Jara-Ettinger and Dunham too quickly assimilate the mentalistic stance to theory-like mindreading. They therefore leave open their institutional stance to a familiar reply from theory-theorists, namely, that the target phenomena can be recovered by enriching the tacit theory. A more principled rebuttal of the theory-theory is required to motivate the institutional stance, and that, in turn, requires an alternative mentalist capacity. Simulation theory is one such alternative.
Simulation-based mindreading offers a mentalistic but non-theory-like alternative to theory-theory. To explain and predict others’ behavior, one imaginatively places oneself in the other’s epistemic and motivational situation. Much mental-state attribution thus proceeds by redeploying one’s own practical and affective resources (Goldman, Reference Goldman1992; Reference Goldman2006; Gordon, Reference Gordon1986; Gordon & Cruz, Reference Cruz, Gordon and Nadel2003; Heal, Reference Heal and Butterfield1986; Todd & Tamir, Reference Todd and Tamir2025).
This matters here because the “surprise” cases the authors emphasize invite a standard theory-theory reply. The maneuver is to accommodate the data by updating the theory to make it more granular and reflect changes in context. The Kate and Eric vignette (sect. 1) illustrates this. When Kate asks her workplace assistant Eric to get her snacks after running into him at a movie theater, Jara-Ettinger and Dunham argue that our surprise reflects reliance on the institutional stance. On their view, this is an interpretive framework appropriate to the workplace that does not fit the informal setting, so we are surprised when we observe the institutional behaviors manifested. Since the theory-theory attribution of Kate and Eric’s beliefs and desires can be readily carried over to the informal context, the theory-theorist allegedly has no explanation for our surprise.
A theory-theorist can readily offer a reply. Nothing in the theory-theory requires attributing the same belief–desire profile across contexts. The theory can be tuned so that work and leisure contexts carry different default expectations. Contemporary defenses of theory-theory often stress this flexibility, especially in cases that are unexpected or involve idiosyncratic agents (Jara-Ettinger, Reference Jara-Ettinger2019; Jern, Lucas, & Kemp, Reference Jern, Lucas and Kemp2017). Kate can be treated as just such an agent, which would motivate a more complex attribution of her beliefs and desires. For example, she may have confused beliefs about the boundaries of an assistant’s obligations. On this approach, surprise arises because Kate behaves as if the work context belief–desire profile applies in a leisure setting.
Both the institutional stance and the theory-theory have an explanation for the data. A way out of this standoff is to appreciate that the institutional stance is attractive because it explains the surprise without relying on unconstrained enrichment of a mentalistic theory. More generally, the motivation for the institutional stance consists in highlighting behavioral phenomena that resist, or are less parsimoniously explained, by the mentalistic stance. But the theory-theory always has resources to accommodate the instances that motivate the institutional stance. This flexibility is precisely why surprise cases do not, by themselves, force a choice between theory-like mentalizing and the institutional stance. Jara-Ettinger and Dunham effectively concede as much when they note that a mentalistic interpretation is available (e.g., sect. 6.3.2), but that it requires increasingly complex and intuitively implausible mental-state attributions. They contrast this with the institutional stance, which makes the same interactions appear seamless by appeal to role-based expectations. The institutional stance is therefore appealing not because it refutes the theory-theory on its own terms but because it explains the surprise without relying on open-ended enrichment of a tacit mentalistic theory. The worry about the theory-theory is not complexity per se, but case-by-case accommodation without any independent constraints on the theory. And there is a ready alternative to the theory-theory version of mentalism in the simulation theory.
Accordingly, I propose a division of labor. In structured contexts shaped by roles and norms, the institutional stance plus script knowledge does much of the explanatory work that is often assigned to theory-like mentalizing. In less structured contexts, simulation-based mindreading supports flexible and agent-specific interpretation. On this picture, the institutional stance does not primarily compete with mentalism as such. It mainly competes with a theory-like construal of mentalism. The range of cases in which theory-theory is the most parsimonious option is therefore considerably narrowed. These are cases in which behavior is genuinely novel or nonstandard and is not readily explained by folk physics, as when someone trips and stumbles, not captured by a familiar script, and not made predictable by institutional role structure. They are also cases in which simulation is unreliable because the target is dissimilar, atypical, or otherwise not well modeled by first-person redeployment.
A second motivation Jara-Ettinger and Dunham offer is computational economy. They argue that role-based reasoning can reduce the cognitive costs and uncertainty of mentalizing and can reduce the online computational burden of tracking beliefs and desires in real time (sect. 6.3.1). This is a useful point, but it further highlights what is missing from their picture of the mentalistic stance. Simulationist accounts were motivated in part by the same computational contrast, since simulation predicts others by redeploying first-person resources rather than by applying an increasingly complex theory (Cruz, Reference Cruz1998). The institutional stance and simulation theory therefore look like parallel forms of cognitive offloading that apply in different domains.
Jara-Ettinger and Dunham suggest that mentalism combined with the institutional stance yields richer possibilities for representing and creating institutional structure (sect. 6.3.3). Simulation-based mindreading can play that same enriching role, since it supports agent-specific expectations about how particular individuals will interpret and negotiate institutional roles, without committing us to the theory-theory.
Once we acknowledge how readily the theory-theory can be modified, its treatment of the authors’ surprise cases becomes difficult to state without an overly permissive enrichment strategy. If the theory can be freely revised to match each surprise, it no longer looks like the kind of stable, learnable resource that could guide online interpretation. The institutional stance treats the same phenomena by deploying independently learnable representations of social structure in terms of roles and norms, rather than by case-by-case elaboration of an underlying theory. It looks stronger still once simulation-based mindreading is counted as a mentalistic alternative.
