Classical theories of rationality often assume complete knowledge of decision-relevant factors. However, this assumption contradicts apparent constraints on decision-making in the real world, where people need to simultaneously search for information and determine its relevance for choices at hand. Because the quantity of information in many decisions far outstrips an individual's information processing capacity, selective attention is required to maintain representations of information one piece at a time, essentially highlighting different frames at different times during choice (Kiyonaga & Egner, Reference Kiyonaga and Egner2013; Moore & Zirnsak, Reference Moore and Zirnsak2017; Myers, Stokes, & Nobre, Reference Myers, Stokes and Nobre2017; Smith & Krajbich, Reference Smith and Krajbich2019). While this can theoretically result in a process of sequential frame selection using rational goal-driven attention, attention is also frequently exogenously constrained by the environment: What is attended is as often as not stimulus-driven as opposed to goal-directed (Corbetta & Shulman, Reference Corbetta and Shulman2002; Vanunu, Hotaling, Le Pelley, & Newell, Reference Vanunu, Hotaling, Le Pelley and Newell2021). Importantly, these attentional processes may interact in dynamic ways over time: the decision context primes particular frames of evaluation (Diederich & Trueblood, Reference Diederich and Trueblood2018; Maier, Raja Beharelle, Polanía, Ruff, & Hare, Reference Maier, Raja Beharelle, Polanía, Ruff and Hare2020), prior frames differentially enhance and constrain the accessibility of subsequent framings (Johnson, Häubl, & Keinan, Reference Johnson, Häubl and Keinan2007; Nook, Satpute, & Ochsner, Reference Nook, Satpute and Ochsner2021), and executed decisions frame and bias post-choice evaluation (Chaxel, Russo, & Kerimi, Reference Chaxel, Russo and Kerimi2013; Navajas, Bahrami, & Latham, Reference Navajas, Bahrami and Latham2016). We argue here that determining when, and if, framing effects are rational requires a thorough consideration of these components of frame selection.

First, we argue that the external environment may disproportionately impact initial frames compared to internally represented goals, because attention tends to be drawn first toward salient information in the environment. Indeed, this is the mechanism for most framing studies, which induce frames by highlighting specific information in the decision problem itself (Kühberger, Reference Kühberger1998; Levin, Schneider, & Gaeth, Reference Levin, Schneider and Gaeth1998; McDonald, Graves, Yin, Weese, & Sinnott-Armstrong, Reference McDonald, Graves, Yin, Weese and Sinnott-Armstrong2021). This is true not only in classical framing studies, but even in the studies that Bermúdez cites as evidence for the potential rationality of framing effects. For example, studies showing that framing marshmallows as “puffy clouds” facilitates rational choice and self-regulation work by explicitly encouraging participants to adopt these frames. It is not clear that people would typically select such “cool” frames in real-world contexts, particularly as the initial frame. Instead, research suggests that foods' appetitive qualities (e.g., sweet and tasty) are the most immediately salient dimension of evaluation (i.e., “hot frames”; see Maier et al., Reference Maier, Raja Beharelle, Polanía, Ruff and Hare2020; Sullivan, Hutcherson, Harris, & Rangel, Reference Sullivan, Hutcherson, Harris and Rangel2015), and that these appetitive frames may be rapidly represented regardless of people's efforts to refocus on healthy frames (HajiHosseini & Hutcherson, Reference HajiHosseini and Hutcherson2021). Effortful attentional control is thus usually required to refocus attention away from initially appetitive frames in order to regulate one's choice (Papies, Stroebe, & Aarts, Reference Papies, Stroebe and Aarts2008; Rangel, Reference Rangel2013). Studies like the ones Bermúdez cites circumvent the need for regulatory control by presenting “cool” frames in advance, effectively off-loading the work of controlled attention onto the environmental context.

Second, although we fully agree that self-control may facilitate the decision-maker's ability to reframe decision problems in alignment with their goals, we note that exogenously determined initial frames can also constrain the accessibility of subsequent frames. For example, query theory accounts of the endowment effect show that framing a mug first as being owned by the decision-maker led people to consider its value-enhancing aspects more than when the mug was first framed as one of two possible choice options (Johnson et al., Reference Johnson, Häubl and Keinan2007). Moreover, certain frames of evaluation may be even more strongly constrained by sequence due to their emergent nature. For example, Bermúdez discusses two frames of evaluation in a strategic interpersonal interaction where people can cooperate for maximal joint outcomes or selfishly choose to minimize potential losses for themselves. Here, the “I-frame” provides strategic reasons for selfish behavior by emphasizing self-relevant outcomes of options while the “We-frame” provides reasons for cooperative behavior by emphasizing joint outcomes of options. Yet Bermúdez's discussion does not consider that the sequence of frames in this decision problem is directionally constrained: People have to first separately acquire information about their own outcomes (“I-frame”) and their partner's outcomes (“You-frame”) in order to evaluate joint outcomes (“We-frame”). This is highly consequential for decisions because recent work suggests that the information search necessary to adopt a frame incurs a cost (i.e., time and effort; see Callaway, Rangel, & Griffiths, Reference Callaway, Rangel and Griffiths2021; Jang, Sharma, & Drugowitsch, Reference Jang, Sharma and Drugowitsch2021). Construction of the complex “We-frame” incurs a greater cost by requiring two separate information samples. People may thus be less inclined to spontaneously adopt this frame, especially under time constraints that limit the number of possible frames and increase the cost of each frame (Roberts, Teoh, & Hutcherson, Reference Roberts, Teoh and Hutcherson2022; Teoh & Hutcherson, Reference Teoh and Hutchersonin press; Teoh, Yao, Cunningham, & Hutcherson, Reference Teoh, Yao, Cunningham and Hutcherson2020).

Finally, we highlight that constraints on frame-selection extend beyond the immediate choice and into processes of post-choice evaluation. Evidence suggests that people may continue to acquire more information about forgone options and sometimes reframe the chosen option in light of new information (Shani & Zeelenberg, Reference Shani and Zeelenberg2007; Teodorescu, Sang, & Todd, Reference Teodorescu, Sang and Todd2018). This process could promote more rational choice, by providing decision-makers with the opportunity to pre-emptively frame future decisions in service of rational goals (see Braver, Reference Braver2012; Brick, MacIntyre, & Campbell, Reference Brick, MacIntyre and Campbell2016 for discussions of proactive and reactive control). However, just as prior frames constrain the accessibility of subsequent frames during choice, frames adopted during the decision process may also continue to bias post-choice framing of the decision problem. For example, research finds that people tend to seek out confirmatory information to justify the decisions they made (Brehm & Wicklund, Reference Brehm and Wicklund1970; Qin et al., Reference Qin, Kimel, Kitayama, Wang, Yang and Han2011; Scherer, Windschitl, & Smith, Reference Scherer, Windschitl and Smith2013), diminishing the potential for goal-directed attention to explore alternative frames that would promote more rational framings in future decisions.

Therefore, while different frames may provide different reasons that lead to distinct choices and people may strategically reframe choices in alignment with their goals, we emphasize here that exogenous environmental and contextual factors strongly constrain these strategies. We have highlighted thus far how these constraints gate the accessibility of particular frames during and after choice. Understanding these constraints on frame-selection will prove critical to any theoretical account of how framing effects can be strategically used to promote human rationality.