Framing effects get a bad rap. Arguably, they have pride of place within the narrative that has taken hold in recent decades about our irremediable and predictable irrationality. But is susceptibility to framing effects always revealing of faulty rationality? In Frame it Again: New Tools for Rational Decision-Making, José Luis Bermúdez argues persuasively that it’s not. Contra standard theories of rationality, he argues that being sensitive to how things are framed can be perfectly rational and that, once we understand why this is so, the value of frame-dependent reasoning for rational decision-making comes into view.

Bermúdez’s book is a welcome contribution to the important, broader task of problematizing and challenging the powerful narrative that he labels the ‘litany of irrationality’ (8). This is grounded in the work of psychologists – most famously, Amos Tversky and Daniel Kahneman – which has shaped theoretical and public debates over rational decision-making over the past 40 years or so. The experimental ‘discovery’ of numerous cognitive biases and heuristics formed the basis of Tversky and Kahneman’s (Reference Tversky and Kahneman1986) prospect theory and this in turn influenced the development of behavioural economics and nudge theory. The narrative generated by these findings is one of ‘human fallibility’ (Thaler and Sunstein Reference Thaler and Sunstein2009: 40): there is ‘an ineliminable gap’ (76) between how rational agents ought to make decisions and how we actually do reason practically, where the latter is ‘often divorced from rationality’ (Ariely Reference Ariely2009: xi). However, this litany of irrationality only makes sense against the backdrop of the normative theory of rationality assumed by this experimental and theoretical paradigm – known variously as rational choice theory, Bayesian decision theory or expected utility theory (17). Taking framing and framing effects as his focus, Bermúdez sets out to defuse this narrative by showing that frame-dependent reasoning can be ‘rational in the normative sense’ (18). He offers an insightful and engaging argument to this end, much of which I find plausible. I outline this argument, noting some places where questions remain for me and some others where I think there are further opportunities to link his innovative analysis with existing and burgeoning philosophical debates.

Framing is a ubiquitous part of our mental lives. Salience plays a key role in how we sort and understand our experience (e.g. de Sousa Reference De Sousa1997); and a frame is the lens through which we see and interpret this experience once one aspect of it has been picked out as particularly salient. This continuous but often unconscious mental activity is significant because the lens through which we see the world affects how we evaluate our different options and the outcomes they might bring about (10). Shifting frames, then, can lead to evaluative shifts – such as having preference A over B when a situation is viewed through one frame, but preference B over A in another frame – which is what produces framing effects.

Explained in this way, arguably framing effects do not seem particularly unreasonable. Yet, according to the dominant way of thinking about rationality and rational decision-making in the social sciences, there is ‘absolutely no room for susceptibility to framing effects to be anything other than the height of irrationality’ (66–67). For they contravene the principle of extensionality which Bermúdez claims is, from the perspective of rational choice theory (hereafter RCT), a ‘non-negotiable requirement of rationality’ (75).Footnote ¹ This principle states that a rational agent will value the same thing equally, regardless of how it is described. Or, in other words, that rational thinking should be frame-independent. Bermúdez aims to refute the justification for this principle, at least as regards its application in a certain set of cases.

Chapters 5 and 6 comprise the heart of this argument. Bermúdez’s first move is to highlight that the discussion of framing effects and their rationality has been distorted by the kinds of scenarios used by experimenters.Footnote ² As soon as we step out of the lab, the decision problems we face are generally much more complicated and multifaceted. Any illusion of frame-independence is burst by the fact that outcomes are not neutrally characterizable, showing that decision problems are ‘typically constructed, not given’ (120). Framing is integral to this mental process of construction, since frames are a way of fastening onto one feature of a complex, multi-dimensional process. In itself, this is not a novel point; but Bermúdez develops it in new ways.

He introduces the idea of quasi-cyclical preferences via examples of moral dilemmas faced by figures such as George Orwell, Macbeth, and Agamemnon. Reflecting on an episode from his time fighting in the Spanish Civil War, Orwell tells us this about his decision not to shoot at a particular Fascist soldier: ‘I had come here to shoot at ‘Fascists’; but a man who is holding up his trousers isn’t a ‘Fascist’, he is visibly a fellow creature, similar to yourself, and you don’t feel like shooting at him’ (Orwell Reference Orwell1957: 193–194; cited at 101–102). According to Bermúdez, this is an example of a rational agent having what he calls ‘quasi-cyclical preferences’. This occurs when someone prefers A (in this case, shooting a Fascist) to B (lowering his rifle) and B to C (shooting a fellow human being), despite knowing that A and C are different ways of framing the same outcome (113). Understood thus, the question of whether framing effects can be rational is the same as the question of whether quasi-cyclical preferences can be rationally held (113).

