NOTE-393: Epstein — Social Ontology (SEP)

Contribution

A field survey with a thesis about how the field is organised. The entry treats the many theories of the social as answers to two different questions. One asks what grounds social facts: their constituents or determination base. The other asks what anchors social categories: what sets up the conditions. It argues that much confusion comes from running them together. That division is the author’s, and it is the frame of The Ant Trap (LIT-478).

Key insight

For any social category there are two questions. Given that kosher (or foxtrot, or Supreme Court justice) has the conditions it has, what makes this thing satisfy them? And what put those conditions in place? Collective acceptance, convention, practice and patterns are answers to the second question. Individualism and its rivals are answers to the first. A theory can be individualist about one and not the other.

Assumptions

• Survey stance. The entry reports positions and the main objections to each. It does not adjudicate, except where its organising distinction is at stake.
• Grounding is taken in the standard metaphysics sense (Fine, Rosen, Audi), as non-causal determination, with the entry on metaphysical grounding as background (§2.3).
• The two inquiries are separable. The entry calls the sharpness of the distinction “a matter of some controversy” but uses it to organise §§3–4.

Key results

• Demarcation (§2.1).
  • No agreed mark of the social: Gilbert locates sociality in joint commitment and hesitates over crowds; others deny there is a criterion.
  • “Epstein 2016” separates socially grounded from socially constructed. Crowds are grounded but not constructed, phenotypic racial categories constructed but not grounded, fiat money both.
• Building relations (§2.2).
  • Supervenience gave precise distinctions (local versus global supervenience of the Senate’s properties) but has known shortcomings.
  • Grounding has displaced it.
  • Epstein 2015 treats determination by grounding and construction by anchoring. Schaffer treats both as grounding. Mikkola argues against grounding approaches, and Passinsky answers her.
  • Epstein 2019b argues anchoring is neither grounding nor meta-grounding. Haslanger proposes anchoring as epistemic “scaffolding”. Hawley and Schaffer would reduce it to grounding. Schaffer models the distinction on Dretske’s structuring versus triggering causes.
• Ontology versus causation (§2.3). Ontological determination can be diachronic, as a footprint depends on a past foot-strike and a president on a past election. Epstein 2015 and Schaffer favour diachronic determination; Bernstein opposes it. Searle’s account is synchronic.
• Determination bases (§3).
  • Psychologism (Mill, Popper). Its local-dependence form says group attitudes depend only on members’ attitudes (Gilbert, Tuomela & Miller, Bratman, Wray, Lackey); Bird 2010 and Epstein 2022 challenge local dependence.
  • Atomism; bodies, actions and relations (List & Pettit 2011 widen the base to actions and dispositions).
  • Material building blocks and physicalism. Ontological individualism “has recently come under criticism” (Epstein 2009, Hindriks 2013, Ylikoski, Elder-Vass, Haslanger 2022).
  • Mixed views (Pettit’s externalist psychologism, institutional individualism), socially constituted individuals and practices, holism and flat ontologies (Actor-Network Theory).
• Anchoring theories (§4).
  • Individual mental states (Lockean nominal essences).
  • Community-wide mental states:
    • agreement;
    • Lewis’s convention: solutions to coordination problems with several equilibria, needing no agreement, only beliefs, expectations and preferences;
    • Hart’s accepted rules;
    • Searle’s status functions, imposed by collective acceptance of constitutive rules (“X counts as Y in C”) through declarations, with deontic powers. Searle differs from Hart in three ways: collective rather than individual attitudes, no requirement of behaviour, and one form put in place by declarations.
  • Pattern theories: functional roles, Millikan’s reproductively established families, Boyd’s homeostatic clusters, Hacking’s looping kinds, Mallon.
  • Practice theories, and pluralism. Epstein 2014a: there are several “anchoring schemas”, and anchors are heterogeneous.
  • Language: word meanings, speech acts, discourse.
• Domains (§5).
  • Groups (§5.1): sets, fusions, pluralities, structures, or entities constituted by members. Coinciding groups are a problem for the fusion view, and Epstein 2019a argues there can be distinct groups whose members necessarily coincide.
  • Group minds and agency (§5.2): Gilbert’s joint commitment; Bratman’s meshing intentions; Searle’s primitive we-intentions; Tuomela & Miller’s we-mode; the functionalist model of group agency as a realised “system of practical activity” (Bratman, List & Pettit, Theiner, Epstein 2022); critics who deny groups phenomenal states; Hutchins and Huebner on large complex groups.
  • Institutions (§5.3); race, gender and disability (§5.4); artifacts (§5.5); money (§5.6); law, with Hart’s rule of recognition and Epstein 2015 ch. 7 (§5.7); language (§5.8).

