Walking down the street, someone you don’t recognize makes eye contact, smiles, and waves while excitedly approaching you. You panic: Should I know this person? What do they want? What should I do? Then you see their clipboard and the non-profit logo on their t-shirt. You avert your gaze and scurry by.

Cases like this show how everyday social cognition efficiently integrates both institutional and mentalistic information as it navigates us through complex social interactions. The stranger’s clipboard and t-shirt provide crucial cues about their social role – a canvasser for a non-profit organization. Grasping these cues enables you to infer the mental causes of their behavior. You conclude that they are a stranger who does not really recognize you, and that their friendliness is a charade. Rather, they are enacting a familiar social script because they want to persuade you to sign on to their cause. (But you aren’t fooled.)

Jara-Ettinger and Dunham’s account of the institutional stance leaves plenty of room for such collaborations between mentalistic and institutional reasoning. However, their account goes further, positing that in some social interactions, we may rely entirely upon the institutional stance. This is nicely illustrated in the examples of Kate and Eric, of Laurie, Paul, and Alex, and of our vending machine-like interactions with baristas, which depict contexts where we can reliably predict and interpret behavior using only structured role representations, and where relying upon mentalizing would be unnecessary, cognitively inefficient, and even misleading.

We are mildly skeptical about this aspect of their account. We suspect that social cognition is isotropic (Fodor, Reference Fodor1983): it holistically integrates all kinds of information when predicting and interpreting behavior – including mentalistic and institutional cues, but also cues about social group membership, artifacts and their functions, social norms, the broader physical environment, etc. Different kinds of information might be more relevant in certain contexts, but we remain unconvinced that it’s useful to theorize about this in terms of distinct socio-cognitive “stances.”

But we are curious to hear more. More specifically, we wonder what kinds of contextual factors might lead us to rely solely (or just more heavily) upon the institutional stance rather than mentalizing. The authors give some hints, highlighting how certain proximal cues – uniforms, verbal disclosures of institutional status, linguistic markers like the t/v distinction, and more subtle signs of automatic, norm-governed, or scripted behavior – might activate institutional reasoning. They also suggest that reliance upon role-based reasoning will be moderated by familiarity with the institution in question. We think there could be more to the story.

Here is one possibility: perhaps, all else equal, we should expect preferential reliance on the institutional stance in familiar institutional contexts. This idea still does not generate very useful predictions, though, as all else that could be relevant is rarely equal. There are many familiar institutional contexts in which grasping someone’s role leaves open crucial questions about how to interact with them that can only be answered through sophisticated mentalistic reasoning (e.g., a student wondering how to respond to a stern teacher, or faculty strategizing about how to handle an erratic provost). Institutions are sites of highly complex social interactions, and we typically need all of the socio-cognitive tools at our disposal to navigate them. Maybe some institutional contexts require little or no mentalizing, but it’s not at all obvious which these will be.

So, it is unlikely that institutional reasoning dominates mentalizing in every familiar institutional context. What Jara-Ettinger and Dunham need, then, is a more detailed theory (and set of predictions) about when and why certain institutional contexts invite more or less mentalizing than others.

Our suggestion: the predictions and interpretations produced by mentalizing and institutional reasoning differ greatly in terms of their specificity. Mentalistic reasoning is individualistic: it most naturally explains why a particular individual is engaged in a specific action and is well suited to predict what they are likely to do (Alice likes to surf in the mornings, so she’ll schedule her work meetings in the afternoons). In contrast, institutional reasoning is inherently generic: it applies generally to whomever happens to be occupying a kind of role (cashiers scan food items while customers place them in bags; servers ask for drink orders before food orders). This suggests that mentalistic reasoning will be more useful and reliable in contexts where the target is known to have greater scope for individual agency, while role-based reasoning will be more useful and reliable in contexts where that scope is constrained or absent.

Notably, this prediction also holds for target individuals who are deeply ensconced in an institutional context. Some roles afford greater scope for individual agency than others. Often, these sit higher on the institutional hierarchy: bosses have more agency than workers; customers have more agency than service staff; generals have agency more than infantrymen. The more institutional power an individual has, the more they are able to bring the idiosyncrasies of their own personalities to bear on their decisions and actions, and the more predicting their behavior, even within the context of their institutional role, will require reasoning about their particular attitudes and traits. Conversely, the less institutional power an individual has, the more their behavior will be constricted by their institutional role, and hence the more apt and reliable role-based representations will be for predicting their behavior.

Our prediction aligns with the spirit of many examples the authors use to illustrate the institutional stance in action, from Eric the hapless flunky, to the interchangeable customer service workers Alex and Paul, to the barista-as-vending machine. These are all low-status roles that afford little to no individual agency. Little personalized knowledge of their occupants’ mental states is needed to predict what they will (or, perhaps, must) do.

Our suggestion represents only one route the authors might pursue; surely there are others. But we think it exemplifies the sort of amendment that will generate more specific predictions about when we rely more heavily upon the institutional stance: a more detailed theory of what, respectively, role-based predictions and mentalistic predictions are good for.