Resolving the puzzle of the changing past

ABSTRACT

Barlassina and Del Prete argue that the past can change, on the basis that there is no other explanation for the truth values of certain claims involving the past-tense predicate ‘won the Tour de France in 2000’. To establish this, they argue that no contextualist account of this predicate will be able to explain these truth values. I show that their argument straightforwardly fails. Not only does a tweak to the contextualist account they consider suffice to explain these truth values, there is in fact an even simpler and more plausible non-contextualist account that can do the same work. Put simply: there is no puzzle of the changing past.

KEYWORDS:

• The past
• change
• past tense
• contextualism

In ‘The Puzzle of the Changing Past’, Barlassina and Del Prete (2015) argue that the past can change, on the basis that there is no other explanation for (1)–(3) (pp. 61, 65):

1. ‘Lance Armstrong won the Tour de France in 2000’ was true in 2002.

2. ‘Lance Armstrong won the Tour de France in 2000’ is false in 2022.

3. ‘It is no longer the case that Lance Armstrong won the Tour de France in 2000’ is true in 2022.

(Armstrong was stripped of his Tour de France titles by the governing body of the Tour de France, the Union Cycliste Internationale (UCI), in October 2012.) They conclude that we ought to “stop asking whether the past can change and start to inquire on how to make sense of this.” (p. 66)

Some appear to have accepted this surprising conclusion.¹ Others have sought to deny the truth of (3) (and either (1) or (2)), and/or have felt the need to engage in substantive metaphysical and semantical reflection in their attempts to finesse the issue.²

However, to establish their conclusion, Barlassina and Del Prete argue that no contextualist account of the predicate ‘won the Tour de France in 2000’ can explain (1)–(3) (pp. 64–66). And this is straightforwardly mistaken. (§I) Moreover, with that error corrected, an even simpler and more plausible account of the predicate can be brought into view. (§II) The upshot is that there simply is no puzzle of the changing past.

I

Consider the sentence ‘According to UCI, Lance Armstrong won the Tour de France in 2000’. Clearly, claims (1′)–(3′) are all correct:

1. ‘According to UCI, Lance Armstrong won the Tour de France in 2000’ was true in 2002.

2. ‘According to UCI, Lance Armstrong won the Tour de France in 2000’ is false in 2022.

3. ‘It is no longer the case that, according to UCI, Lance Armstrong won the Tour de France in 2000’ is true in 2022.

Why are (1′)–(3′) correct? Well, back in 2000, UCI took an official view on the matter when it declared Armstrong the winner of that year’s Tour de France. And in 2002, it still held him to have won—hence the correctness of (1′). But in 2012, as already mentioned, UCI changed its official view by issuing the declaration stripping Armstrong of his titles, from which point on it no longer held him to have won. And that is how things still stand today, in 2022—hence the correctness of (2′) and (3′).

There is nothing semantically unusual or mysterious going on here. Compare a sentence such as ‘According to Biden, Aristotle taught Plato’. Suppose this is something Biden used to claim in 2021, whereas now, having learnt of his error, he claims that Plato taught Aristotle instead. Then ‘According to Biden, Aristotle taught Plato’ was true in 2021; but it is now false; and ‘It is no longer the case that, according to Biden, Aristotle taught Plato’ is now true.

But if (1′)–(3′) are all correct, then an account of ‘won the Tour de France in 2000’ according to which it is more or less semantically equivalent (in the relevant contexts) to ‘according to UCI, … won the Tour de France in 2000’ will be able to explain (1)–(3). And such an account is in fact exactly what we get with only the smallest of tweaks to the very contextualist account that Barlassina and Del Prete argue against.

How so? Well, according to the account they consider, ‘won the Tour de France in 2000’ is more or less semantically equivalent to ‘according to δ, … won the Tour de France in 2000’, where δ is a contextually relevant declaration.³ And what they point out – quite rightly – is that no single declaration would then render (3) correct. (pp. 65–66) For it was never the case that, according to UCI’s 2012 declaration, Armstrong won the Tour de France in 2000. So it cannot be true that this is no longer the case. Similarly, it is still the case that, according to UCI’s initial 2000 declaration, Armstrong won the Tour de France in 2000. And so it cannot be true that this is no longer the case, either.

