Philosophical Agreement and Philosophical Progress

Abstract

In the literature on philosophical progress it is often assumed that agreement is a necessary condition for progress. This assumption is sensible only if agreement is a reliable sign of the truth, since agreement on false answers to philosophical questions would not constitute progress. This paper asks whether agreement among philosophers is (or would be) likely to be a reliable sign of truth. Insights from social choice theory are used to identify the conditions under which agreement among philosophers would be a reliable indicator of the truth, and it is argued that we lack good reason to think that philosophical inquiry meets these conditions. The upshot is that philosophical agreement is epistemically uninformative: agreement on the answer to a philosophical question does not supply even a prima facie reason to think that the agreed-upon view is true. However, the epistemic uninformativeness of philosophical agreement is not an indictment of philosophy's progress, because philosophy is valuable independent of its ability to generate agreement on the correct answers to philosophical questions.

1. Introduction

Lack of agreement among philosophers is often taken to be a threat to philosophical progress. Some authors have argued that the widespread disagreement among experts about the answers to major philosophical questions shows that philosophy as a discipline has failed to progress, or has not progressed as much as other disciplines which exhibit greater agreement (Chalmers 2015; Dietrich 2011). Others have argued there is much more philosophical agreement than may initially seem to be the case, and therefore we can be more optimistic about philosophical progress (Stoljar 2017a, 2017b). In both cases, the shared assumption is that agreement among philosophers is a necessary condition for progress within the discipline.

If the assumption that agreement is necessary for progress strikes us as plausible, it is presumably because we take agreement to be a proxy for truth. We should not value agreement for agreement's sake; we should value it when it is a reliable sign that we're getting things right. If philosophers came to agree on falsehoods, this would not constitute progress.¹

If we are interested in philosophical progress, then, we should ask not only whether agreement has been achieved, but also about the conditions under which agreement would be a reliable sign of truth, and whether philosophical inquiry fulfills those conditions. This question is crucial because, if it turns out that agreement among philosophers is not a reliable sign of truth, then the question of whether philosophers do agree is much less interesting. Surprisingly, the importance of the question about whether philosophical agreement is (or would be) a good indicator of truth has not been fully appreciated in the literature on philosophical progress. Even while the importance of philosophers agreeing on true theories is readily acknowledged, the focal question tends to be whether there is in fact agreement. But of course, if agreement turns out to have little correlation with truth in philosophy, we can stop placing so much stock in agreement.

This paper asks whether philosophical agreement would be likely to be a reliable indicator of truth. I defend the view that for many philosophical questions, agreement among philosophical inquirers is in fact (or would be if it existed) epistemically uninformative: it does not supply even a prima facie reason to believe that the agreed-upon views are true. To argue for this claim, I draw on insights from social choice theory to identify the conditions under which majority agreement among group members is a reliable guide to truth. If the group comprised of philosophical inquirers meets these conditions, then majority agreement in philosophy will generally be epistemically informative: it will supply a prima facie reason to think the agreed-upon views are correct.

One such condition is the competence condition: for majority agreement in a group of reasoners to be a reliable sign of truth within some domain, the group must collectively have a certain level of competence with respect to questions in that domain.² I argue that, due to features of human psychology, the group comprised of philosophical inquirers is likely to fail to meet the competence condition for many philosophical questions. This argument appeals to empirical work in the psychology of human reasoning that demonstrates that people do not process evidence impartially when it comes to inquiry into questions on which they have prior convictions. These biased reasoning tendencies interfere with philosophers' collective competence on certain philosophical questions. Even so, there may still be some philosophical questions for which philosophers are collectively competent in the relevant sense, and thus there may be some philosophical questions for which expert agreement is a reliable sign of truth. However, in practice there are significant barriers to figuring out which philosophical questions are those for which philosophers enjoy a high level of collective competence. It follows that for any given philosophical question, we don't have good reason to believe that expert agreement will be a reliable guide to truth for that question.

