Frames group choices into categories, thus modifying the incentives for them. This effect makes framing itself a motivated choice rather than a neutral cognition. In particular, framing an inferior choice with a high short-term payoff as part of a broad category of choices recruits incentive to reject it; but this must be motivated by its being a test case.

Bermúdez's “rational framing effects” are consequences of a counterintuitive phenomenon that I call “normative polyphony”: the reality that a single action may, with logical consistency, sustain diverse positive and negative judgments. I show that normative polyphony emerges from “ontological polyphony” – that is, diverse possible framings of relevant details – and illustrate this “polyphony principle” through a reading of Dostoevsky's (1993) Crime and Punishment.

Bermúdez convincingly argues that framing effects are ubiquitous and that this is not a sign of human irrationality, but an unavoidable feature of any intelligent system. The commentary adds that framing effects arise even in formal domains, such as chess and mathematics, which appear paradigms of rational thought. Indeed, finding and attempting to resolve clashes between different frames is a major impetus for deliberative cognition.

We distinguish two types of cases that have potential to generate quasi-cyclical preferences: self-involving choices where an agent oscillates between first- and third-person perspectives that conflict regarding their life-changing implications, and self-serving choices where frame-based reasoning can be “first-personally rational” yet “third-personally irrational.” We argue that the distinction between these types of cases deserves more attention in Bermúdez's account.

There is much to like in Bermúdez's analysis, yet it is incomplete and at times problematic for social decision making and, by extension, interpersonal conflict. Here I explain how four frames – gains, losses, me, we – operate in conjunction and how humans gravitate toward a “me–loss” frame that, without intervention, leads to a breakdown of cooperation and an arguably tragic funeral of the commons.

To assess whether behaviors like framing effects are rational, researchers need to consider decision makers' goals. I argue that researchers should broaden the scope of analysis to include impression management goals. Under predictable conditions, behaviors traditionally considered irrational (e.g., loss–gain framing effects on risk preferences) can be reputationally rewarding, casting doubt on strict claims of irrationality.

Bermúdez claims that agents think about framed outcomes, not outcomes themselves; and that seemingly incoherent preferences can be rational, once defined over framed outcomes. However, the agents in his examples know that alternative frames describe the same outcome, neutrally understood. This undermines the restriction of their preferences to framed outcomes and, in turn, the argument for rational framing effects.

Bermúdez persuasively argues that framing effects are not as irrational as commonly supposed. In focusing on the reasoning of individual decision-makers in complex situations, however, he neglects the crucial role of the social-communicative context for eliciting certain framing effects. We contend that many framing effects are best explained in terms of basic, rational principles of discourse processing and pragmatic reasoning.

Bermúdez's arguments in favour of the rationality of quasi-cyclical preferences conflate reasons, desires, emotions, and responses with genuine preferences. Rational preference formation requires that the decision-makers not only identify reasons, but also weigh them in a coherent way.

Bermúdez suggests that agents use framing to succeed in self-control. This commentary suggests that frames are effective in steering behavior because they modulate information salience. This analysis extends to self-control strategies beyond framing, raising the question whether there remains an explanatory role for dual process theories for self-control.

When communication is not disinterested, seemingly inconsistent preferences are predictable from language pragmatics and information non-equivalence. In addition, the classic risky choice framing effect found in the Asian disease task – risk-aversion with gains and risk-seeking with losses – applies to gambles, but tends to be overgeneralized to non-gambling situations.

Bermúdez argues for rational framing effects in the form of quasi-cyclical preferences. This is supposed to refute the extensionality principle in standard decision theory. In response, I argue that it is better to analyze seemingly quasi-cyclical preferences as ceteris paribus preferences. Furthermore, if frames are included as objects of choice, we can acknowledge rational framing effects without rejecting extensionality.

Framing effects are held to be irrational because preferences should remain stable across different descriptions of the same state of affairs. Bermúdez offers one reason why this may be false. I argue for another: If framing provides implicit testimony, then rational agents will alter their preferences accordingly. I show there is evidence that framing should be understood as testimonial.

Bermúdez's case for rational framing effects, while original, is unconvincing and gives only parenthetical treatment to the problematic assumptions of extensional and semantic equivalence of alternative frames in framing experiments. If the assumptions are false, which they sometimes are, no valid inferences about “framing effects” follow and, then, neither do inferences about human rationality. This commentary recaps the central problem.

Frames for interpreting situations are necessary in the face of time constraints for action and indeterminacy of the “right or optimal thing to do” given multiple objectives but not all frames are equally useful. We need a way of modeling representational frames according to the informational gain of using them and the computational cost of synthesizing a decisive reason for acting from them.

The quasi-cyclical preferences that Bermúdez ascribes to Agamemnon and others in analogous situations do not best represent them. I offer two alternative accounts. One works best if the preference ordering is taken to be the agent's personal betterness ordering of acts; the other works best if it is taken to provide a summary of the agent's dispositions to act.

Bermúdez argues that a framing effect is rational, which will be true if one accepts that the biased editing phase is rational. This type of rationality was called procedural by Simon. Despite being procedurally rational in the evaluation phase framing effect stems from biased way we set a reference point against which outcomes are compared.

Bermúdez describes the extensionality principle as being “almost unquestioned.” This claim might come as a surprise to philosophers who work on agency and ethics. In Kantian deontological ethics and in Platonic or Aristotelian virtue ethics, our preferences for outcomes can be rationally affected by how those outcomes are framed in terms of maxims and character traits.

We connect Bermúdez's arguments to previous theorizing about “leaky” rationality, emphasizing that the decision process (including decision frames) “leaks” into the experience of decision outcomes. We suggest that the implications of Bermúdez's analysis are broadly applicable to the study of virtually all real-world decision making, and that the field needs a substantive and not just a formal theory of rationality.

The normative principle of description invariance presupposes that rational preferences must be complete. The completeness axiom is normatively dubious, however, and its rejection opens the door to rational framing effects. In this commentary, we suggest that Bermúdez's insightful challenge to the standard normative view of framing can be clarified and extended by situating it within a broader critique of completeness.