Paul Horwich makes the following observation at the beginning of his landmark book Truth:

Perhaps the only points about truth on which most people could agree are, first, that each proposition specifies its own condition for being true (e.g., the proposition that snow is white is true if and only if snow is white), and, second, that the underlying nature of truth is a mystery. (Horwich 1998a, b, c, p. ix [emphasis original])

The aim of Horwich’s minimalist theory (MT) is to turn one of these touchstones signifying the people’s view of truth against the other.Footnote 1 Accordingly, ‘truth is entirely captured by its triviality’ leaving nothing mysterious, unusual, or puzzling about it.Footnote 2 For Horwich, the propositional form of the equivalence schema, (ES):

〈〈p〉is true iff pFootnote 3

is conceptually, explanatorily, epistemologically, and logically fundamental.

Recent work in experimental philosophy and the experimental work of Arne Næss in the 1930s and 1950s show that people’s view of truth can be quite diffuse and can vary according to individual differences (Barnard and Ulatowski 2013, 2019, 2021; Barnard et al. 2017; Reuter 2022; Ulatowski 2017). When we further analyse the data, however, interesting phenomena arise. According to a new interpretation of the data, many of us have a non-inferential commitment to practical forms of ES; it need not be justified by anything more immediately known or anything more obvious to us. It seems, then, that we have preliminary evidence of ES’s epistemological fundamentality. Given that this prevailing assumption of MT is well-supported by the empirical data, I argue that the disposition to accept ES is axiomatic, providing a relatively stable foundation upon which to build a minimalist theory of truth.Footnote 4 Here is how the paper will unfold. §1 is a brief explanation of the empirically-informed methodology that this paper has employed to show that people have an unmediated commitment to ES. This paper analyses publicly available data sets and provides a new interpretation of them to support the view that ES is axiomatic. In Sect. 2, I summarise Horwich’s MT and show how intuitive responses among ordinary language users may be a guide to epistemological fundamentality. Then, Sects. 3, 4, and 5 report empirical data that support Horwich’s MT, from experimental work that spans almost eighty years, i.e., from Næss’ (1938) and (1953) studies to recent experimental work on the nature of truth. Section 6 will address challenges that concern whether it is appropriate to derive conclusions about practical variants of ES that serve as axioms for alethic minimalism from empirical data. That will permit me in §7 to conclude that a view such as Horwich’s MT that takes ES as its epistemologically fundamental starting point may come closest to the claim of a folk theory of truth, people's intuitive responses to practical variants of ES.

1 Methodology

The methodological approach that I employ in this paper assesses publicly available data to determine whether people have a disposition to embrace ES, or something approaching it. The argument here calls upon data reported by recent work in experimental philosophy and a predecessor of that movement, Arne Næss. My analysis gives rise to a novel empirically-informed conclusion. The empirical data provide evidential support for the argument that people are committed to practical variants of ES, which is something that Horwich anticipated in his own work.

In 1938 and 1953, Næss did not have an arsenal of quantitative statistical tests at his disposal. Empirical work in psychology and other social sciences around that time was relatively new, and statistics was merely considered to be a branch of the natural sciences (cf. Hald 1998, 2003; Salsburg 2001; Stigler 1999). In this paper, I provide a quantitative analysis of Næss’ early experimental work to show that non-philosophers have a disposition to accept practical forms of ES.

One may worry that since Næss’ (1938) studies were mostly qualitative, a quantitative analysis of the data set may be impossible to perform. If, for example, Næss’ early qualitative studies were coded by philosophers who share Næss’ views, then any data will likely show that the author’s view is correct. First, the concern over coding may be mitigated by the fact that Næss’ data sets were disseminated by people who did not share his view. Just as in contemporary qualitative studies, Næss employed research assistants unaffiliated with him to “code” the data.Footnote 5 Next, a scrupulous review of Næss’ (1938) monograph reveals that elementary descriptive statistics were reported by him. Næss’ anachronistic writing style, however, has prevented readers from appreciating his results. A careful read of Næss requires one to move nimbly between different sections of the text, much as one might read a choose-your-own-adventure book. Reading Næss (1938) cover-to-cover or sequentially from section-to-section would yield a tremendous misunderstanding of the empirical data he collected. Neither of these worries about analysing Næss’ data, therefore, should concern us.