Claims

id	claim	strength	support
C1	Social ontology divides into an inquiry into determination (grounding) and one into construction (anchoring)	the entry’s organising thesis	argued by example (animal sacrifice, foxtrot), §§2.2, 4; called controversial in §2.2
C2	Anchoring is a distinct relation, not grounding or meta-grounding	reported as the author’s view	§2.2.2, cited to Epstein 2019b; three dissents named in a sentence each
C3	Grounding has replaced supervenience as the main tool for determination	reported	§2.2.1, with the shortcomings of supervenience cited
C4	Ontological individualism is widely assumed and false on recent arguments	reported; the entry’s sympathy is clear	§§3.1–3.2, five critics cited, no argument given here
C5	Group attitudes are not exhaustively determined by member attitudes	reported as challenged view	§3.1.1, Bird 2010 and Epstein 2022
C6	Lewis-style conventionalism faces three critiques: not all social phenomena solve coordination problems; it is too mentalistic; it requires behavioural regularity	reported	§4.2.2, Burge 1975, Millikan 2005
C7	Social categories are anchored in several distinct ways, by heterogeneous anchors	reported as the author’s view	§4.5, Epstein 2014a; one example (US legal categories)
C8	Group agency is widely modelled as a functionally realised system of practical activity	reported	§5.2, five sources

Concepts

• Grounding (ontological determination) — the non-causal relation between a social fact and the facts that make it obtain.
• Anchoring — the relation between a frame principle (the conditions for a fact of some kind to be grounded) and the facts that put it in place. The précis gives the example x is kosher: whether x came from a pig grounds it, and what anchors the condition might be religious authority, acceptance, practice or divine command.
• Frame, frame principle (précis) — a frame is a set of possible worlds; a frame principle says how a kind of social fact is grounded across it. Searle’s constitutive rules are frame principles of one form.
• Ontological individualism — the social is exhaustively determined by facts about individual people. Anchor individualism (précis; the entry cites it only by the title of Epstein 2014b) — the parallel individualist claim about how frame principles are anchored. The précis says both are false and argues only the first in the book.
• Local versus global dependence — whether a social fact about some people depends only on them or on a wider population (§3.1.1).
• System of practical activity — the functional organisation of belief, intention, planning, deliberation and action that an agent realises, used for individual and group agency alike (§5.2).

Connections

• Hindriks & Guala (LIT-481, NOTE-395). The entry reports Searle’s constitutive rules as an anchoring theory. Hindriks & Guala reduce them to regulative rules held in equilibrium. On the framework, their rules-in-equilibrium are an answer to the anchoring question: equilibrium plus representation is what puts the conditions for property in place. That is my placement, not the entry’s or theirs.
• Lewis (LIT-473) appears as the anchoring theory of §4.2.2, and Bicchieri (LIT-493) in its list of convention theorists, dated 2005, which matches Cambridge’s print date.
• North (LIT-489, NOTE-392) is cited by his 1990 book and spelled “Douglas” in §5.3.
• List (LIT-401, NOTE-340) and Schwitzgebel (LIT-159, NOTE-131) are cited in §5.2; see Corrections for how Schwitzgebel is placed.
• Griesemer & Shavit (LIT-188) give a typology of collections, conexuses and communities by how members interact. §5.1’s question, whether groups are sets, fusions, pluralities or structures, and Epstein 2019a’s case against simple typologies, are the metaphysical side of the same question. The entry does not cite them.
• Tomasello (LIT-125, NOTE-185) builds obligation on joint commitment and a “we”. §5.2 places joint commitment (Gilbert) as one of four accounts of collective intention and reports that psychologists now study distinctively social attitudes, citing Tomasello 2014, a different work from the one in the record.