The argument that the group of philosophical inquirers often fail to meet the competence condition appeals to psychological characteristics that all reasoners share. However, the conclusion that agreement is epistemically uninformative does not generalize to other disciplines. Rather, distinctive features of philosophical inquiry make philosophy as a discipline uniquely susceptible to the effects of biased reasoning tendencies, whereas these effects are mitigated in other disciplines.

Contrary to what one might expect, the conclusion that agreement within philosophy is epistemically uninformative is good news. Recognizing it frees us to stop holding philosophy to a standard – namely, producing consensus on the “big questions” of philosophy – that it was never capable of achieving. When we come to accept some persistent, intractable disagreement as a permanent feature of philosophical inquiry, we can focus instead on measuring progress by other means.

Regardless of whether my argument is successful, I hope to convince the reader of the importance of questions about the epistemic informativeness of agreement for the debate about philosophical progress. In addition to considering whether agreement has been achieved, we should be considering whether agreement would be a reliable indicator of the truth.

2. Optimism about the epistemic informativeness of agreement

To ascertain the epistemic significance of agreement within philosophy, we need to know whether there is any reason to think that agreement among professional philosophers is, or would be, agreement on the truth. A natural place to look for resources to address this question is the field of social choice theory, the study of collective decision-making procedures. A major ambition of social choice theory has been to determine the conditions under which agreement within a group tracks the truth. The seminal theorem of this field, Condorcet's jury theorem (CJT), establishes that for a group of independent thinkers (or “voters”) each of whom have a greater than 0.5 chance of getting the answer to some factual question right, the chance of the majority being right will approach 1 (certainty) as the group size increases (Condorcet 1785; Grofman et al. 1983).³

The two conditions identified in Condorcet's theorem are known as independence and competence. The independence condition says that voters' opinions must be independent of one another such that any individual group member's vote (i.e., her judgment about which option is correct) must not depend on the vote of any other group member. If, in a group of 100 voters, 99 voters believe P only because the one-hundredth voter believes P, group members' opinions will not be independent in the relevant sense. In this case, the unanimous opinion of the group will be no more likely to be true than the opinion of any randomly selected group member. When independence is violated, higher levels of agreement may be explained by the influence of one voter (or group of voters) rather than the collective wisdom of the crowd. Under these conditions, agreement is not guaranteed to be epistemically significant.

The competence condition says that the average judgmental competence of the group must be greater than chance. Each person in the group has a judgmental competence: a probability of getting the correct answer to the question she is inquiring into that falls somewhere between 0 (no chance of getting the correct answer) and 1 (certain of getting the correct answer). For the CJT result to obtain, the average of the individual judgmental competence levels of all group members must exceed 0.5.⁴

Applying Condorcet's insight yields one way to argue for the epistemic informativeness of philosophical agreement. If a group of reasoners meets the independence and competence conditions of the theorem, majority agreement exhibited by the group will indicate that the agreed-upon answer is likely to be correct, and the agreement will thereby be epistemically informative.⁵ Therefore, if one could establish that the group of professional philosophers satisfy the independence and competence conditions, it would follow that the majority opinion among philosophers will be a reliable indicator of the correct answers to philosophical questions (more reliable, at least, than the opinion of any individual philosopher). According to this Condorcetian justification, agreement on some philosophical view gives us prima facie reason to think that that view is true – that is, it implies that philosophical agreement is epistemically informative.

If successful, this argument would have the result that any degree of agreement among philosophers is epistemically informative, rather than requiring philosophers to meet the higher bar of consensus.⁶ The term “consensus” connotes unanimous or near-unanimous agreement, or at a minimum, higher levels of agreement than a mere majority. Condorcet's theorem implies that, even in cases where the group majority falls short of consensus, the agreed-upon opinion is still more likely to be correct than the opinion of any group member. Therefore, assuming independence and competence conditions are met, any amount of agreement among philosophers provides a prima facie reason to believe the agreed-upon answer, and the larger the majority, the stronger the prima facie reason. So, for example, if approximately 60% of philosophers accept the analytic–synthetic distinction and approximately 70% of philosophers accept the existence of a priori knowledge, all else being equal, the agreement concerning a priori knowledge would constitute a slightly stronger reason to believe than the agreement concerning the analytic–synthetic distinction.⁷ In both cases, the agreement would be epistemically informative.