In addition to the empirical studies by Næss, there is some extant data from more recent empirical studies that may be employed to reinforce the view that practical variants of ES, the axioms of Horwich’s minimalism, are prevalent in available data sets. These empirical studies may be interpreted as replications of Næss’ earlier studies. One should not think of this essay as historical since the data are being put into the service of the conclusion that practical variants of ES is the intuitive view, shared not only by philosophers, such as Horwich, but also by non-philosophers. Thus, the argument of this paper calls upon data in a fresh and exciting way.

2 Minimalism About Truth and the Disposition to Accept ES

Unlike its substantive counterparts, such as correspondence and coherence theories of truth, minimalism denies that there is something mysterious underlying the nature of truth. According to Horwich, we need not look any further than how the truth predicate behaves in natural language for us to appreciate what truth is. At the very least, however, an appreciation of the behaviour of the truth predicate in natural language may include some empirically informed story of how truth is expressed by users of natural language or how they respond to inquiries about truth. Therefore, this section argues that considerations of theoretical economy need not be a barrier to explore whether alethic minimalism is epistemologically fundamental.

To understand truth, according to Horwich’s MT, is to comprehend how ES works. Truth’s functional role is captured by three core theses:

  1. (i)

    truth includes non-paradoxical instances of ES;

  2. (ii)

    truth is a property; and

  3. (iii)

    we accept a generalised form of ES.

First, the propositions expressed by non-paradoxical instances of ES, i.e., 〈〈p〉 is true if and only if p〉, make up the axioms of Horwich’s MT. Second, the inferential role of ‘true’ as a logical predicate is not purely redundant. Finally, since there is not an uncontroversial way of generalising from instances of ES, Horwich proposes that we have a disposition to accept the conjunction of all uncontroversial and non-paradoxical sentences instantiating ES as capturing the entirety of MT.Footnote 6

If we accept the three guiding principles of MT, then minimalism is explanatorily, conceptually, logically, and epistemologically fundamental. It is explanatorily fundamental because everything we are able to do with the truth predicate can be explained in terms of instances of ES. Second, MT is conceptually fundamental since no substantive concept underwrites instances of ES; such instances are brute.Footnote 7 Third, it is natural for the truth predicate to enable the explicit formulation of schematic generalisations, and, therefore, allow MT to be logically fundamental. Finally, instances of ES need not be justified by anything more immediately known or anything more obvious to us than non-paradoxical cases of ES; thus, MT is epistemologically fundamental.

The focus of this paper is the last of these theoretical pillars of alethic minimalism. Either non-philosophers have a disposition to accept practical variants of ES or they do not. If non-philosophers accept instances of ES without any evidential support, then it seems clear that MT is epistemologically fundamental. Likewise, if non-philosophers accept the equivalence of p and ‘p’ is true without any intervening discussion, say, of what makes ‘p’ true or of there being a correspondence between p and how entities stand to one another in the world, then the correspondence relation between propositions and the world do not mediate one’s acceptance of the concept of truth. If non-philosophers do not accept instances of ES, then MT’s epistemological fundamentality applies to philosophers but not necessarily to those outside the discipline. Epistemological fundamentality appears to be accessible through systematic study of a group of respondents by asking them questions concerning what their intuitions are about practical instances of ES.

First, let me be clear that I am not suggesting Horwich should offer a consensus view. Horwich never claimed that MT would be the consensus view upheld by all philosophers and non-philosophers. Moreover, consensus is troubling because sometimes it means that there is unanimous consent while at other times it means that a significant majority agree with a prevailing view. First, Horwich is clear that we should not expect MT or practical variants of ES to be 100% accepted by language users.Footnote 8 Whilst a consensus is reached by unanimous vote, reaching a consensus need not be equivalent to a unanimous vote.

On an alternative view of consensus, there might be dissenters to the majority’s consensus. Horwich has explicitly warned us against the idea that the debate be settled by consensus.

