1. Introduction

There is a widespread narrative in contemporary Anglo-American literature on the history of epistemology that, until Gettier’s seminal paper (1963), the concept of (propositional) knowledge—since Plato—was universally understood as justified true belief (JTB). This narrative has recently been challenged, convincingly, in my view, by scholars such as Antognazza (2015, 2020), Dutant (2015), and Le Morvan (2017, 2023). This shift in the scholarly narrative provides an opportunity to reexamine the interpretations of historical figures shadowed by this narrative, scrutinize their underlying assumptions, and consider how these assumptions influence our understanding of historical issues. If the traditional narrative has indeed exerted a misguided influence, we should not only revise it but also propose a more accurate alternative. In this regard, Locke’s theory of knowledge serves as a pertinent case study.¹

Lex Newman (2007) argues that Locke’s view of knowledge can largely be reduced to the justified true belief (JTB) account.² In his article ‘Locke on Knowledge’ (2007), Newman begins by outlining two key characteristics of Locke’s definition of knowledge—the perception of (dis-)agreement and its restriction to ideas—before addressing the question of how Locke’s official definition of knowledge relates to the JTB account.³ At first glance, Locke’s definition appears incompatible with JTB: as Newman notes, ‘the perception of agreement as wholly constitutive of knowledge’ (2007: 319) seems to have little connection to the JTB framework.

Newman divides the prima facie divergence between Locke’s definition of knowledge and the JTB account into verbal and substantive senses. The verbal difference concerns terminology and is of minor significance; it can be easily resolved by substituting Locke’s term ‘assent’ with ‘belief’. Under this substitution, knowledge as justified true assent neither alters the core of JTB nor seems alien to Locke’s framework. However, the substantive difference remains: as Newman notes, the doxastic state ‘is in no way constitutive of knowledge’ (2007: 320). Nevertheless, Newman argues that this difference lacks practical significance because ‘Locke holds that the mind always assents to what it knows’ (2007: 320–21). Thus, Locke’s broader definition of knowledge can still be mapped onto the JTB account. In a lengthy footnote, Newman (2007: 321n.14) further emphasizes that his conclusion relies on an assumption of the so-called ‘two-act theory’ regarding Locke’s view of judgement and propositions. It claims that Locke has a clear awareness of three distinct cognitive elements: (i) the justifying apprehension (the first act); (ii) the true proposition (the agreement or disagreement of ideas) as the object/content of act;⁴ (iii) the mind’s assent/dissent to that proposition (second act).

In what follows, I will primarily challenge the two-act interpretation and argue that, for Locke, apprehension/perceiving (the first act) is identical to knowing/assenting (the second act), and that the mental proposition is not an object prior to any act of assent or dissent. In this process, I will also clarify several foundational concepts in Lockean epistemology. Moreover, I will contend that, even if the two-act reading holds some plausibility, the fundamental divide between Locke’s conception of knowledge and the justified true belief (JTB) account remains irreconcilable.

2. An Overview of the Debate

The debate over the one-act versus two-act theory centres on whether Locke conflates the process of constructing or entertaining a proposition with the act/state of holding a specific attitude toward that proposition. Geach, who first criticized this conflation, describes the confusion of an unasserted proposition with assertion as a ‘monstrous and unholy union’ (Geach 1962: 51), highlighting its serious implications for understanding conditionals and disjunctions.⁵ In response to Geach’s critique, one might defend Locke’s position in four ways:

1. Admit that Locke conflates the two acts and that Geach’s critique is valid, but argue that this conflation is largely due to the limitations of pre-Fregean logic, making Locke’s error excusable. This view is supported by Owen (2007: 415–16) and Hill (2008: 200–201).

2. Defend the one-act interpretation, arguing that there is no confusion or blame to be assigned. The one-act theory is neither a mistake nor does it lead to absurd consequences within Locke’s philosophical framework or that of his contemporaries. Marušić (2014) has recently provided a strong defence of this position, and Weinberg (2016: 56–57; 2021: 4n.8) and Powell (2018: 27) have endorsed her arguments.