The Condorcetian justification, if it is correct, would vindicate the value we place on agreement by showing that agreement is a sign of the thing that we really value: truth. It would explain why the lack of consensus in philosophy is troubling to many, and why it ought to be troubling. Many hope that as time goes on, philosophers will achieve some measure of consensus on a greater number of philosophical questions (e.g., Williamson 2006; Wilson 2017). This is a sensible thing to hope for only if greater agreement is evidence that we are finding the correct answers to philosophical questions. If the Condorcetian justification is correct, greater agreement is such evidence. The Condorcetian justification doesn't provide an explanation for why levels of agreement among philosophers are so low compared to other disciplines. But that's okay: its purpose is not to explain why levels of agreement in philosophy aren't as high as they are in the sciences, or as high as we would like. Its purpose is to secure the value of agreement among philosophers by establishing a connection between agreement and likely truth, whether that agreement is great or slight.

However, one might reasonably wonder whether real-world groups, including the group of philosophical experts, can ever satisfy the idealized assumptions of jury theorems and subsequently, about our ability to draw meaningful conclusions about the significance of actual agreement and disagreement from these theorems. In fact, whether ordinary groups of voters can satisfy the competence and independence conditions is a matter of debate.

One suggestion is that real-world groups violate the independence condition, which concerns mutual influence within the group of inquirers. Determining whether a group satisfies independence will involve determining in what sense, exactly, individual inquirers' opinions must be independent for the CJT result to hold. Some commentators have assumed that any conditions that allow for deliberation among group members are sufficient to violate independence (Grofman and Feld 1988: 570; Rawls 1971). But more recently, it has been argued that some forms of dependence within deliberating groups are benign. For instance, Jeremy Waldron argues that voters are independent in the sense required for CJT so long as they do not causally interact after their individual competencies have been assigned; if the competence of one voter depends to some degree on the competence of another, this is irrelevant according to Waldron. “[T]he sort of interaction between voters that would compromise independence would be interaction in which voter X decided in favor of a given option just because voter Y did” (in Estlund et al. 1989: 1327–28). And David Estlund (1994) has shown that even the presence of opinion leaders who are followed to some degree by a proportion of voters within a community is compatible with the CJT result. The presence of this kind of dependence has effects equivalent to that of reducing the size of the group. These results show that there are cases in which the majority agreement of a group can be epistemically significant even when there is some degree of mutual influence among the group members.

How do these considerations apply to the group comprised of expert philosophers? An indispensable component of the philosophical method is engaging with philosophical work that has already been done. Philosophical inquiry is a dialogue in which philosophers routinely read each other's work, talk to each other, and generally allow their opinions to be influenced by the arguments of peers, both past and present. It is beyond a doubt that philosophers' opinions are to some degree mutually dependent. But is this dependence pernicious, or benign? It is uncharacteristic of philosophers to simply fall in line with opinion leaders in the field, so an accusation that philosophical inquiry violates independence because some philosophers “decided in favor of a given option just because [other philosophers] did” is implausible.⁸ Philosophers interact with their peers and take the arguments and evidence that they present seriously, but in the end form their opinions based on their own estimations of the probative force of the relevant evidence. As already mentioned, it has been argued that majority agreement of a group can be epistemically significant under these conditions (Estlund 1994; Ladha 1992). So we have some reason to think that philosophers' opinions are suitably independent.

