Many cognitive biases are due to framing. Bermúdez agrees, but he thinks (1) there are rational framing effects in the form of quasi-cyclical preferences. Furthermore, (2) these effects refute the extensionality principle – the widely accepted normative assumption that rational preferences should not depend on how outcomes are framed. I shall argue against both conclusions. Seemingly quasi-cyclical preferences are better construed as ceteris paribus preferences. Moreover, if frames are included as objects of choice, we can accept rational framing effects without rejecting extensionality.

Consider Bermúdez's example of Agamemnon. Bermúdez claims that the following sentences are true:

1. (A) Agamemnon prefers Following Artemis's Will to Failing his Ships and People.

2. (B) Agamemnon prefers Failing his Ships and People to Murdering his Daughter.

Furthermore, Bermúdez thinks that Agamemnon is not irrational, even though Agamemnon knows that Following Artemis's Will and Murdering his Daughter are two different ways of framing the same outcome.

I think Bermúdez is mistaken to treat this as demonstrating quasi-cyclical preferences. A better analysis is that (A) and (B) express ceteris paribus general preferences, rather than strict preferences about specific outcomes (Hansson, Reference Hansson1996; Van Benthem, Girard, & Roy, Reference Van Benthem, Girard and Roy2009). If I say I prefer coffee over tea, we normally take this to involve an implicit qualification – all else being equal, I prefer coffee to tea. This preference is defeasible and not absolute. I am not being inconsistent if I happen to choose tea over an overpriced, watery coffee. If (A) and (B) express ceteris paribus preferences, we can see why they can both be true, even when Agamemnon decides to sacrifice his daughter. All else being equal, Agamemnon prefers Failing his Ships and People to Murdering his Daughter. But in this unfortunate instance, taking everything into account, it is indeed rational for him to murder his daughter. It does not matter whether the outcome is framed in terms of Murdering his Daughter or Following Artemis's Will. Extensionality is consistent with (A) and (B) being true. The same analysis applies to Bermúdez's Macbeth example, and his examples of gun-violence and energy-independence, as they all share the same structure.

Bermúdez discusses rational framing effects in three types of everyday situations. The first relates to self-control. For example, an agent might be tempted to choose a smaller immediate reward over a larger future reward. Bermúdez proposes that the agent can resist temptation through self-control by reframing the future reward as a case of having successfully resisted the immediate reward. This is because the agent might consider resisting temptation as demonstrating highly valuable traits such as virtue and commitment. I suspect this misrepresents typical cases of temptations, where considerations about virtues often lack motivational power (“Lord make me chaste but not yet,” as Saint Augustine might say).

In any case, even if Bermúdez is correct that reframing allows the agent to overcome temptation, this only goes to show that the agent underestimated the full value of the future reward earlier on. The value of overcoming a temptation should also have been included, as they belong to the same outcome, regardless of how it is framed. Failure to compute utility correctly does not constitute evidence for rational quasi-cyclical preferences.

The same problem arises for Bermúdez's second type of cases concerning strategic decision in game theory. In his snowdrift example, cooperation is not the preferred outcome under an “I”-frame ranking, but it eventually becomes the preferred outcome when the subject changes to a “we”-frame ranking. It is unclear why this supports rational quasi-cyclical preferences. Bermúdez says the subject changes frame “as considerations of fairness start to take hold” (target article, sect. 5.3, para. 9). This suggests that the subject failed to fully assess the fairness of the situation when cooperation was rejected under “I”-frame thinking. But this does not show that the rationality of cooperation is frame-relative. As Rawls (Reference Rawls1971) has pointed out, reflective equilibrium requires working back and forth among our considered judgments. Reflecting on the principles of justice can lead us to revise our earlier beliefs as to whether something is fair, but this is consistent with extensionality.

Bermúdez's third type of example concerns interpersonal conflicts. He argues that discursive deadlocks in the public domain often involve clashes of frames, and their resolution requires techniques such as reflexive decentering and imaginative simulation. I am sympathetic to Bermúdez's framework, as I think it complements related proposals, such as the use of emotional regulation in resolving intractable conflicts (Halperin, Cohen-Chen, & Goldenberg, Reference Halperin, Cohen-Chen and Goldenberg2014). These proposals can help us promote rationality and objectivity in public reasoning. However, we can acknowledge these insights without accepting rational quasi-cyclical preferences.

I agree with Bermúdez that there can be rational framing effects, but this is compatible with extensionality. Consider how mindsets affect learning and performance. A mindset at its core is a set of frames, incorporating a system of concepts, principles, and values for interpreting the world. There is considerable evidence that a “growth” mindset that views intelligence as a malleable rather than fixed trait is more likely to lead to success (Dweck, Reference Dweck2006; Yeager & Dweck, Reference Yeager and Dweck2012). Similarly, having a “stress-is-enhancing” mindset – perceiving stress as functional and adaptive – might improve coping behavior and performance (Jamieson, Nock, & Mendes, Reference Jamieson, Nock and Mendes2012). If these findings are correct, there will be times when a project will be successful only if we embrace it with an appropriate mindset. In these situations, whether it is rational to pursue the project seems to be frame-relative.

This does not refute extensionality, for the simple reason that the proper object of choice in these situations is not solely the project but the project coupled with a commitment to a particular mindset. Bermúdez says reasons are always frame-relative, that it is “frames all the way down” (target article, sect. 6.1, para. 3). But then whether a decision is rational relative to a given frame is presumably also relative. Does it mean there is no objective fact as to whether a decision is rational? However, if mindset can be an object of rational choice, we can then preserve extensionality and the objectivity of reason, and still recognize the existence of rational framing effects.