Frege–Geach problem
On this page
The challenge, pressed by Peter Geach after Frege, that non-cognitivist theories of moral language cannot explain the meaning of moral sentences in embedded contexts or the validity of ordinary moral arguments.
The problem at a glance
A guide to the entry below, quoting its own words. Skip to the full entry and its sources.
- The reasoning that leads to it , Standard form of the problem, as given in the entry; the entry does not say who numbered the steps
-
Geach's example is an argument with the logical form of modus ponens. Premise
- : if tormenting the cat is bad, getting your little brother to do it is bad. Premise
- : tormenting the cat is bad. Conclusion
- : getting your little brother to torment the cat is bad. (the entry continues below)
- The puzzling result
- As van Roojen presents it, the problem is less a flat refutation than an explanatory burden: once propositional content is denied to moral sentences, the simple explanation of consistency, inconsistency and valid inference is no longer available, and the non-cognitivist owes a substitute.
Objections and responses
- Principal replies
-
Simon Blackburn treated conditionals and other embeddings as expressions of higher-order attitudes, first as complex beliefs (1971) and then as further non-cognitive commitments (1984), so that someone who accepts the premises and rejects the conclusion is incoherent. Allan Gibbard modelled the content of normative judgements with norm-acceptance (1990) and then with plans and hyperplans over 'fact-prac' worlds (2003). Mark Schroeder (2008) distinguished inconsistency in which one attitude type is directed at inconsistent contents (A-type) from inconsistency in which incompatible attitude types are directed at the same content (B-type), and argued that A-type models are preferable if they can be built. Minimalists about truth, following Frank Ramsey and Horwich, can explain embedding under 'is true', though van Roojen notes that this handles only that one embedded context; hybrid theories assign moral sentences a descriptive component that generates most of the required logical relations, supplemented by an account of the non-cognitive component.
- Further discussion in the entry
-
Principal objections to the replies. Hale (1986) and others, including Schueler, Brighouse, Zangwill and van Roojen himself, argue that such non-cognitivist proposals conflate genuine inconsistency with mere pragmatic incoherence. Dorr's wishful-thinking objection holds that inferring a factual consequent from a conditional and an accepted moral antecedent would change one's beliefs on the strength of an attitude alone; van Roojen reports that "the licensed inference is really a form of wishful thinking" on such views, records David Enoch's early response (2003) and Schroeder's criticism of it (2011), and argues (2019) that the objection needs more than the Frege–Geach problem to succeed. Van Roojen (2005) and Schroeder (2009) question how the descriptive content of hybrid theories is fixed.
Coordinate 3.1.2Coordinate 6.10Peter GeachGottlob FregeJohn SearleW. D. RossSimon BlackburnFrank RamseySources and primary textsRead the full entry
What it means
The Frege–Geach problem is the leading objection to non-cognitivist and expressivist accounts of moral language. It addresses the question whether a theory on which a moral sentence such as 'lying is wrong' serves to express an attitude rather than to state a fact can explain what the sentence means when it is not asserted on its own: when it occurs inside a conditional, a negation, a question or a disjunction.
Origin. Mark van Roojen's Stanford Encyclopedia entry on moral cognitivism and non-cognitivism names Peter Geach (1965) as the source. Geach drew on a point he credited to Gottlob Frege, which is why the problem carries both names; van Roojen quotes Geach's own formulation of what is called the Frege point: "A thought may have just the same content whether you assent to its truth or not". John Searle (1962) raised a version independently, and the entry credits W. D. Ross (1939) with an earlier statement.
Structure. Geach's example is an argument with the logical form of modus ponens. Premise (1): if tormenting the cat is bad, getting your little brother to do it is bad. Premise (2): tormenting the cat is bad. Conclusion (3): getting your little brother to torment the cat is bad. The argument is plainly valid. The problem, in a reconstruction: (a) on a simple non-cognitivist view, the meaning of 'tormenting the cat is bad' in (2) consists in its expressing disapproval; (b) in the antecedent of (1) the same words express no disapproval, since someone can assert (1) without disapproving of anything; (c) so on that view the words mean different things in (1) and (2); (d) an argument whose terms change meaning equivocates and is not valid; therefore (e) the simple view cannot explain the validity of an obviously valid argument. Van Roojen states the general demand as a set of desiderata: an account of embedded moral sentences should be compositional, should preserve logical relations, and should not attribute implausible attitudes to speakers. "Geach thought that the second and third desiderata would be especially hard to accomplish simultaneously."