[T]here’s the idea that truth be identified with what most people think—or, better perhaps, with what most people would come to think if they investigated thoroughly enough to reach a consensus. But, intuitively, we could all agree with each other yet all be mistaken. And, intuitively, there are facts on which we would not converge no matter how thorough the investigation. (Horwich 2010, p. 3)

Horwich’s rejection of a consensus view consists of two complaints: (a) it could be mistaken and (b) it might never be reached, regardless of how well informed the people that form the consensus are. If consensus leads to a mistaken view of truth, then there is no reason for truth to be identified with ‘what most people think’. We should abandon whatever the consensus says the truth is and adopt whatever the appropriate view is. If consensus is never reached, then it seems natural to say that we ought to drop our attempt to reach consensus.

Horwich’s analysis leaves out an alternative view of consensus. Consensus is somewhere in between outright acceptance and straightforward rejection of a select view by a significant majority of people. Should a sizable number of language users who are not philosophers disagree with instances of ES, that fact should not be ignored or taken lightly by Horwich. Horwich may have to reconsider whether practical instances of ES are axiomatic of MT. He would even have to consider whether the context or some other feature of the proposition under consideration may be adversely affecting people’s view of ES. One could imagine participants in an empirical study who affirm that ‘abortion is always wrong’ but deny the claim that ‘it is true that abortion is always wrong’. To modify a sentence with ‘It is true that…’ may be interpreted by a non-philosopher as suggesting that the sentence should be universally accepted. Respondents may be reluctant to go so far as to suggest that the two are equivalent. Moral discourse, especially about matters at the borders of life, may be ripe with counterexamples for the axioms of Horwich’s minimalism. These outliers should not necessarily dictate whether practical variants of ES are accepted by the consensus.

If a statistically significant majority of people favour a variety of instances of ES, then there is reason to believe that ES is axiomatic and has the support of non-philosophers. Under such an interpretation, we may take practical instances of ES to be axiomatic. This opens the possibility that among non-philosophers the three core theses of Horwich’s minimalism about truth may be supported empirically, and at least two of the four fundamentality theses of minimalism are founded upon natural language variants of truth-talk. More empirical research will need to be done to serve as evidence for the core theses and fundamentality theses of minimalism, but what this article intends to show is the starting point from which we should consider all future research into this phenomenon. Finding evidence of non-philosophers intuitively agreeing with practical instances of ES suggests that ES is axiomatic at a practical level, as Horwich has hypothesised.

Yet, one may further object to the salience of the study by pointing out that Horwich did not have non-philosophers in mind when he talks about the intuitiveness of practical instances of ES. Perhaps Horwich has a special form of ‘intuition’ in mind when he says that practical variants of ES are epistemologically fundamental. In a paper, not on truth but on the basic methods of belief formation, Horwich discusses the notion of ‘rational intuition’, which mirrors what he has said about epistemological fundamentality:

[W]e have certain fundamental convictions—beliefs and inferences that come from neither experience nor reasoning. We just feel that they are obviously right. And we might then express ourselves by saying, “That’s just the way things intuitively seem.” But there is no characteristic experience, no phenomenological “glow of plausibility,” that lies behind those certainties. Introspection reveals no such thing—but would presumably reveal it if it were there. (Horwich 2010, p. 215)

Accordingly, rational intuitions do not comprise a sound evidential basis of beliefs or convictions because whenever a philosopher calls upon these intuitions she does so not dispassionately but with a sense of urgency to unify the theory the philosopher has proposed. That intuitions for Horwich are just how things seem to be aligned with what he has said of MT’s epistemological fundamentality, and with what experimental philosophers seek to explore through systematic empirical studies.

It seems, then, that Horwich should abdicate some authority over the question of whether practical variants of ES capture the non-philosopher’s notion of truth to a philosophically-informed empirical investigation. The issue becomes an empirical one: are folk committed to practical variants of ES?