However, if we take δ to be a contextually relevant institution, rather than a contextually relevant declaration, this problem disappears. For unlike declarations, institutions change. As we saw above, UCI adopted one view in 2000, holding Armstrong to be the winner of that year’s Tour de France, and then changed its view in 2012, after which it no longer held him to have won. While no declaration went from holding that Armstrong won the 2000 Tour de France, to not holding that Armstrong won the 2000 Tour de France, UCI did. And so, again, ‘According to UCI, Lance Armstrong won the Tour de France in 2000’ was true in 2002; but it is now false; and ‘It is no longer the case that, according to UCI, Lance Armstrong won the Tour de France in 2000’ is now true.⁴

II

We’ve seen that a contextualist account of ‘won the Tour de France in 2000’, of the sort suggested by (1′)–(3′), can resolve the supposed puzzle concerning (1)–(3). But is it correct? Not quite.

Suppose some other institution, call it ‘ABC’, replaced UCI as the governing body of the Tour de France in 2012. Further, suppose that ABC had never held Armstrong to be the winner of the Tour de France in 2000, and, upon becoming the relevant authority, issued a declaration officially stripping him of his titles. Finally, suppose UCI not only continued to exist after 2012, but continued to this day to regard Armstrong as the winner of the 2000 Tour de France, despite no longer being the official authority.

Under these suppositions, (1)–(3) seem to be just as correct as they actually are. However, the problem facing the declaration-based contextualist account now also faces the institution-based contextualist account. For neither ABC nor UCI is such that it no longer holds that Armstrong won. (ABC never held this to be true; UCI still holds it to be true.) And so ‘It is no longer the case that, according to δ, Lance Armstrong won the Tour de France in 2000’ comes out as false in 2022, whichever institution is taken to be the contextually relevant value for δ.

Again, however, a small tweak—and, this time, a simplifying one—resolves the issue. For consider the sentence, ‘Officially, Lance Armstrong won the Tour de France in 2000’. ‘Officially’, in this perfectly standard and familiar use, means something like ‘according to the official authority’. And clearly, (1*)–(3*) are all correct, both in actuality and in the alternative scenario outline above:

1. ‘Officially, Lance Armstrong won the Tour de France in 2000’ was true in 2002.

2. ‘Officially, Lance Armstrong won the Tour de France in 2000’ is false in 2022.

3. ‘It is no longer the case that, officially, Lance Armstrong won the Tour de France in 2000’ is true in 2022.

Why are (1*)–(3*) actually correct? Well, given that the official authority in every actual context at issue is UCI, it follows that, both in 2002 and in 2022, for something to be officially the case (/to be the case according to the official authority) is for it to be the case according to UCI. And so the explanation given above for why (1′)–(3′) are actually correct serves equally as an explanation for why (1*)–(3*) are actually correct.

Why are (1*)–(3*) correct in the alternative scenario outlined above, in which the governing body changes? Well, suppose we are in that scenario and it is 2022. Back in 2002, UCI was the official authority and still held Armstrong to be the winner of the 2000 Tour de France—hence the correctness of (1*). Now, in 2022, ABC is the official authority, and does not hold Armstrong to be the winner of the 2000 Tour de France—hence the correctness of (2*). But we can also say now, in 2022, that it is no longer the case that the official authority holds Armstrong to be the winner of that race—hence the correctness of (3*).⁵ The difference is simply that, in this circumstance, the change in the official verdict between 2002 and 2022 – the difference in what holds according to the official authority – occurred due to a change in which institution had official authority, rather than due to a change in what any single institution’s verdict was.

There is still nothing semantically unusual or mysterious going on here. Compare the sentence, ‘According to the President, Aristotle taught Plato’. This time, suppose that, whereas Biden has long claimed that Plato taught Aristotle, Trump has long claimed that Aristotle taught Plato. Then, ‘According to the President, Aristotle taught Plato’ was true in 2019; but it is false in 2022; and so ‘It is no longer the case that, according to the President, Aristotle taught Plato’ is true in 2022.