What about the competence condition? The competence condition of the CJT is concerned with the reliability of group members. To satisfy it, enough group members will have to have high enough judgmental competencies that the group average is greater than 0.5. The question of whether the group of philosophical inquirers meets the competence condition is essentially a question about whether a great enough proportion of individual philosophers are good enough at tracking the truth.⁹

It may seem obvious that the group comprised of professional philosophers satisfies the competence condition. Philosophers are experts who have spent years cultivating the ability to assess philosophical claims and arguments. And as experts in their field, we should expect philosophers to be much more reliable than chance when it comes to answering philosophical questions. If so, then the group comprised of philosophers will satisfy the competence condition.

Sometimes, features of a group's evidence can negatively impact the group's judgmental competence. For example, when a group's evidence is massively misleading, this can drastically decrease individual and group judgmental competence.¹⁰ But we have no reason to think that evidence that bears on philosophical questions in particular is massively misleading. So the possibility of misleading evidence poses no threat to the claim that philosophers satisfy the competence condition – or at least, no threat that is distinct from the threat posed by global skepticism generally.

From the discussion so far, it appears that the Condorcetian justification may provide a basis for thinking that agreement among philosophers is epistemically informative. As I will argue in the next section, however, this conclusion is premature. The Condorcetian justification fails because philosophical inquirers have psychological tendencies that render them collectively unlikely to meet the competence condition of the CJT.

3. Pessimism about the epistemic informativeness of agreement

Recall that the competence condition says that the average judgmental competence of the group of inquirers must exceed 0.5. If a group of inquirers fails to meet this condition, majority agreement within the group cannot be a reliable indicator of the truth. So in order to secure the significance of philosophical agreement, we need to have good reason to believe that philosophers typically satisfy the competence condition for philosophical questions. But we don't have good reason to think that philosophers typically satisfy the competence condition. Therefore, we don't have good reason to think that philosophical agreement is typically a sign of truth. But if we don't have good reason to think that philosophical agreement is typically a sign of truth, then we can't take agreement among philosophers as a reason to think that the agreed upon view is true – in other words, philosophical agreement is not epistemically informative. More formally:

Argument against epistemic informativeness

(1) We lack good reason to think that the group comprised of philosophical inquirers has an average judgmental competence that exceeds 0.5 with respect to many philosophical questions.

(2) If we lack good reason to think that a group of inquirers has an average judgmental competence that exceeds 0.5 for a question, agreement within that group is epistemically uninformative with respect to that question.

Conclusion: For many philosophical questions, agreement among philosophers is epistemically uninformative.

The argument is deductively valid. (2) expresses the Condorcetian requirement that a group's meeting the competence condition is a necessary (but not sufficient) condition for agreement within the group to reliably indicate the correct answer to the question. This section and the next are devoted to defending (1) – the claim that, for many philosophical questions, we lack good reason to think that philosophers meet the competence condition.

Why think (1) is correct? A central piece of the answer is that philosophers, like all human beings, are subject to certain reasoning tendencies that impair our ability to treat the evidence for and against a given proposition impartially. These tendencies pertaining to how we gather and evaluate evidence often go under the heading of myside bias: a well-documented tendency to gather, generate, and evaluate evidence in a way that privileges one's prior beliefs and attitudes.¹¹

Myside bias has been extensively documented in numerous empirical studies. In a classic study, Lord et al. (1979) had participants examine two fictitious studies: one which provided evidence that capital punishment is a deterrent to murder, and another which provided evidence that capital punishment is not a deterrent to murder. Lord et al. found that those individuals who were, prior to the study, proponents of the death penalty found the pro-deterrence study significantly more convincing, whereas those who were, prior to the study, opponents of the death penalty found the anti-deterrence study significantly more convincing. The authors referred to this phenomenon as the biased assimilation of evidence, whereby peoples' existing theories and expectations will influence their judgments about the reliability, relevance, and meaning of evidence. “Even a random set of outcomes or events can appear to lend support for an entrenched position, and both sides in a given debate can have their positions bolstered by the same set of data” (Lord et al. 1979: 2099, emphasis added).

