Jara-Ettinger and Dunham propose that the institutional stance is supported by a generative grammar on the basis of (i) the fact that we can represent and create an unbounded number of arbitrary institutional structures; (ii) the existence of a unique type of representation that is specialized for thinking and reasoning about institutions; and (iii) the fact that these resources can be used to think about social collectives from an institutional perspective whether or not they actually have institutional structure. Similar arguments can be made for other stances (e.g., essentialist and mentalist) potentially leading to the suggestion that our conceptual systems consist of multiple generative grammars. Such a view is problematic for multiple reasons.

I illustrate the problems for such a view by considering the institutional stance, though the problems apply equally to other stances. First, the institutional stance is not self-contained. Institutions cannot be fully characterized without making reference to representations of things that are not institutions (e.g., universities have buildings; baristas make coffee). Second, whereas Jara-Ettinger and Dunham do a great job of characterizing the ways in which institutional representations are distinctive, the institutional representations must also support modes of thought that are shared with concepts in other domains. For example, the concepts that represent institutions and their roles support instance-of-kind and kind perspectives, just as non-institutional concepts such as dog and table do. Thus, the concept barista allows us to think and talk about particular persons as baristas as well as to think and talk about a particular kind of profession and thus baristas generally. Whereas instances of the kind barista may be characterized via predicates such as male or female, the kind itself cannot. On the other hand, baristas may be said to be widespread, even though instances of the kind baristas cannot be so characterized. Furthermore, whereas two instances of an institutional, or non-institutional, kind can be qualitatively identical, differing only numerically, two institutional, or non-institutional, kinds cannot differ only numerically. Thus, two baristas, like two dogs, can be qualitatively identical but numerically distinct, but baristas and schmaristas, like dogs and schmogs, cannot be distinct kinds if they are merely numerically distinct, but qualitatively identical. In these, and other formal characteristics, kind and instance-of kind representations of institutional kinds do not differ from such representations for non-institutional kinds such as dog and table. If our conceptual systems contain multiple generative grammars, these shared conceptual capacities would have to be built into each grammar separately, a decidedly unparsimonious possibility. Finally, we can form an unlimited number of coherent thoughts involving both institutional and non-institutional kinds and their instances (e.g., the school does not allow dogs in the classroom) suggesting the need for a common format in which we can combine both institutional and non-institutional concepts.

A better solution than positing multiple generative grammars is to have a generative grammar that can capture both the unity and the diversity found in the ways we think and reason about things from different content domains. One attempt to formulate such a theory can be found in the generative theory of conceptual form (Prasada, Reference Prasada2026). According to that theory, all our concepts provide perspectives (stances) for thinking and talking about things and their properties. The perspectives are encoded in the formal structure of concepts – their conceptual form – which provides instructions for how to think about the contentful properties of the concept. Kind concepts are constituted of three types of formal components, a generative component, a descriptive component, and indexical elements. Kind concepts from formally distinct domains are generated via different combinations of the formal components of kind representations and how those components are realized. Prasada (Reference Prasada2026) develops a fragment of the theory of conceptual form which generates kind representations for a wide variety of non-social kinds. It also suggests that the theory may be extended to generate social kind concepts by allowing that instances of social kinds are constituted of a multiplicity of persons, with some social kinds being constituted by persons in roles (e.g., family), while others do not require roles (e.g., community), and with some allowing independence between roles and the persons occupying those roles (e.g., committee) and others not (e.g., family). Such extensions capture distinctive properties of formally distinct classes of social kind concepts. Those distinctive elements work in concert with formal components that characterize kinds across content domains. For example, both living kinds and artifact kinds are partially characterized via principled connections between the kind and certain functional characteristics (Prasada & Dillingham, Reference Prasada and Dillingham2006; Prasada, Reference Prasada2017, Reference Prasada2026). Given the normative dimension of principled connections and the relevance of functions for characterizing non-institutional kinds, it appears that the machinery needed for characterizing non-institutional kinds is well suited for characterizing institutional kinds as well as the roles of which those kinds are constituted. Incorporating Jara-Ettinger and Dunham’s analysis of institutional kinds with the theory of conceptual form is likely to yield new insights into the mechanisms that guide our thoughts about institutions as well as other kinds of things. The resulting theory would have the characteristics that motivate Jara-Ettinger and Dunham’s proposal that we have a generative grammar for institutional concepts without the drawbacks of thinking of it as being distinct from the formal dimension of commonsense kind concepts more generally. On that view, the institutional stance would be one of the many perspectives for thinking and talking made available by the generative grammar of conceptual form.