3. Claim that Locke developed an embryonic account of propositional attitudes and should not be interpreted as making such an obvious error. This view is supported by Ott (2004: 34ff.), Newman (2007: 321), and Anstey (2021: 41ff.).

4. Concede that Locke must be interpreted as a two-act theorist, but argue that this does not require imposing a Fregean model on his work. Instead, Locke’s two-act theory is non-Fregean and does not aim to address the problem of propositional attitudes. Van der Schaar (2008) adopts this position.⁶

I will defend the second interpretation by critically engaging with the fourth. Given that the first group also endorses a one-act theory, many arguments advanced by Owen and Hill apply equally to the second position. Besides, I find considerable merit in van der Schaar’s non-Fregean approach to Locke’s epistemology and have drawn significant insights from her work on his concept of knowledge. Some puzzles related to the one-act thesis may be better clarified by adopting her model; apart from the part that supports the two-act theory, van der Schaar’s arguments are fully compatible with, or even better aligned with, the second position.

3. One-Act Theory

Before proceeding, two preliminary remarks are necessary. First, much of the debate relies heavily on the connection between Locke’s theory of judgement and Arnauld and Nicole’s Logic or The Art of Thinking (1662/1996). While discussing Locke’s one/two-act theory in this context is insightful, I cannot explore the Port-Royalians’ views in detail here. Instead, I will assume—without arguing—that the Port-Royalians endorse the one-act theory⁷ and that their influence on Locke is significant.⁸

Second, I do not aim to provide a systematic analysis of the pros and cons of the one-act theory or to defend my support for it in full range. Marušić (2014) has already done excellent work in this area, and there is no need to repeat her arguments. My goal, therefore, is largely focused on identifying the weaknesses and misleading aspects of her argument and proposing a way to address the resulting perplexities.

We now begin with an examination of three central concepts in Locke’s epistemology.

4. Judgement: Propositional Attitude or Judgement-Candidate?

Beyond the aforementioned sense of judgement, scholars frequently discuss another use of the term: a generic notion of ‘taking-to-be-true’ or ‘holding-to-be-true’. This sense of ‘judgement’ is consistent with its modern epistemological usage—judging or believing something to be true or false. It is this notion that I take issue with. In my view, the difficulty with Owen and Marušićs’ one-act interpretation stems from their problematic attribution of this concept to Locke as a propositional attitude.

Owen argues that, although Locke never explicitly uses the term ‘judgement’ to address this broader sense of judgement, he is indeed concerned with it. Marušić describes this broader use as a kind of doxastic attitude: ‘coming to believe’. Both scholars contend that Locke posits a generic attitude of taking-to-be-true that applies across cases of both knowledge and judgement.²³ I do not deny their supposition that, through knowing or judging something, we are taking it to be true, nor do I deny that Lockean affirming or denying constitutes a mental act of taking-to-be-true. What I reject is the explanation of this ‘taking-to-be-true’ as a modern notion of doxastic or propositional attitude, which inevitably involves—whether implicitly or explicitly—a modern concept of proposition.²⁴

What, then, is the serious problem with applying the post-Fregean model of propositional attitude while associating the one-act theory with Locke?²⁵ If, as one-act theorists argue, every act of forming a proposition inherently involves believing its truth or falsehood, then there is no room for merely conceiving a proposition—such as understanding or considering it without adopting any attitude. It seems that Locke, like all of us, requires the ability ‘to consider, suppose, or entertain propositions prior to committing ourselves to their truth or falsity’ (Owen 2007: 416). Owen (2007: 416–17) attempts to resolve this issue by introducing the notion of degrees of probabilistic judgement: entertaining a proposition is positioned as a midpoint between full belief (assent) and disbelief (dissent). Marušić (2014: 272) follows this approach. However, van der Schaar’s objection is compelling: if presumption (judgement) ‘were the genus of all intellectual, epistemic attitudes toward a proposition, there would be no reason to exclude knowledge as a species of this genus’ (van der Schaar 2008: 339). Yet, for Locke, judgement is fundamentally distinct from knowledge and cannot serve as its genus. If this is the case, then, following van der Schaar, we should acknowledge that on some occasions Locke does seem to presuppose a neutral proposition prior to knowing and judging, and thus embrace a two-act theory.