Our tendency to treat evidence impartially is not limited to our evaluations of existing evidence, as in Lord et al. (1979), but also manifests itself in a tendency to generate arguments for positions that we agree with (Macpherson and Stanovich 2007; Perkins 1985; Toplak and Stanovich 2003) and a tendency to actively seek out evidence and sources of information that are likely to support what we already believe (Hart et al. 2009; Kunda 1990; Taber and Lodge 2006).

The prevalence of myside reasoning tendencies is by now extremely well evidenced, the effect having been confirmed by multiple research groups in different settings using a variety of experimental designs. It has also been shown that this is a robust effect that exists throughout the population: intelligence (measured by cognitive and verbal ability) does not correlate with myside reasoning tendencies, so being highly intelligent does not inoculate against processing the evidence in a biased way (Perkins 1985; Perkins et al. 2012; Stanovich et al. 2013; Toplak and Stanovich 2003).¹² In fact, there is some evidence that individuals with high intelligence are better able to put their verbal and cognitive abilities to work justifying their pre-existing convictions (Kunda 1990; Perkins et al. 2012).

This body of work strongly suggests that philosophers, like the rest of the population, will not assess the evidence and arguments for philosophical positions in a neutral, impartial way. They will tend to construct arguments for positions that they already believe and to offer their strongest criticisms of philosophical positions and arguments that they already disagree with. When engaging with new arguments that bear on a strong conviction they already hold, philosophers will tend to reason in a way that is driven by a desire to maintain and defend their pre-existing beliefs.¹³ These realities mean that we should expect the prior beliefs of a philosophical inquirer to impact the philosophical views they ultimately arrive at.

The claim that philosophers routinely engage in motivated reasoning is not a new one, and some philosophers writing about philosophical progress and methodology have already signaled their openness to it. David Chalmers writes: “When we address arguments against our views, we sometimes work backwards from our rejection of the conclusion to see which premises we have to deny, and we deny them” (2015: 18). William Lycan, in his book on philosophical methodology, offers a number of observations (in his terms, “cynical sociophilosophical observations”) about how philosophers tend to privilege their own views and arguments (Lycan 2019). And there is the well-known adage, “One man's modus ponens is another man's modus tollens.” So to some degree, the presence of motivated and partial reasoning tendencies is already recognized within the profession. To this extent, my claims so far should not be surprising.

What does the presence of myside bias mean for the judgmental competence of individual philosophers? The reality of motivated reasoning means that philosophers considering matters on which they have prior opinions will tend to emerge from their inquiries having strengthened their confidence in whatever opinion they espoused prior to inquiry. This suggests that a philosopher's judgmental competence with respect to a philosophical question on which she has a strong prior opinion will depend on whether her initial opinion on the matter is correct. If her initial answer to the question is correct, then she will have a high judgmental competence with respect to the question. But if her initial answer to the question is incorrect, then she will have a low judgmental competence. The presence of myside bias means that whether one comes to inquiry with the correct view about some question impacts one's judgmental competence on that question.

Recall that premise (1) is about group judgmental competence. How myside reasoning tendencies impact the average judgmental competence of the group will depend on the distribution of prior opinions within the group. In a group where the pre-inquiry opinions of a majority of members are mostly correct, the average judgmental competence of the group will likely exceed 0.5. In a group where the pre-inquiry opinions of a majority of group members are mostly incorrect, the average judgmental competence of the group will likely fall below 0.5. In a group where the pre-inquiry opinions of group members are more evenly distributed, the average judgmental competence of the group will be closer to 0.5. In short, we won't be able to make any claims about the average judgmental competence of the group without knowing further facts about what beliefs the group members had prior to inquiry.