But if (1*)–(3*) are straightforwardly correct, both in actuality and under the suppositions above, then an account of ‘won the Tour de France in 2000’ on which it is more or less semantically equivalent to ‘officially, … won the Tour de France in 2000’ will get all the right results for (1)–(3). And such an account is surely plausible. After all, it is precisely the change in the official view that prompted people to make claims such as those in (2) and (3). No one who accepts (1)–(3) on this sort of basis would reject (1*)–(3*). And the account is nice and simple. For the content of ‘won the Tour de France in 2000’ (and, more generally, appropriate uses of ‘won’) need no longer be thought of as varying from context to context in the way supposed above.

Given the availability and plausibility of this account, we should accept it. (1)–(3) do not actually reflect the past somehow changing as a result of UCI’s 2012 declaration. All they reflect is a contemporaneous change in the official verdict about what happened in the past.

In sum: there is no puzzle of the changing past.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This paper was completed whilst on a project that has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme [Grant Agreement No. 679586].

Notes

1 See Effingham (2020, p. 1) and Deasy (forthcoming, p. 4, n. 7). (And Del Prete (2020, pp. 62–65) still seems to hold the puzzle to be genuine.)

2 See Iacona (2016), whose response is endorsed by Langton (2018, p. 157, n. 65) and perhaps Meyer (forthcoming, p. 2, n. 1); also Büttner & Dolby (2017), Jaszczcolt (2018, pp. 12–14), Spolaore & Del Prete (2019, pp. 138–139) and Torrengo (2018).

3 They also consider a very similar relationist account of the predicate, which assigns the same propositional contents to the sentences at issue (in the relevant contexts) as this contextualist account. The points made here could have been made about either.

4 It would be a mistake to object on the basis that, in order to evaluate the sentence ‘According to UCI, Lance Armstrong won the Tour de France in 2000’ at a time t, ‘UCI’ must be understood as referring to something time-indexed or otherwise t-specific, rather than to the persisting institution. One form of this objection would be that, because (let us suppose) objects persist and change by having temporal parts, ‘UCI’ must be understood as referring to UCI’s 2002 temporal part in 2002, but to UCI’s 2022 temporal part in 2022. Another form of the objection would be that, because the truth of the sentence at a time turns on what UCI’s position is at that time, ‘UCI’ must be understood as elliptical for something like ‘UCI’s 2002 position’ in 2002, and for something like ‘UCI’s 2022 position’ in 2022. (Either way, this would undermine our explanation of (3′)’s (and so (3)’s) correctness. For what would ‘UCI’ then pick out in 2022, such that the sentence mentioned in (3′) is true?) But there is no more reason to think either of these things than there is to think that, in order to evaluate ‘Ada holds that p’ or ‘Ada is pale’ at t, we must understand ‘Ada’ as referring to a temporal part of Ada, or as elliptical for something like ‘Ada’s t position’ or ‘Ada’s t appearance’—and that we therefore face the problem of explaining how sentences such as ‘Ada no longer holds that p’ or ‘Ada is no longer pale’ can be true. We face no such problem. All of these sentences refer to, and ascribe properties to, persisting objects, not to temporal parts or time-indexed positions/appearances. Even if it is true that (say) persisting objects have properties at times in virtue of having temporal parts that have those properties simpliciter, this is a metaphysical thesis, and it is entirely compatible with a standard semantic picture on which we refer to, and ascribe properties (at times) to, the persisting wholes. So there is simply no theoretical problem with taking δ to receive the very same contextually-determined persisting entity as its value both in 2002 and in 2022, with both the change in truth value reflected in (1′)–(2′), and the specific truth evaluation recorded in (3′), being explained by that one entity having undergone an appropriate change. (Thanks to two anonymous referees for raising the objections addressed in this footnote.)

5 This is the case when ‘the official authority’ takes narrow scope with respect to ‘it is no longer the case that’—which is the reading that corresponds to the most (only?) natural reading of ‘It is no longer the case that, officially, Lance Armstrong won the Tour de France in 2000’, in which ‘officially’ takes narrow scope with respect to ‘it is no longer the case that’. (If there are natural scope ambiguities in the sentences mentioned in (1*)–(3*), then it may be better to speak of them as having suitable true readings, rather than as being true, in the relevant contexts. But any such ambiguities would not cast doubt on the present suggestion. They would simply provide us with extra resources to account for any differences of opinion that competent speakers may have concerning (1)–(3).)