense is standard decision theory a theory of rational decision? Although it is neutral about the content of people's preferences, the theory imposes a few constraints on their shape: It requires that preferences are “well-ordered” or “consistent,” as specified by axioms such as completeness, transitivity, and independence. This axiomatic conception of rationality, however, conceals a deeper sense in which the agents of decision theory make rational decisions. Significant decisions typically involve conflicting reasons – there are reasons to do X, but also reasons to do Y, when X and Y are available. An abundant snowfall followed by a spell of sunshine may give me a good reason to go skiing, for example, but the fact that I haven't visited my parents for a while may give me a good reason not to do so. As Bermúdez correctly points out, we often become aware of such conflicts by looking at the same situation from different perspectives – the perspective of the ski-lover versus the perspective of the good son, in the example above. Endorsing (partially, and preliminarily) different frames is indeed an effective way to make sure that all reasons – the reasons that can possibly matter to us – are taken into consideration in the process of decision-making.
Why “partially” and “preliminarily”? Rational decision theory requires that preferences are consistent. Consistency, in turn, requires that each option is assigned a stable value, and that the value of each option reflects the relative value of different aspects of the option. The value of each aspect and each option must then be weighed against the values of other options (and their aspects). This weighing process is the truly difficult part of decision-making, as we all know from personal experience. It is a common complaint that the standard theory does not offer much help in making up one's own mind and weighing different options. But it does give some help, if only as a warning: When the preferences that are produced by the weighing process turn out to be inconsistent, then we know that something wrong must have happened. Some aspect of an option, for example, must have been evaluated differently in contexts that are effectively identical – as in the framing effects described by Bermúdez in his paper.
The point of framing, to put it differently, is not simply to see things from a different perspective, however intellectually pleasing this may be. The point is to make up one's own mind, to decide what the relative value of different options (and aspects of the options) really is. A rational agent thus cannot simply endorse one frame and then another. The rational decision-maker must compare the (partial, frame-dependent) reasons or valuations that each frame has elicited, and come up with an all-things-considered evaluation of the alternative options.
According to a prominent proposal, a preference just is an “all-things-considered evaluation” (Hausman, Reference Hausman2011). Whether this conception of preference is adequate in the descriptive psychological sense is controversial (e.g., Angner, Reference Angner2018), but there is little doubt that it fits the standard account of rational decision-making. Bermúdez unfortunately constantly conflates preferences with cognate entities, such as emotions, desires, or “responses.” One of his working hypotheses goes, for example:
(H3) Framing effects and quasi-cyclical preferences can be rational in circumstances where it is rational to have a complex and multi-faceted response to a complex and multi-faceted situation.
But a “response” is not a preference. It may be a gut reaction, an emotion, or a pro tanto evaluation, in which case it may constitute part of the input for preference-formation. It may even be a choice made impulsively before the process of due diligence has been properly completed. But, alas, in such a case it would be an irrational choice, not a preference in the proper, all-things-considered sense.
Examples can be found easily in Bermúdez's paper: Agamemnon may want to follow Artemis's will (under the grip of frame A), and may want to fail his ships and people (under the grip of B), but he cannot prefer both. Macbeth may have a desire or a reason to fulfil his double duty to Duncan, and another desire or reason to take the throne, but he cannot prefer both, in the sense of rational preference. Another way to put it is that a rational decision-maker must not only “be sensitive to the full range of potential reasons that there might be for choosing one way rather than another” (sect. 3.4, para. 3); she must also make up her mind about the relative importance of such reasons. A rational decision-maker reasons across frames, and not just within each frame. She looks at the various reasons (pros and cons) from a bird's-eye point of view, and decides which ones are more important for her, all-things-considered. Although “reasons are frame relative” (sect. 3.4, para. 4), in other words, the comparative evaluation of reasons cannot be made within a single frame.
This is, in a nutshell, why quasi-cyclical preferences cannot be rational preferences. As Bermúdez correctly points out, a rational preference must inherit the rationality of the process that generated them, and quasi-cyclical preferences are the output of a process that fails to accomplish what rationality requires.