According to the above reasoning, there will be cases in which group members are impacted by myside bias and yet the average judgmental competence of the group exceeds 0.5. This will typically happen when a large proportion of group members start out with the correct view. Perhaps some philosophical questions are like this. One plausible candidate is the question of whether there are mind-independent external objects. Suppose (as is likely) that most philosophers come to study the issue of skepticism with a pre-existing belief in the existence of a mind-independent reality. Suppose further that there really are mind-independent external objects. Then, the presence of myside reasoning tendencies won't decrease the judgmental competence of individual philosophers with respect to this question; rather, their tendency to reason in a motivated way will mean that individuals will have better-than-chance odds of getting the answer right, and that the group's average judgmental competence will be high.¹⁴

If most philosophical questions are like this – if, prior to philosophical inquiry on the question, most philosophers already believe the correct answer – then we have good reason to reject (1). In that case, for many philosophical questions, philosophers would be likely to have an average judgmental competence that exceeded 0.5. Unfortunately, however, we have no reason to think that most or many philosophical questions are ones for which philosophers somehow serendipitously start out with the correct views. What's more likely is that, for most philosophical questions, the landscape of pre-inquiry beliefs is mixed. If so, then we have yet to find a reason to think that the group comprised of philosophers generally has an average judgmental competence that exceeds 0.5 – that is, we have yet to find a reason to think that philosophers generally meet the competence condition of the CJT.

One way we might make more progress on the question of when philosophers have a high judgmental competence with respect to a philosophical question is by further probing the conditions under which myside reasoning tendencies manifest. As already noted, myside bias shows up when people have a strong opinion about the matter under investigation prior to beginning their inquiry. This suggests that we can make some headway on the question of philosophers' group judgmental competence by figuring out which kinds of philosophical questions tend to elicit strong opinions prior to conducting philosophical inquiry. The next section takes up this question.

4. Which philosophical questions are implicated by myside bias?

In the previous section, we saw that the existence of myside reasoning tendencies means that the assessment of a group's average judgmental competence will depend on facts about what views group members begin with, and in particular, whether those views are correct. Of course, there is no easy way to say which philosophical questions are ones for which philosophers start out with a strong conviction of the correct answer. While it is comparatively easy to get a good sense of what beliefs each philosopher holds (or held) about the answers to philosophical questions before engaging in philosophical inquiry, it is near impossible to say with any certainty whether those initial beliefs are correct. To do so, we would need to know the answers to philosophical questions. But of course, if we already knew the answers to philosophical questions, there would be no need to figure out whether agreement about the answers to those questions is a good guide to truth. Indeed, there would be no point in engaging in further inquiry into those questions at all. This means that facts about both individual and group judgmental competence for particular questions will often elude us.

Because of the considerations adduced in the previous paragraph, it's fruitless to try to determine whether philosophers meet the competence condition for specific philosophical questions. However, perhaps we can draw general conclusions about the kinds of philosophical questions that tend to elicit motivated reasoning tendencies. We know that people are most often susceptible to motivated reasoning about an issue when they already have strong beliefs (or “convictions”) concerning that issue.¹⁵ If it turns out that there are some philosophical questions that tend not elicit strong pre-theoretical beliefs, then there remains hope that philosophers have comparatively high judgmental competence with respect to that subset of philosophical questions.

It is a sociological question whether there are any philosophical questions that tend not to elicit strong beliefs among those who have not yet undertaken inquiry into those questions. Nevertheless, it seems fairly clear that there are such questions. For example, it seems unlikely that, prior to engaging in philosophical inquiry, most people have convictions about philosophical issues that are highly theoretical or abstract, such as the rules of modal logic, the nature of abstract objects, the existence of the analytic–synthetic distinction, or the nature of epistemic justification. Likewise, it is evident that prior to engaging in philosophical inquiry many people have strong convictions about philosophical matters that are more concrete or applied in nature – for instance, questions about what it is to live a good or meaningful life, the existence of a God or Gods and what God requires of humans, or the proper role of the government. If so, motivated reasoning will play a bigger role in determining the distribution of philosophical views for the latter questions than it will for the former. If this is correct, then agreement may still be a sign of truth for the former kind of philosophical question (abstract, theoretical), but not for the latter (concrete, applied).