However, I do not believe this view can be sustained. As we have seen earlier, the Lockean proposition is process/production-ambiguous, mind-dependent, event-like, and takes the full form of a declarative sentence. How can these characteristics be reconciled with the post-Fregean realist view of logic, which treats propositions as mind-independent, object-like entities expressed in the linguistic form of a that-clause? Van der Schaar accurately identifies the crux of this problem and proposes her traditional alternative to the modern concept of ‘proposition’. According to her, ‘proposition’ is ambiguous, potentially denoting three things: (1) the act of joining or separating mental ideas; (2) the result or product of this act; and (3) the judgement-candidate. While the first two are relatively clear, the third concept is crucial, as it leads van der Schaar to her two-act interpretation.

In brief, a judgement-candidate is an unasserted proposition—a linguistic meaning-entity that can be asserted but need not be. It takes the form of a declarative sentence with an indicative mood yet lacks assertive force. Unlike Frege’s Gedanke, which, as a Platonic abstract entity, exists beyond space and time and adopts a that-clause structure, a judgement-candidate differs primarily in its agent-relative correctness.²⁶ For instance, while the truth of Goldbach’s conjecture, in a Platonic sense, is wholly independent of any cognitive agent, a judgement-candidate’s validity—though potentially pre-existing, as with the conjecture before Goldbach—remains tied to an agent capable of demonstrating it.²⁷

Compared to the post-Fregean model, van der Schaar’s ‘troika’—judgment as act, candidate, and product— resonates more closely with Locke’s thought. First, both ‘proposition-made’ and ‘judgement-candidate’ take the linguistic form of declarative sentences. Second, they do not presuppose a fully mind-independent meaning-entity. Third, and relatedly, the ‘true’ in ‘taking-to-be-true’ does not signify an epistemically unconstrained property of such an entity; thus, there is little justification for imposing a post-Fregean propositional attitude framework on Locke. Yet this alone is insufficient: even a moderate meaning-entity like the judgement-candidate is not an essential presupposition for Locke’s account. In my estimation, van der Schaar errs in ascribing to Locke the notion of ‘judgement-candidate’—a quasi-Fregean Gedanke—and thereby concluding that he endorses a two-act theory of judgement.

I offer four grounds for this critique.

First, the ‘judgement-candidate’, conceived as an ontologically neutral meaning-entity, though an improvement over Gedanke, remains alien to Locke and the late-scholastic tradition he inherits. In late-scholastic logic and semantics, ‘proposition’ (propositio or enuntiatio) does not denote a sentence-level meaning-entity; rather, it is conventionally classified into three types—spoken, written, and mental²⁸—where its signification is merely the aggregate of its categorematic terms’ meanings.²⁹ Locke himself does not develop a semantic theory of propositional compositionality or signification.³⁰ Given his discussion of mental propositions, which closely aligns with Ockhamist doctrine,³¹ it is implausible to assume he would accept an abstract entity like the ‘judgement-candidate’ as the signification of a mental proposition.³²

Second, the textual evidence from the Essay provided by van der Schaar does not conclusively demonstrate that Locke adopts the concept of a ‘judgement-candidate’. The combination of proposition-as-act and proposition-made is sufficient for understanding Locke’s view. For example, in one passage discussing habitual knowledge, Locke states:

A Man is said to know any Proposition, which having been once laid before his Thoughts, he evidently perceived the Agreement, or Disagreement of the Ideas whereof it consists; and so lodg’d it in his Memory, that whenever that Proposition comes again to be reflected on, he, without doubt or hesitation, embraces the right side, assents to, and is certain of the Truth of it. (IV.i.8, 527–28)

This passage does not, as van der Schaar concludes, confirm the existence of a neutral judgement-candidate prior to knowledge and judgement.³³ All Locke emphasizes here is that in habitual knowledge, the agreement or disagreement of ideas has already been perceived. Once you recall what you have perceived (proposition-made), you immediately affirm it again (perceiving). In this case, the subject matter before the mind is not a judgement-candidate but the product of a propositional act. This interpretation coheres with the term ‘habitual knowledge’: ‘habit’ refers to what people have regularly done—the completed act itself—rather than the object acted upon.³⁴ Locke’s description of the second degree of habitual knowledge reaffirms this point:

[T]he three Angles of a Triangle are equal to two right ones, one, who has seen and clearly perceived the Demonstration of this Truth, knows it to be true, when that Demonstration is gone out of his Mind; so that at present it is not actually in view, and possibly cannot be recollected: But he knows it in a different way, from what he did before … He remembers, i.e. he knows (for remembrance is but the reviving of some past knowledge) that he was once certain of the truth of this Proposition, that the three Angles of a Triangle are equal to two right ones. (IV.i.8, 529; my italics)

When someone forms, perceives, or demonstrates a proposition such as ‘the three angles of a triangle are equal to two right ones’, they possess actual knowledge of it. Later, they may forget the proof or be unable to form or perceive it again, but they can still claim to know it by appealing to the memory-based certainty that they once performed that process—in other words, by appealing to the propositional act they once completed. For example, if asked why they still believe a geometric theorem learned in school, they might respond: ‘Although I forget how, I am still convinced of its truth because I remember proving it before’. It is the memory of the previous action that ensures their conviction; there seems to be no need to interpret ‘proposition’ in these contexts as a pre-existing, mind-independent entity.³⁵ A judgement-candidate is thus a redundant assumption here.³⁶ Therefore, I suggest that we remove the concept of a judgement-candidate from Locke’s framework: his proposition is always known or judged as identical to either the proposition-made (the product of a propositional act) or the act itself.

Third, without presupposing the concept of a judgement-candidate or the modern notion of proposition, would Locke be trapped in an unavoidable predicament? Not at all. I fully agree with Owen when he emphasizes that ‘it is a fundamental error’ for post-Russellians ‘who are immersed in propositional attitude psychology’ (Owen 2003: 17) to impose the distinction between forming and understanding a proposition and taking it to be true onto Locke. If my two objections against van der Schaar are valid, I propose that we go further and more radically reject the assumption of a post-Fregean (Russellian) proposition in interpreting Locke.

Let me first address Owen’s concern. Citing Locke’s statement—‘We have the Ideas of a Square, a Circle, and Equality; and yet, perhaps, shall never be able to find a Circle equal to a Square, and certainly know that it is so’ (IV.iii.6, 540) —Owen (2007: 416) argues that Locke was aware (or should have been aware) of the need to merely consider a proposition. Otherwise, we could not even ask the question, ‘Is a square equal to a circle?’ without first understanding its meaning.³⁷

However, if we set aside the post-Fregean concept of proposition and consider a question-sentence, it becomes clear that, from Locke’s perspective, asking a question simply involves uttering an interrogative sentence with an interrogative mood. In other words, we use words to construct a question-sentence to inquire about the relationship between two ideas. For example, taking any two inherently distinct and unrelated ideas—such as ‘matter and thinking’—we can ask, ‘whether any mere material Being thinks, or no’ (IV.iii.6, 541). It is unnecessary, as Owen imagines, to posit a meaning-entity or sentence-type like ‘a square is equal to a circle’ existing prior to the mind, waiting to be grasped or perceived,³⁸ and then, in a separate act, questioned, desired, believed, assented to, and so on.³⁹

The process, rather, is as follows: through experience (observation or reflection), we acquire separate ideas (such as ‘square’ and ‘circle’), use various categorematic terms to signify them, and form sentences with corresponding moods (interrogative, imperative, indicative, conjunctive, etc.).⁴⁰ Only sentences with an indicative mood qualify as candidates for Locke’s propositions and are connected to his notions of knowledge and judgement. Before these sentences are composed, the objects of thinking and understanding are limited to ideas.⁴¹ The notion of a proposition as a pre-existing meaning-entity, prior to the mind’s construction, is thus an illusory concept imposed on Locke by contemporary epistemologists.⁴² This also explains why scholars search in vain in Locke’s Essay for accounts of other propositional attitudes, such as hoping and desiring. The answer is that desire, joy, and hope are not propositional attitudes for Locke, as he lacks the modern (standard) concept of proposition and propositional attitude psychology.⁴³ Instead, he discusses these concepts—fear, desire, joy, hope, and anger—together in ‘Of Modes of Pleasure and Pain’ (Essay II, xx), categorizing them as passions caused by ideas.⁴⁴

Finally, van der Schaar consistently argues that her concept of a judgement-candidate reflects the traditional understanding of ‘proposition’ and has its roots in scholastic thought, particularly in Ockham’s philosophy. I find this connection inaccurate and misleading.