3 On Dispositions to Accept Practical Variants of ES

Horwich’s notion of intuition amounts to non-technical intellectual seemings almost anyone would have a disposition to accept, and, at least among theories of truth, he suspects that the correspondence theory would be most appealing. Non-paradoxical propositional instances of ES are conceptually, explanatorily, logically, and epistemologically fundamental. For truth to be epistemologically fundamental, ‘we accept … instances [of ES] in the absence of supporting argument’ (Horwich 1998c, p. 103) and consequently ‘[t]he idea is that just as [ES] tells us everything there is to tell about truth, so the schemas for ‘refers’ and ‘true of’ tell us everything there is to tell about reference and the ‘true-of’ relation’ (Simmons 2018, p. 207): ‘there is nothing more to our concepts of being true of and reference than is conveyed by our acceptance of [ES]’ (Horwich 1995, p. 360). Acceptance of practical instances of ES, then, is constitutive of a theory that is empirically adequate in terms of capturing how non-philosophers employ truth-talk. Horwich may have explored the truth-talk of non-philosophers unsystematically, for example by eavesdropping on conversations at a café. If systematic studies have been conducted on non-philosopher’s employment of truth-talk, then that data may be used to support or to deny what Horwich has claimed regarding non-paradoxical practical instances of ES. In this section, I report some fascinating data that have some bearing upon Horwich’s MT because, at least for matters of fact, people tend to agree with practical instances of ES, which may further warrant Horwich’s claim that ES is axiomatic.

When we look at Næss’ early studies, we seem to find a thin or deflated account present among the participants’ responses. Næss reports that respondents are far more likely to agree with a variety of ES than with a non-philosopher’s conception of truth (cf. Næss 1938). When subjects are presented with a non-philosopher’s account of truth and an informal variety of ES, Næss says that study participants who received “PAf 148”: ““p” ist wahr, wenn p” were divided; it “received much criticism as well as appraisal” (Næss 1938, p. 148). Næss does not include the responses of all participants, but here are a few examples:

P 1: “It says truth is truth! A definition must be otherwise: one cannot use the same expression. It explains nothing. The same is said twice.”

P 55: “This seems to turn round in circles.”

Unidentified P: “But “viewed from outside”, judged automatically by the thing itself, independent of any person, the formula is all right. A thing, an event, “something”, is true if it exists in reality.” (Næss 1938, p. 145)

PAf 148 was identified as ‘Tarski,’ and Næss treated it as an informal interpretation of Tarski’s Convention-T, which is equivalent with what I have been calling ES. He writes in a footnote, ‘The “PAf 148” (Tarski) is not to be identified with the so-called “semantic notion of truth.” To construct this notion the method of formalisation is essential. There is, however, a tendency to look at PAf 148 as a definition of non-formal truth’ (Næss 1938, p. 148). We should think about PAf 148 as a non-formal variant on instances of ES or what Horwich uses as axioms for MT.Footnote 9

Several participants in Næss’ (1938) study, in fact, offered formulations compatible with instances of ES. Næss’ findings do not just show that people accept a multichoice answer but respond in ways compatible with practical instances of ES since the interviewers asked study participants for their views and they volunteered their views. In this sense, the views that Næss’ respondents give are fundamental in the way that Horwich must have meant. For example, formulation-root Nos. 40,3 and 37,3 in Gr1-8 included the following:

8 | 40,3: “it rains now” is true if it actually rains now etc. (“p” is true, if p).

8 | 37,3: “outside is warm” is a true statement, if it is warm outside etc. (“p” is true if p). (Næss 1938, p. 42)

Of the first 150 study participants Næss and his examiners queried, Test-persons #40 and #37 offered on at least three different occasions formulations of the common characteristic of truth to be something like instances of ES. One might not believe that this result is remarkable because only two test-persons of 150 converged upon instances of ES. That would be true if it were not for the fact that participants rarely offered formulations on multiple occasions. Only 22 participants offered the same formulation three or more times in response to the leader’s questions, so approximately 10% of them were wedded to a variation of an instance of ES. Thus, Næss’ (1938) studies provide evidence that practical variants of ES are prevalent in the responses offered by study participants.