Intended conclusion. As van Roojen presents it, the problem is less a flat refutation than an explanatory burden: once propositional content is denied to moral sentences, the simple explanation of consistency, inconsistency and valid inference is no longer available, and the non-cognitivist owes a substitute.
Principal replies. Simon Blackburn treated conditionals and other embeddings as expressions of higher-order attitudes, first as complex beliefs (1971) and then as further non-cognitive commitments (1984), so that someone who accepts the premises and rejects the conclusion is incoherent. Allan Gibbard modelled the content of normative judgements with norm-acceptance (1990) and then with plans and hyperplans over 'fact-prac' worlds (2003). Mark Schroeder (2008) distinguished inconsistency in which one attitude type is directed at inconsistent contents (A-type) from inconsistency in which incompatible attitude types are directed at the same content (B-type), and argued that A-type models are preferable if they can be built. Minimalists about truth, following Frank Ramsey and Horwich, can explain embedding under 'is true', though van Roojen notes that this handles only that one embedded context; hybrid theories assign moral sentences a descriptive component that generates most of the required logical relations, supplemented by an account of the non-cognitive component.
Principal objections to the replies. Hale (1986) and others, including Schueler, Brighouse, Zangwill and van Roojen himself, argue that such non-cognitivist proposals conflate genuine inconsistency with mere pragmatic incoherence. Dorr's wishful-thinking objection holds that inferring a factual consequent from a conditional and an accepted moral antecedent would change one's beliefs on the strength of an attitude alone; van Roojen reports that "the licensed inference is really a form of wishful thinking" on such views, records David Enoch's early response (2003) and Schroeder's criticism of it (2011), and argues (2019) that the objection needs more than the Frege–Geach problem to succeed. Van Roojen (2005) and Schroeder (2009) question how the descriptive content of hybrid theories is fixed.
Key philosophers and texts. Gottlob Frege; Peter Geach (1965); John Searle (1962); W. D. Ross (1939); Simon Blackburn (1971, 1984); Allan Gibbard (1990, 2003); Mark Schroeder (2008).
Current status. Every leading non-cognitivist has offered a solution, but van Roojen judges that the task is difficult and "generates much controversy"; there is no agreed solution, and any expressivist semantics is still judged by how it handles embedded moral sentences.
On this site it belongs to Coordinate 3.1.2 (Moral Semantics) and bears on Coordinate 6.10 (Philosophy of Language).
See also the glossary entries Expressivism, Noncognitivism, Validity and Sense and reference.
The questions it belongs to
- 3.1.2 Moral SemanticsDoes "wrong" describe an act, or do something to the listener?
- 6.10 Philosophy of LanguageHow do noises come to be about things?
Thinkers
To read
- Being Good: A Short Introduction to Ethics — Simon Blackburn
- F. P. Ramsey: Philosophical Papers — Frank Ramsey (D. H. Mellor, ed.)
- Minds, Brains and Science — John Searle
Sources and further reading
Cite this page
Responsibility: Coordinates of Thought Editors. Each statement was drafted from the sources listed above and checked against them in a separate pass before publishing; this page is not peer-reviewed by outside specialists. Source-checked 12 November 2026. How to cite this site · How we write and check.
APA (7th edition)
Coordinates of Thought Editors. (2026, November 12). Frege–Geach problem (Arguments). Coordinates of Thought. Retrieved [date accessed], from https://coordinatesofthought.com/arguments/frege-geach-problem/
MLA (9th edition)
“Frege–Geach problem (Arguments).” Coordinates of Thought, 12 Nov. 2026, coordinatesofthought.com/arguments/frege-geach-problem/. Accessed [date accessed].
Chicago (17th edition)
Coordinates of Thought Editors. “Frege–Geach problem (Arguments).” Coordinates of Thought. Source-checked November 12, 2026. https://coordinatesofthought.com/arguments/frege-geach-problem/.
BibTeX
@misc{cot-glossary-frege-geach-problem-2026,
author = {{Coordinates of Thought Editors}},
title = {Frege–Geach problem (Arguments)},
howpublished = {Coordinates of Thought},
year = {2026},
month = nov,
url = {https://coordinatesofthought.com/arguments/frege-geach-problem/},
note = {Source-checked 12 November 2026}
}
RIS
TY - ELEC AU - Coordinates of Thought Editors TI - Frege–Geach problem (Arguments) T2 - Coordinates of Thought PB - Coordinates of Thought PY - 2026 DA - 2026/11/12 UR - https://coordinatesofthought.com/arguments/frege-geach-problem/ LA - en ER -
For reference managers (Zotero, Mendeley, EndNote): BibTeX file · RIS file