What’s the case for thinking that quasi-cyclical preferences can be rationally held? First, Bermúdez explains how complex decision-making situations are ‘fundamentally different’ (8) from the oft-cited examples given as evidence of the irrationality of framing effects. Following Sher and McKenzie (Reference Sher and Mackenzie2006), he locates this difference in the fact that the two (or more) frames are not informationally equivalent, even while being logically equivalent, in these richer cases (92). An important part of the story for Bermúdez is that different ways of framing a situation or outcome can engage reasoners’ emotions, motivations or values in different ways (126–128). Consequently, alternative frames often bring into play distinct, and potentially conflicting, reasons for action.

Second, constructing a decision problem ‘requires active thought and reflection’ and these processes are themselves subject to standards of rationality (121). Specifically, a plausible normative theory of rationality will have a due diligence requirement. That is, when mentally constructing a decision problem (in the kinds of more complicated situations that require us to do this), we ought to be ‘appropriately sensitive to as many potential consequences of the different courses of action available to [us] as possible’ (121). Not to do this, Bermúdez maintains, would put us in the category of the imagined character ‘Agamemnon-minus’, who only frames his decision problem about whether to sacrifice his daughter to the goddess Artemis in one way – either as the leader of the Greek forces but not as Iphigenia’s father, or vice versa (114–115). We meet the due diligence requirement when we recognize the distinct reasons in play, and entertaining different frames enables us to do this.

Lastly, Bermúdez argues that it is rationally permissible for someone who views a scenario under multiple frames to end up with quasi-cyclical preferences as a direct result of her different, frame-dependent emotional and rational engagement with these frames (136–137). This holds whenever the alternative frames are legitimate – by which he means complementary (i.e. each reflecting a genuine aspect of a complex phenomenon) and consistent (i.e. they don’t strictly contradict each other) – and are not based on false beliefs (134–135).Footnote ³ When frame-dependent reasoning results in quasi-cyclical preferences, the process of coming to have this preference set is rational, according to Bermúdez, in virtue of being responsive to reasons in the way specified by the due diligence requirement. That is, quasi-cyclical preferences arising from satisfying this requirement of reason ‘inherit the rationality of the process that led to them’ (123–124). What follows is that quasi-cyclical preferences are not only rationally permitted, but that in certain kinds of cases it would be irrational not to have them. If correct, this reins in the principle of extensionality’s legitimate role within a normative theory of rationality.

Bermudez takes himself to have presented ‘a different way of thinking about rationality [according to which] what makes decision-makers rational is how they reason’ (242).Footnote ⁴ This general way of thinking about rationality is not novel to philosophers. Nonetheless, given the dominance of instrumental theories among his target audience of social scientists, his case for a more process-focused normative theory of rational decision-making is undoubtedly an important contribution to the ongoing debate over rationality. It questions the justification for the principle of extensionality in a novel way, by making a convincing case for the value of frame-sensitive reasoning for rational decision-making.

Nonetheless, questions remain for me about precisely how we ought to understand Bermúdez’s main theoretical contribution. One set of questions revolves around the rationality of due diligence, and how exactly we get from that to the rationality of quasi-cyclical preferences. I find Bermúdez’s suggestion that rationality imposes informational, in addition to procedural and consistency-based requirements (118) very plausible for the reason he gives: how one sets up a decision problem directly fixes the content of her reasoning (122). But at times the way Bermúdez puts the point about due diligence seems overly demanding. For example, he claims that the argument from due diligence places decision-makers ‘under a requirement of rationality to think through as deeply as possible all the potential ramifications of the different acts available to them’ (123). This seems less plausible; further specificity on this matter is needed in my view, especially given the centrality of this requirement to his argument.

I also found myself unsure about the key move relating to rationally permitted violations of the principle of extensionality. Bermúdez admits that having quasi-cyclical preferences need not follow from recognizing the complexity of the decision problem. Following Frederic Schick, for instance, we might think that ‘seeing’ (a term of art for Schick) the soldier as a fellow human being was a ‘wake up call’ for Orwell that resolved the conflict in his mind such that Orwell didn’t have quasi-cyclical preferences when he decided against shooting (Schick Reference Schick2003: 3, 10; cited at 103). But Bermúdez does maintain that one plausible version of what went on in this case is that ‘one frame came to dominate the other [in Orwell’s mind] without the other one disappearing’ (105). His view requires that rational agents ‘knowingly and simultaneously’ (100) hold quasi-cyclical preferences even in cases where they ultimately come to decide in favour of one of the competing option sets. I found the author’s discussion of the ‘rational regret’ (265) that agents often feel after making such a decision to be suggestive with respect to this point. Nevertheless, precisely what the difference is between one frame coming to dominate the other and the rational process of assigning more weight to one set of reasons (made salient by one frame) than another (made salient by an alternative frame) remains somewhat opaque to me.