One may point out that Næss’ participants who offered responses to the qualitative questionnaire that amount to practical instances of ES need not uphold the stronger view that these instances exhaust a theory of truth. It is one thing to offer an instance of ES in response to a carefully crafted questionnaire but another thing to say that it constitutes a definition of truth. Minimalism about truth is an interesting test case because Horwich has said that ‘we accept instances [of ES] in the absence of supporting argument’ (1997, p. 96). That means non-philosophers need not be burdened with having to say that instances of ES are constitutive of a definition of truth. Moreover, they do not need to provide reasons for offering instances of ES. Our acceptance of instances of ES, on Horwich’s view, is constitutive of accepting minimalism.

4 Further Evidence for a Practical Variant of ES in Non-philosophers from Næss’ 1953 Studies

In ES, when we replace a variable, such as p, with specific content about empirical matters of fact, predictions, or a broad description of a popular scientific theory, the data tell us a majority of non-philosophers judge that ‘p’ and ‘It is true that p’ are synonymous.Footnote 10 This implies that non-philosophers accept instances of ES and the axioms of Horwich’s MT, thus suggesting that the view is epistemologically fundamental.

Much of what Næss found in his earlier PAf148 studies carries over to later studies focusing upon the semantic or—broadly speaking—the practical rendering of Tarski’s conception of truth. Using these earlier studies as a backdrop for later ones, Næss hypothesised that:

There is a more pronounced tendency to affirm the synonymity of “p” and “It is true that p” in case “p” is a sentence expressing a (supposed) matter of fact than in case “p” expresses, (1) a prediction, or (2), a theory. (Næss 1953, p. 16)

In the 1953 studies, Næss’ participants received the following two criteria to assist them in deciding whether two statements are synonymous or express two different ideas:

A sentence A is for you expressive of a different assertion from that of another sentence B, if and only if you can imagine possible (but perhaps not actual) circumstances (conditions, existing state of affairs) of such a kind that if they were present you would accept A as warranted but reject B, or vice versa.

A sentence A expresses for you one and the same assertion as another sentence B, if and only if you can not imagine such circumstances, that is, if you under all conditions whatsoever either would accept both A and B as warranted, or reject both A and B. (Næss 1953, p. 38)

Then, Næss proceeded to ask 129 participants whether two (or more) statements expressed the same thing. For purposes of this paper, I will restrict myself to Næss’ questions #1, #3, and #7 (See Table 1.). For each question, Næss asked participants:

Are some of the following sentences for you expressive of the same assertion, and if that is the case, which ones: (Næss 1953, p. 39)

Table 1 Synonymity affirmations

Here are the statements:

1A: There is at least one copy of the Bible in the University Library.

1B: It is true that there is at least one copy of the Bible in the University Library.

3A: It will be raining on May 17.

3B: It is true that it will be raining on May 17.

7A: (Here you may imagine that the selection theory of Darwin be formulated.)

7B: Darwin’s selection theory is true. (Næss 1953, p. 39)

For #1 and #3, a majority of study participants responded that A and B express the same thing, 87% of 126 respondents for #1 and 81% of 114 respondents for #3. Since #1 concerned a straightforward present tense empirical claim about a book in the University Library, people seem to find no difficulty affirming ‘It is true that there is at least one copy of the Bible in the University Library’ is synonymous with there is at least one copy of the Bible in the University Library. Likewise, people tend to affirm that ‘It is true that it will be raining on May 17’ is synonymous with the fact that it will be raining on May 17. Call #1 a ‘matter of fact’ and #3 a ‘prediction’.

Næss found that people are more likely to agree that prompts involving matters of empirical fact are synonymous with their alethically modified counterparts than prompts involving statements about scientific theory and their alethically modified counterparts. As Næss correctly reported, his hypothesis did not pan out when it compared the ‘matter of fact’ with both (1) a prediction and (2) a theory [χ2 = 1.336, ns], largely because the prediction, Question 3, had as many people agree that 3A and 3B expressed the same assertion as did the participants who responded to Question 1, A and B. But, if we remove the prediction from Næss’ hypothesis, the data confirm a statistically significant difference among study participants [χ2 = 27.9, p < 0.01].