You have
Access
Jara-Ettinger and Dunham too quickly assimilate the mentalistic stance to theory-like mindreading. They therefore leave open their institutional stance to a familiar reply from theory-theorists, namely, that the target phenomena can be recovered by enriching the tacit theory. A more principled rebuttal of the theory-theory is required to motivate the institutional stance, and that, in turn, requires an alternative mentalist capacity. Simulation theory is one such alternative.
Simulation-based mindreading offers a mentalistic but non-theory-like alternative to theory-theory. To explain and predict others’ behavior, one imaginatively places oneself in the other’s epistemic and motivational situation. Much mental-state attribution thus proceeds by redeploying one’s own practical and affective resources (Goldman, Reference Goldman1992; Reference Goldman2006; Gordon, Reference Gordon1986; Gordon & Cruz, Reference Cruz, Gordon and Nadel2003; Heal, Reference Heal and Butterfield1986; Todd & Tamir, Reference Todd and Tamir2025).
This matters here because the “surprise” cases the authors emphasize invite a standard theory-theory reply. The maneuver is to accommodate the data by updating the theory to make it more granular and reflect changes in context. The Kate and Eric vignette (sect. 1) illustrates this. When Kate asks her workplace assistant Eric to get her snacks after running into him at a movie theater, Jara-Ettinger and Dunham argue that our surprise reflects reliance on the institutional stance. On their view, this is an interpretive framework appropriate to the workplace that does not fit the informal setting, so we are surprised when we observe the institutional behaviors manifested. Since the theory-theory attribution of Kate and Eric’s beliefs and desires can be readily carried over to the informal context, the theory-theorist allegedly has no explanation for our surprise.
A theory-theorist can readily offer a reply. Nothing in the theory-theory requires attributing the same belief–desire profile across contexts. The theory can be tuned so that work and leisure contexts carry different default expectations. Contemporary defenses of theory-theory often stress this flexibility, especially in cases that are unexpected or involve idiosyncratic agents (Jara-Ettinger, Reference Jara-Ettinger2019; Jern, Lucas, & Kemp, Reference Jern, Lucas and Kemp2017). Kate can be treated as just such an agent, which would motivate a more complex attribution of her beliefs and desires. For example, she may have confused beliefs about the boundaries of an assistant’s obligations. On this approach, surprise arises because Kate behaves as if the work context belief–desire profile applies in a leisure setting.
Both the institutional stance and the theory-theory have an explanation for the data. A way out of this standoff is to appreciate that the institutional stance is attractive because it explains the surprise without relying on unconstrained enrichment of a mentalistic theory. More generally, the motivation for the institutional stance consists in highlighting behavioral phenomena that resist, or are less parsimoniously explained, by the mentalistic stance. But the theory-theory always has resources to accommodate the instances that motivate the institutional stance. This flexibility is precisely why surprise cases do not, by themselves, force a choice between theory-like mentalizing and the institutional stance. Jara-Ettinger and Dunham effectively concede as much when they note that a mentalistic interpretation is available (e.g., sect. 6.3.2), but that it requires increasingly complex and intuitively implausible mental-state attributions. They contrast this with the institutional stance, which makes the same interactions appear seamless by appeal to role-based expectations. The institutional stance is therefore appealing not because it refutes the theory-theory on its own terms but because it explains the surprise without relying on open-ended enrichment of a tacit mentalistic theory. The worry about the theory-theory is not complexity per se, but case-by-case accommodation without any independent constraints on the theory. And there is a ready alternative to the theory-theory version of mentalism in the simulation theory.
Accordingly, I propose a division of labor. In structured contexts shaped by roles and norms, the institutional stance plus script knowledge does much of the explanatory work that is often assigned to theory-like mentalizing. In less structured contexts, simulation-based mindreading supports flexible and agent-specific interpretation. On this picture, the institutional stance does not primarily compete with mentalism as such. It mainly competes with a theory-like construal of mentalism. The range of cases in which theory-theory is the most parsimonious option is therefore considerably narrowed. These are cases in which behavior is genuinely novel or nonstandard and is not readily explained by folk physics, as when someone trips and stumbles, not captured by a familiar script, and not made predictable by institutional role structure. They are also cases in which simulation is unreliable because the target is dissimilar, atypical, or otherwise not well modeled by first-person redeployment.
A second motivation Jara-Ettinger and Dunham offer is computational economy. They argue that role-based reasoning can reduce the cognitive costs and uncertainty of mentalizing and can reduce the online computational burden of tracking beliefs and desires in real time (sect. 6.3.1). This is a useful point, but it further highlights what is missing from their picture of the mentalistic stance. Simulationist accounts were motivated in part by the same computational contrast, since simulation predicts others by redeploying first-person resources rather than by applying an increasingly complex theory (Cruz, Reference Cruz1998). The institutional stance and simulation theory therefore look like parallel forms of cognitive offloading that apply in different domains.
Jara-Ettinger and Dunham suggest that mentalism combined with the institutional stance yields richer possibilities for representing and creating institutional structure (sect. 6.3.3). Simulation-based mindreading can play that same enriching role, since it supports agent-specific expectations about how particular individuals will interpret and negotiate institutional roles, without committing us to the theory-theory.
Once we acknowledge how readily the theory-theory can be modified, its treatment of the authors’ surprise cases becomes difficult to state without an overly permissive enrichment strategy. If the theory can be freely revised to match each surprise, it no longer looks like the kind of stable, learnable resource that could guide online interpretation. The institutional stance treats the same phenomena by deploying independently learnable representations of social structure in terms of roles and norms, rather than by case-by-case elaboration of an underlying theory. It looks stronger still once simulation-based mindreading is counted as a mentalistic alternative.
Financial support
None.
Competing interests
The author declares no competing interests.