First, the inaccuracy: van der Schaar claims that Ockham (in Quodlibetal Questions, Quodlibet 5, Questio 6) distinguishes between propositio apprehensiva and propositio iudicativa—a proposition merely entertained and an asserted proposition.⁴⁵ This reference is inaccurate. Ockham never uses the terms propositio apprehensiva/iudicativa in this passage.⁴⁶ The title of the fifth Quodlibet, question 6, is: ‘utrum actus apprehendendi et actus iudicandi differant realiter (Is an act of apprehending really distinct from an act of judging?)’ (my italics). The focus here is on the distinction between propositional acts of apprehension and acts of judgement—or, in other words, between entertaining and judging a proposition. For Ockham, as echoed by Locke, a mental proposition is primarily treated as a mental act, not as an object or content of the act. Ockham identifies four types of acts: two apprehensions and two judgements (assents),⁴⁷ none of which corresponds to van der Schaar’s description of a judgement-candidate as a ‘sentence type together with its meaning’ (van der Schaar 2015: 316). Moreover, Ockham’s robust nominalist ontology only admits individual substances and individual qualities, making it highly unlikely that he would accept such an abstract meaning-entity.⁴⁸

Now, the misleading part. By distinguishing two kinds of apprehension and two kinds of assent, Ockham appears to endorse a two-act theory. It is reasonable to assume that the apprehensiva/iudicativa distinction is one of the philosophical commonplaces Locke shares with Ockham’s doctrine. However, the matter is more complex. To avoid delving into intricate issues, I will highlight two key points. On the one hand, the view that knowing or judging is identical to propositional formation or apprehension (the one-act theory) was not uncommon during the late-scholastic period.⁴⁹ On the other hand, regarding this theme, Locke seems less likely to side with Ockham and Buridan than with Gregory of Rimini, another prominent medieval scholastic thinker. Gregory is generally regarded as a proponent of proposition-judgement identity theories and, on several issues, disagrees with his contemporaries, Ockham and Buridan.⁵⁰

Locke’s affinity to Gregory also provides an opportunity to address a significant puzzle that scholars have long struggled with: the alleged absurdity of Locke’s one-act theory in explaining complex propositions such as conditionals and disjunctions. The charge of absurdity regarding conditionals can be summarized as follows: a conditional can be accepted as true even if its antecedent is not accepted as true. However, under the one-act interpretation, forming a conditional requires accepting both the antecedent and the consequent as true, since both are propositions. This would render the truth-functionality of conditionals impossible.

Scholastic thinkers were well aware of this problem. One of their responses is that the antecedent and consequent of a conditional, as the principal components of a hypothetical proposition, are not themselves propositions and therefore are neither true nor false.⁵¹ But what are they, then? To clarify this, we need to introduce a dichotomy between non-ultimate and ultimate mental language.⁵² I will borrow Nuchelmans’ words to illustrate: the former consists of ‘those mental terms and propositions that are natural likeness of spoken or written expressions’, and the latter ‘contains the concepts and thoughts which render the expressions of a particular language meaningful and are natural signs of things’ (Nuchelmans 1980: 16; my italics).

The first group of terms is non-ultimate in the sense that they cannot produce proper ‘thoughts corresponding to their expressions’. Instead, they only ‘form indirect mental images of the sounds and letters’, which fail to achieve ‘ultimate cognitive success’ (Nuchelmans 1980: 16). This is akin to an English speaker parroting German words or sentences, or someone speaking heatedly, using words or sentences without truly understanding them. The second group of terms is ultimate because they carry conventional meaning based on what they signify and are universally understood.