A second type of question that I have concerns the scope of the critique of the principle of extensionality. Bermúdez seeks to show that this principle should play a more limited role in a theory of rational decision-making than is standardly thought. Although I found the general shape of his argument to this end plausible, I wasn’t always clear on whether he sees his account of frame-sensitive reasoning as a revision to the existing dominant way of thinking about rational decision-making, or whether it represents (part of) a more wholesale departure. He critiques others who propose alternatives to RCT for being ‘ambiguous about what they are actually proposing’ (82). Despite his efforts against this, at times I read Bermúdez as offering an alternative normative theory (i.e. in proposing the due diligence requirement as a standard of rationality and setting out his account of the model frame-sensitive reasoner). Yet, at other times I read him more as showing how we can make room for framing and framing effects within RCT, while generally accepting it as our normative theory of rationality (i.e. by loosening the grip of the principle of extensionality). Both represent a contribution to the debate; but I would have appreciated more clarity on which type of contribution he is making.

As a philosopher interested in the ethics of influence, and cognizant of the burgeoning interest in salience as a philosophical topic (see Archer Reference ArcherForthcoming), I was also mindful of some opportunities for developing Bermúdez’s work on framing in ways that would further underline its interest and value for philosophers.

For example, Bermúdez only references the ethical issues around framing in passing when he comes to applying his account to ‘deadlock’ in public discourse in Chapter 10 (i.e. he mentions that it can be ‘a tool for manipulation’) (215). This is due primarily to his focus on rationality. Nonetheless, the issue of who (or what) is involved in the ‘framing’ process never seemed far away. Bermúdez zones in on the role of the individual thinker, giving them a lot of agency in the framing process – e.g. Orwell ‘conceptualizing the soldier in two different ways’ such that his decision will depend ‘upon how he frames the scene in front of him’ (102–104; my emphasis). This differs from Schick’s analysis, for instance, which is more passive in tone – certain ‘understandings’ get passed on to us by the world around us, such as how ‘seeing’ the soldier holding up his trousers changed Orwell’s intention to shoot.Footnote ⁵ Of course, we can also actively reframe things in our minds to change how we see them; but this doesn’t seem to be what is going on a lot of the time. Rather, things are made salient to us (or framed for us, you might say) by socio-environmental factors. Sometimes this salience-adjustment is done intentionally, with other agents wishing to frame things so that we see or understand them in a particular way (as is the case in the experiments that initially generated the evidence of framing effects); other times it is contingent, as in the Orwell case. Accordingly, I think that further conceptual analysis of these more active and passive types of framing – including a deeper understanding of salience and how it functions or is involved in the different cases – would be valuable.

Furthermore, this type of analysis would also direct us towards a set of interesting questions that are related specifically to the book’s focus on the rationality of framing effects and to framing as a tool for rational decision-making. Once we include in our analysis the role of other actors in (variously) affecting the frames we adopt, we might ask whether the processes and skills of frame-sensitive reasoning could be externally scaffolded. Could certain kinds of salience-adjusting nudges play a legitimate role in supporting the development or exercise of the skills required as a frame-sensitive reasoner (Noggle Reference Noggle2018; Adams and Niker Reference Adams, Niker, Niker and Bhattacharya2021)? We might think it possible, for example, to use framing techniques to nudge us into seeing a situation in a different way, which could prompt us to reason across frames (Niker Reference Niker, Harrison and Walker2018). And, if so, would this frame-sensitive reasoning and any quasi-cyclical preferences it generated still count as rational in virtue of this (scaffolded) process? In other words, do Bermúdez’s ‘new tools for rational decision-making’ need always to be internal to the individual agent? Addressing these questions would connect his in-depth study of the rationality of framing with the growing interest in socio-ecological accounts of rational agency and responsibility in philosophy (e.g. Hurley Reference Hurley, Knight and Stemplowska2011; Schmidt Reference Schmidt2019). And I, for one, would be very interested to get Bermudez’s view on these questions.