Let’s turn our attention to Questions 1 and Question 7. The sentence of question 7 involves a scientific theory—specifically Darwin’s evolutionary theory. A majority of people affirmed the synonymy of statements in Question 1A and B, while a majority of respondents denied that synonymy was present for Question 7A and B (see Table 1.). The results are statistically significant (x2 = 40.9116, df = 1, p < 0.01). The results seem to suggest that non-philosophers accept practical instances of ES when its content concerns matters of fact, but reject practical instances of ES when its content concerns scientific theories.Footnote 11

The explanation leaves us with a puzzle. When non-philosophers are asked to judge whether a scientific theory is true, they tend to deny that a formulation of Darwin’s selection theory and its alethically modified counterpart are synonymous. Why does this occur for scientific theories but not for matters of fact or predictions? This might be the result of study participants entertaining an alternative conception of evolution, e.g., a Lamarckian view, or a belief that a theory amounts to nothing more than an unconfirmed hypothesis. To some extent, it may be that when we ask non-philosophers about synonymy they conflate the task of evaluating synonymy between two propositions with the task of expressing agreement with them. Perhaps, too, the results concerning Question 7 should be set aside because Næss never explicitly tells us the content that should be used in place of ‘[A formulation of Darwin’s selection theory]’. Of course, we need not let the puzzle go unsolved, as empirical studies could help us resolve them.

5 A Replication and Extension of Næss’ Studies

Barnard and Ulatowski (2019) have reported data that replicate and extend Næss’(1938) and (1953) studies. For their project, participants were a geographically and demographically diverse group of 300 total participants aged 18–65 years, 183 (60.8%) self-identified as male and 118 (39.2%) self-identified as female. Barnard and Ulatowski employed a between-subjects study design where study participants were randomly assigned to one of four conditions modeled on those from Næss’ 1953 study. Participants were asked whether they believed the two statements were equivalent or different. The responses for each group were then compared. Just as Næss’ data had shown, they hypothesized that non-philosophers’ notions on a practical version of ES varies according to the content of p. As with Næss’ study, Barnard and Ulatowski provided respondents with criteria for how they should judge whether two statements are equivalent or different.Footnote 12

Study participants were given one of the following statements:

Q2. There is at least one copy of the Bible in the University Library.

Q3. It will be raining in Paris on 17 September 2018.

Q4. There is an ether that oscillates in accordance with the wave theory of light or closely related theories.

Q5. Darwin’s selection theory explains the evolution of populations through change in heritable traits over time.

Q2–Q5 represented statement A for each study participant. B is one of the statements prefixed with ‘It is true that…’. For Q2, A is, e.g., ‘There is at least one copy of the Bible in the University Library’ and B is: ‘It is true that there is at least one copy of the Bible in the University Library.’ Respondents were asked: Do sentences A and B express the same thing, or not?

Barnard and Ulatowski replicated Næss’ results after more than sixty years (See Fig. 1.). While the frequency distribution of Barnard and Ulatowski’s study do not (visually) look like what Næss had found in his own data set, statistical analysis shows that there is a statistically significant difference between Q2 and Q5. The data show a significant difference between people’s responses to the Bible case, an empirical matter of fact, and an expression of Darwin’s evolutionary theory (χ2 (1) = 2.65112, p < 0.01, N = 154).

Fig. 1
figure 1

Practical disposition to accept ES, Næss replication study

Fig. 2
figure 2

Judgments of equivalence and difference for Q6–Q13

Barnard and Ulatowski completed a second experiment designed to extend the reliability of Næss’ original empirical results. It was a within-subjects design where respondents were asked whether a statement and its truth-prefixed corollary were equivalent or different, i.e., a practical variant of ES. For example, Q8A is ‘Unicorns tend to sleep underneath oak trees’ and Q8B is: ‘It is true that unicorns tend to sleep underneath oak trees.’ One of the statements included was ES itself, e.g., p and p is true (Fig. 2).Footnote 13 Study participants received eight statements:

Q6. Even numbers are divisible by two without remainder.

Q7. Every even number is the sum of two prime numbers.

Q8. Unicorns tend to sleep underneath oak trees.