You have
Access
In what sense is standard decision theory a theory of rational decision? Although it is neutral about the content of people's preferences, the theory imposes a few constraints on their shape: It requires that preferences are “well-ordered” or “consistent,” as specified by axioms such as completeness, transitivity, and independence. This axiomatic conception of rationality, however, conceals a deeper sense in which the agents of decision theory make rational decisions. Significant decisions typically involve conflicting reasons – there are reasons to do X, but also reasons to do Y, when X and Y are available. An abundant snowfall followed by a spell of sunshine may give me a good reason to go skiing, for example, but the fact that I haven't visited my parents for a while may give me a good reason not to do so. As Bermúdez correctly points out, we often become aware of such conflicts by looking at the same situation from different perspectives – the perspective of the ski-lover versus the perspective of the good son, in the example above. Endorsing (partially, and preliminarily) different frames is indeed an effective way to make sure that all reasons – the reasons that can possibly matter to us – are taken into consideration in the process of decision-making.
Why “partially” and “preliminarily”? Rational decision theory requires that preferences are consistent. Consistency, in turn, requires that each option is assigned a stable value, and that the value of each option reflects the relative value of different aspects of the option. The value of each aspect and each option must then be weighed against the values of other options (and their aspects). This weighing process is the truly difficult part of decision-making, as we all know from personal experience. It is a common complaint that the standard theory does not offer much help in making up one's own mind and weighing different options. But it does give some help, if only as a warning: When the preferences that are produced by the weighing process turn out to be inconsistent, then we know that something wrong must have happened. Some aspect of an option, for example, must have been evaluated differently in contexts that are effectively identical – as in the framing effects described by Bermúdez in his paper.
The point of framing, to put it differently, is not simply to see things from a different perspective, however intellectually pleasing this may be. The point is to make up one's own mind, to decide what the relative value of different options (and aspects of the options) really is. A rational agent thus cannot simply endorse one frame and then another. The rational decision-maker must compare the (partial, frame-dependent) reasons or valuations that each frame has elicited, and come up with an all-things-considered evaluation of the alternative options.
According to a prominent proposal, a preference just is an “all-things-considered evaluation” (Hausman, Reference Hausman2011). Whether this conception of preference is adequate in the descriptive psychological sense is controversial (e.g., Angner, Reference Angner2018), but there is little doubt that it fits the standard account of rational decision-making. Bermúdez unfortunately constantly conflates preferences with cognate entities, such as emotions, desires, or “responses.” One of his working hypotheses goes, for example:
(H3) Framing effects and quasi-cyclical preferences can be rational in circumstances where it is rational to have a complex and multi-faceted response to a complex and multi-faceted situation.
But a “response” is not a preference. It may be a gut reaction, an emotion, or a pro tanto evaluation, in which case it may constitute part of the input for preference-formation. It may even be a choice made impulsively before the process of due diligence has been properly completed. But, alas, in such a case it would be an irrational choice, not a preference in the proper, all-things-considered sense.
Examples can be found easily in Bermúdez's paper: Agamemnon may want to follow Artemis's will (under the grip of frame A), and may want to fail his ships and people (under the grip of B), but he cannot prefer both. Macbeth may have a desire or a reason to fulfil his double duty to Duncan, and another desire or reason to take the throne, but he cannot prefer both, in the sense of rational preference. Another way to put it is that a rational decision-maker must not only “be sensitive to the full range of potential reasons that there might be for choosing one way rather than another” (sect. 3.4, para. 3); she must also make up her mind about the relative importance of such reasons. A rational decision-maker reasons across frames, and not just within each frame. She looks at the various reasons (pros and cons) from a bird's-eye point of view, and decides which ones are more important for her, all-things-considered. Although “reasons are frame relative” (sect. 3.4, para. 4), in other words, the comparative evaluation of reasons cannot be made within a single frame.
This is, in a nutshell, why quasi-cyclical preferences cannot be rational preferences. As Bermúdez correctly points out, a rational preference must inherit the rationality of the process that generated them, and quasi-cyclical preferences are the output of a process that fails to accomplish what rationality requires.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Conflict of interest
None.