Another way of classifying philosophical questions is by their scope. In each sub-discipline of philosophy, there are questions of both wide and narrow scope. The way in which a wide-scope question in a sub-discipline is answered has implications for how a large number of other questions in that sub-discipline (and sometimes in other sub-disciplines) can be answered; in other words, wide-scope questions have answers with many inferential connections to other questions. The way in which a narrow-scope question in a sub-discipline is answered will have fewer implications for other questions in the same sub-discipline; narrow-scope questions have fewer inferential connections. For example, consider two metaethical questions: (i) Is realism about moral properties true?; and (ii) Is the most fundamental division in metaethics between realists and anti-realists, or between intuitionists and non-intuitionists?¹⁶ The former question has a wider scope. This is a rough characterization of the distinction between wide-scope and narrow-scope questions, but I take it that most people will recognize the basic idea behind this distinction, which is just that some philosophical claims have farther-reaching implications than others.

Just as ordinary people seem more likely to have convictions about concrete and applied questions than they are to have convictions about abstract and theoretical questions, we might think that prior convictions about the answers to wide-scope questions are more common than prior convictions about the answers to narrow-scope questions. Many traditional questions of philosophy – epistemological questions about what we can know, ethical questions about how we ought to live, metaphysical questions about free will and God, and so on – are wide-scope questions, and we are much more likely to find people who have strong convictions about these matters in the general population than we are to find people who have strong convictions about specific narrow-scope questions in epistemology, ethics, or metaphysics (e.g., questions about the nature of epistemic warrant, the consistency of a particular version of utilitarianism, whether the consequence argument against compatibilism is sound, etc.). For this reason, we might think that the philosophical questions that tend to produce strong pre-theoretical convictions are wide-scope questions, and subsequently, that reasoning about wide-scope questions is more likely to be subject to bias than reasoning about narrow-scope question. If so, then agreement about the answers to narrow-scope questions might be epistemically informative, even if agreement about the answers to wide-scope questions is not.

While I think there is something to these arguments, things get more complicated when we consider the fact that philosophers sometimes form strong attachments to their philosophical views – even views about the answers to theoretical, abstract, or narrow-scope questions – early on in their careers. An advanced undergraduate or new graduate student sometimes forms a strong opinion that a particular view is correct before developing true expertise on that topic and before they have had a chance to survey a representative sample of evidence. This kind of early attachment may be especially likely when a person is part of a community of thinkers for whom a commitment to the view is highly valued or simply unquestioned, or when a person is educated in a context in which a particular philosophical school of thought or movement dominates. One thinks, for example, of G.A. Cohen's observation that most of his peers at Oxford, like him, believed in the analytic–synthetic distinction, while those in his generation educated at Harvard mostly did not (Cohen 2001). Similar observations might be made about the philosophical views of students at the University of Pittsburgh who were taught by members of The Pittsburgh School (Maher 2014), or the views on animal rights held by students of members of the Oxford Group (Garner and Okuleye 2020).

Moreover, even if it is true that philosophers do not have strong prior convictions on narrow-scope, abstract, or theoretical questions, inquiry on these matters may still be subject to myside bias if other premises concerning which a reasoner does have strong convictions enters into their reasoning on the matter.

The upshot of this is that we need to be cautious in drawing generalizations about which philosophical questions are likely to elicit a biased assessment of the evidence based solely on the kind of philosophical question at play. When it comes to determining which philosophical questions elicit myside reasoning tendencies, wide-scope, concrete, applied questions surely pose more of a danger than narrow-scope, theoretical, abstract questions. But even the latter kind of question is not invulnerable. These considerations establish that we lack good reason for thinking that, with respect to any given philosophical question, the group comprised of philosophical inquirers has an average judgmental competence that exceeds 0.5. This doesn't necessarily mean that individual philosophers' views can't be true, can't be justified, or that philosophers can never have knowledge of philosophical claims. It does mean, however, that the Condorcetian justification fails with respect to many philosophical questions (though we don't know for sure which ones). And, since the CJT articulates conditions necessary for agreement to be a reliable indicator of the truth, it also follows that generally, we shouldn't take agreement in philosophy as a sign of truth.¹⁷