Q9. Emma is honest.

Q10. William Shakespeare’s Romeo and Juliet have matching tattoos.

Q11. Dr. Watson and Sherlock Holmes often share a pipe at afternoon tea.

Q12. The old man’s liver deftly legislated.

Q13. p

The results of Q6–Q12 conform with what Barnard and Ulatowski found in the replication study. The exception was the Q13 result. In this case, a majority of participants judged that ‘p’ and ‘It is true that p’ were different. The result for Q13 was compared with the responses for the seven other examples and a significant difference was found in every case. The results indicate that while people readily agree that instances of ‘It is true that p’ and p are equivalent when content stands in for p, they say that the two are different when no content stands in for p.

Someone may point out that if the disposition to accept all instances of ES is a feature of Horwich’s MT, then people ought to affirm the synonymy of Q13A and Q13B. The respondents of Barnard and Ulatowski’s (2019) study did not register that ‘p’ and ‘It is true that p’ are equivalent and this suggests against Horwich’s view.

There is one critical failure of this potential objection: Horwich anticipated that the non-philosopher would see a distinction between ‘p’ and ‘It is true that p’. Reminder: the truth-predicate is logically, explanatorily, epistemologically, and conceptually fundamental, but its fundamentality does not rule out a distinction between two statements having different functional roles. In the Postscript of Truth (1998) and in a contribution to Blackburn and Simmons’ anthology Truth (1999), Horwich writes:

A … respect in which minimalism improves on the original redundancy theory is in not claiming that ‘p’ and ‘The statement (belief, …) that p is true’ have exactly the same meaning.

When participants were given each of Q6-Q12, they could have just glanced at the alethic modifier without giving it much thought. Noting that the content of whatever sentence replaced p was the same between the proposition itself and its alethically modified counterpart, they may have merely accepted that the two were equivalent. For Q13, however, the situation for the respondent may have been slightly different.Footnote 14 Respondents may have conjectured that p and the alethically modified statement, ‘It is true that p’ (or ‘p is true’), could function differently. I have already mentioned such a case like this at the beginning of the paper. Respondents may be perfectly content to affirm ‘abortion is always wrong’ because it is a statement with which they agree, but they may be reluctant to affirm ‘It is true that abortion is always wrong’ because they believe that by adding ‘It is true that’ changes the content in some way. This would entail that ‘p’ and ‘It is true that p’ have two different meanings, which is exactly what Horwich anticipated.

6 Challenges

Let me address a few criticisms that may be raised against my analysis. First, a critic might challenge the notion of calling upon empirical data to support a particular philosophical theory of truth by arguing that it need not be intuitive like a theory of ethics. While there is plenty of room for disagreement when it comes to concepts of good, right, or value, there is very little room for disagreement when it comes to the concept of truth. Truth is truth!

It seems important for me to state here that I do not disagree with the critic. Theories of truth and theories of ethics are different. While there may be much disagreement between different theories of ethics, there seems very little room for disagreement when it comes to truth. One may think that disagreements in ethics do not constitute a worry of talking past one another, whereas disagreements over truth would. Nevertheless, if we seek a truth theory that is compatible with the folk view, then we cannot ignore the empirical data without at the same time giving up on our view that the two are compatible. The collected data analysed here point out that non-philosophers’ truth-talk seems to align quite well with a prevailing assumption of MT, i.e., that practical instances of ES are axiomatic.

The empirical data collected together in this article provide support for the epistemological fundamentality of ES as the basis or foundation for MT. Just as Horwich has argued in a variety of papers (1995, 1997, 2001), chapters (2010), and books (1990/1998, 1999), ES must be axiomatic, such that all of the other conditions of MT may be derived from them. Horwich writes:

It can be argued that [practical instances of ES] are epistemologically fundamental–we do not arrive at them, or seek to justify our acceptance of them, on the basis of anything more obvious or more immediately known. It can be argued, in addition, that our underived inclination to accept these biconditionals is the source of everything else we do with the truth predicate. (2010, p. 37)