To put it succinctly (though perhaps not entirely accurately), non-ultimate terms/propositions function as signifié (sound-images), while ultimate terms/propositions function as signifiant (concepts).⁵³ Returning to the conditional, its antecedent and consequent cannot properly be called propositions or are only propositions in the non-ultimate sense, as they are pure mental images of spoken or written words. For example, in the sentence ‘the tree is green’ as the antecedent of a conditional, we are not using ‘tree’ or ‘green’ to refer to any concept of an actual tree or green object but only to the concepts of the words ‘tree’ or ‘green’ themselves. Consequently, no proposition in the strict sense is formed, and no issue of assent or dissent arises. Locke himself never addresses conditionals, nor is there direct evidence that he was familiar with this view. However, in response to the alleged absurdity perceived by contemporary philosophers, I believe it makes far more historical and theoretical sense to interpret this issue within a scholastic framework rather than through the lens of twentieth-century theory.⁵⁴ Thus, Locke, influenced by Gregory of Rimini via Smiglecius, could easily have adopted this position had he considered the issue.⁵⁵

5. Concluding Remarks

To summarize, I have argued that, for Locke, the act of knowledge and the propositional act are fundamentally the same (the one-act theory), while judgement, as an inferior form of knowledge, constitutes a categorically distinct act. I then addressed a significant difficulty with Owen (2007) and Marušićs’ (2014) one-act thesis: the impossibility of merely conceiving or inquiring into a proposition without affirming or denying it. Echoing van der Schaar (2008), I argued that their proposed solution to this problem is flawed. Furthermore, I agreed with van der Schaar’s view that Locke’s conception of the proposition does not permit us to resolve the predicament surrounding it by situating it within the contemporary psychological framework of propositional attitudes, given his lack of a post-Fregean understanding of the concept.

Nevertheless, I diverged from van der Schaar’s alternative traditional conception of the proposition as a ‘judgement-candidate’ and her resultant two-act interpretation of Locke, for four reasons:

1. A sentence-level meaning-entity is inconsistent with the philosophical context preceding and including Locke’s era.

2. There is no textual evidence in Locke’s writings to support this assumption.

3. The absence of a Fregean notion of proposition does not place Locke’s thought in a problematic or insurmountable predicament.

4. Van der Schaar’s attempt to link her concept of ‘judgement-candidates’ to the traditional use of ‘proposition’ in the Middle Ages is historically inaccurate and misleading.

Accordingly, I argue that the first step in engaging with the debate over a one-act or two-act theory of Locke is to entirely remove the modern notion of proposition from the discussion. In other words, we should avoid treating Locke’s mental proposition as a self-sustained meaning-entity existing independently of mental construction, and we must cease interpreting Locke’s definition of knowledge through the lens of assumptions embedded in the post-Fregean framework of propositional theory.⁵⁶ Once this point is clarified, the concerns surrounding the one-act interpretation of Locke’s theory of knowledge and judgement can be partially alleviated. What remains is to identify a viable path to address the alleged absurdity of conditionals by drawing on resources from scholastic thought.

If my foregoing arguments hold, Newman’s presupposition of a two-act reading of Locke and his ensuing quasi-JTB claim become unsustainable. Moreover, as noted initially, even if we set aside this debate and allow for a possible two-act interpretation, Newman’s assertion that Locke broadly endorses the JTB account remains unsubstantiated. Locke’s conceptions of knowledge and proposition diverge fundamentally from the post-Gettier framework on three grounds:⁵⁷ (1) in his formal definition, ‘knowledge’ primarily denotes a cognitive act, not a mental state; (2) truth does not constitute an essential element of Locke’s knowledge;⁵⁸ (3) belief is absent from his definition of knowledge, and his ‘taking-to-be-true’ does not align with the contemporary epistemological notion of a generic dispositional attitude.⁵⁹

In a nutshell, Locke’s epistemology defies neat placement within the standard pigeon-holes of the JTB narrative.

Acknowledgements

This work was supported by the Center of Modern Thought, College of Social Sciences, Yangzhou University, Yangzhou, China.

Competing Interests

The author has no competing interests to declare.