Bearing on the record

THEORY-043. THEORY-043 holds that the downward and upward functionalist moves are one move. It finds that what separates them is either an anti-nesting principle or an architectural criterion, and that neither is yet principled. Both routes, as THEORY-043 states them, take the candidate whole to be fixed by its parts: Kammerer’s condition asks whether members consciously represent the whole, and List’s low-Φ argument looks at how members are coupled. The framework, as this entry and the précis state it, bears on that framing in three ways. These are my inferences from those texts, not claims either makes about consciousness. Neither text discusses group phenomenal consciousness beyond §5.2’s two sentences.

• What grounds a group’s states is not only its members. The entry reports a challenge to local dependence: group attitudes need not depend only on member attitudes (§3.1.1; Epstein 2022, not read). The précis says a group’s actions and intentions can be grounded partly in non-members, infrastructure, powers and membership rules. If so, an architectural criterion computed over members and their links, as List’s is, is computed over less than the system whose states are in question. That sharpens THEORY-043’s point that no one has computed the criterion for a group: it also has to be computed for the right group.
• “The group” is not one candidate. Distinct groups can have the same members, even necessarily (§5.1, Epstein 2019a, not read). That is the metaphysical version of the gap Schwitzgebel’s reply opens at Kammerer’s condition A: it is unclear which entity citizens would have to represent (NOTE-381, C2). The framework says the vagueness is not an artefact of the word “United States”. Coinciding groups are a general feature of groups.
• Anchoring is not a third route, but it is a reason to keep the question about experience separate. On the précis’s account, status assignments do not make a group an agent. A group is an agent by realising a system of practical activity. That sides with THEORY-043’s observation that agency survives both directions, and against any view that group agency is conferred. Whether experience could be anchored at all, rather than grounded, the texts do not say. Nothing in them suggests it.

Net: no bearing on promote_when, which asks for a criterion of phenomenal consciousness. The framework adds a constraint on any such criterion: it must say which of several coinciding wholes it applies to, and it must count the non-member grounds of a group’s functional organisation. Reading Epstein 2022 (“Why Group Mental States are not Exhaustively Determined by Member States”, Philosophical Issues 32(1):417–433, not in the record) would test the first point directly.

No instruction for machine-learning practice appears, so nothing here points to the Anthology of the SOTA.

Limitations

• A survey: positions are reported in a sentence or two and objections are listed, not weighed. The anchoring dispute (§2.2.2) gets one paragraph for both sides.
• The organising distinction is the author’s own thesis. Readers who reject anchoring as a distinct relation (Hawley, Schaffer) would organise §§3–4 differently.
• §5 is thin by design; each subsection points to other SEP entries.
• The 2024 revision was read. Whether the 2018 first version made the same claims was not checked.

Open questions

• Is anchoring a distinct relation from grounding? The entry names the dispute and does not settle it (§2.2.2).
• Are group mental states exhaustively determined by member states? Epstein 2022 argues not; the record does not hold it.
• Can the anchoring–grounding split be applied to non-social kinds whose boundaries are set by practice, such as scientific kinds (§4.5 notes pluralism about scientific kinds)?

Corrections

• none to a seeded skim (there was no seed)
• “Epstein 2016” is cited in §2.1 but missing from the bibliography, which lists Epstein 2008, 2009, 2010, 2014a–b, 2015, 2019a–c and 2022. The intended work could not be confirmed from the entry.
• Schwitzgebel 2015 is listed with critics of group minds. §5.2 says Rupert 2014 argues groups cannot have phenomenal states, “(see also Schwitzgebel 2015, List 2018)”. List 2018 (LIT-401) does deny group phenomenal consciousness, on low Φ (NOTE-340, C5). Schwitzgebel 2015 (LIT-159) argues the opposite conditional: if materialism is true, the United States is probably conscious (NOTE-131, C1). Read as “see also for discussion”, the citation is defensible; read as “see also for the same view”, it is wrong.
• §5.3 spells North’s first name “Douglas”; it is Douglass.