The data show that people have an unmediated commitment to practical instances of ES. People willingly respond that “Snow is white” is true and snow is white are equivalent, without knowing snow’s being white corresponds to the facts. This view is bolstered by some of the empirical data. When, in Barnard and Ulatowski’s experiment, respondents agree that “The old man’s liver deftly legislated” is equivalent to its alethically modified counterpart, it seems odd to infer that respondents believe that the nonsense statement corresponds to the facts. One would suspect that if respondents were endorsing a correspondence account of truth, they would say that the two were not equivalent because there are no facts to which the nonsense proposition corresponds. Horwich writes, “we are required to accept that the statement made by “Dogs bark” is true just in case dogs bark, and so on. It is some such basic principle of use that provides the truth predicate with its distinctive meaning and which equips it to perform its distinctive function” (Horwich 2005, 2010, p. 15). There is no rule or guideline for people to call upon that substantively reinforces the nature of truth in order to accept practical instances of ES.

None of the data reported in this paper approach a theory of truth, and one should not misconstrue the data as being indicative of a full-fledged theory. The finding does not suggest that non-philosophers are minimalists about truth. That would be too presumptuous. Yet, if people were made aware of the conceptual apparatus that supports MT and if we could convince them that their intuitive views should lead them to MT, then they, too, may adopt a deflationary account of truth on the order of Horwich’s minimalism since they assume practical variants of ES.

Second, the inference from empirical results to conclusions about the acceptability of instances of ES deserves to be defended. Some will note that none of the empirical studies asked the question: Do you accept that ‘‘p’ is true if and only if p’? Appropriate experimental design prevented investigators from asking that complex question because such an approach would have presumed that non-philosophers understood quotation devices and the truth functional assessment of the biconditional. We may imagine that the decision by Barnard and Ulatowski to follow Næss and approach the issue through an assessment of synonymy was rooted in a particular understanding of the biconditional. Suppose that we begin with a biconditional such as ‘Humpbacks are mammals if and only if they give birth to live young’. The truth of the biconditional depends upon ‘Humpbacks are mammals’ and ‘Humpbacks give birth to live young’ meaning the same thing (and hence having the same truth conditions). So, for example, Næss asked if two expressions were the same assertion. If participants judge that the expressions are equivalent, then this supports the affirmation of the corresponding biconditional. We need not assume that people actually draw that conclusion; rather, if they judge that the expressions are different, we expect that they reject the biconditional. In other words, affirming the equivalence of A and B is consistent with ‘A iff B’, but affirming the difference between A and B entails that one does not accept ‘A iff B’. This is an indirect way of approaching the matter, but so long as one maintains a semantically driven reading of instances of ES, it seems reasonable.

Horwich’s MT has it that there is nothing more about truth we need to assume than that “p is true” is just an indirect way of saying that p. For Horwich, ES is something approaching a universal generalisation which is inferred indirectly from a potentially infinite number of non-paradoxical instances of ES, e.g.:

  • 〈Spot is a dog〉 is true if and only if Spot is a dog.

  • 〈South Africa won the 2019 World Cup〉 if and only if South Africa won the 2019 World Cup.

  • 〈Bauxite is made of aluminium ore〉 is true if and only if bauxite is made of aluminium ore.

Horwich believed that all non-paradoxical propositional instances of ES are fundamental because our grasp of truth is composed of accepting these biconditionals as the source for everything else we do with the truth predicate, and they explain all the facts we need to know about truth (cf. Horwich 2001).

7 Conclusion

If “our grasp” of truth is somehow settled by instances of ES, to put matters in Horwich’s own terms, then we should be able to empirically test whether people, when asked to consider their views on truth, employ truth-talk that resembles ES. This is not to permit decisions about the theory of truth to be decided by consensus of the majority; rather, it is a matter of discovering what intuitive commitments people have about truth.

If the analysis is correct, and it seems to have been shown to be true in the empirical studies called upon here, then Horwich’s minimalism about truth seems to be reflected in the accumulated data. If there is anything uncontroversial we might conclude from the empirical studies, it is that practical intuitive judgments of non-philosophers affirm practical instances of ES that serve as axioms for Horwich’s minimalism.