5. Two objections

So far, I've argued that for most philosophical questions, agreement among philosophers is epistemically uninformative. Before we consider the implications of this conclusion for philosophical progress in the final section, I want to address two objections. The first is an alleged problem posed by philosophers who change their minds. The second is the objection that the argument proves too much, since it applies not only to philosophical inquiry but also to inquiry generally.

It might be thought that the existence of philosophers who change their minds – even sometimes in spite of having held strong convictions on a topic prior to the conversion – poses a problem for the claim that philosophers, like everyone else, are prone to respond to philosophical evidence and argument with an eye toward fortifying their own pre-existing beliefs. The widespread influence of myside reasoning tendencies seems incompatible existence of “conversions” – cases in which philosophers change their minds about some issue in response to evidence and argument.

There are some striking examples of conversions in the philosophical literature. Wittgenstein famously rejected the picture theory of propositions he put forward in the Tractatus. Laurence BonJour changed his mind about whether the coherence theory of epistemic justification was correct after producing numerous publications defending it. Frank Jackson changed his mind about whether his knowledge argument against physicalism is successful, in addition to changing his mind about the truth of physicalism. These examples are particularly dramatic because they involve philosophers rejecting views for which they had come to be well-known advocates. And these are only the examples that are widely known because the conversions were documented in a very public way; there are surely many more cases of philosophers quietly changing their minds after thinking about philosophical questions more carefully. These kinds of conversions, it might be thought, pose an explanatory obstacle for the view that philosophers typically reason in a motivated way.

In response, we can first note that the only kind of cases that might be thought to pose a threat are cases in which philosophers begin with strong convictions about some philosophical issue and then later change their minds. Most of the examples in the previous paragraph are presumably of this sort. When someone begins their inquiry into some issue with a pre-existing belief to which they do not have a strong attachment, myside reasoning tendencies will not be at play, and they are more likely to be able to give an unbiased assessment of the evidence. So, cases in which philosophers change or refine their views on philosophical issues concerning which they do not have strong prior convictions are perfectly compatible with the argument of the previous section. In addressing this objection, then, we need to focus on cases in which the philosophers who changed their minds had strong convictions that were subsequently overturned.

The existence of some conversions poses no threat to the view that philosophers are often subject to motivated reasoning. Motivated reasoning is a tendency; it is not the only psychological force at work in peoples' reasoning. The scientific work on myside reasoning reveals a positive correlation between individuals' prior convictions and the views they end up with after receiving and processing new evidence; it does not show that every individual with strong convictions is psychologically determined to strengthen those convictions. What's more, the prevalence of myside bias can help explain why conversions of this kind are relatively rare: in cases involving strong prior convictions, the tendency toward motivated reasoning makes it unlikely that people will convert.

The second objection is that the argument against epistemic informativeness proves too much, since it can easily be modified to apply to inquiry in other fields.¹⁸ Since motivated reasoning is a general feature of human inquiry, the argument can be altered to pertain to any discipline we choose. But if the argument applies equally well to any discipline, it appears suspect: surely consensus in the sciences is a reliable indicator of truth.

There is, of course, no good reason for thinking that scientists are less prone to myside reasoning tendencies than philosophers are. So, if it is true that philosophical agreement is typically not epistemically significant, but scientific agreement is, the explanation will be found in some difference between philosophical and scientific inquiry, method, or practice, and not in a difference in the reasoning tendencies or cognitive habits of its individual practitioners.

To give a satisfying answer to the question of why philosophical inquiry is more susceptible to the effects of biased reasoning than scientific inquiry will involve addressing the question of what (if anything) distinguishes philosophical inquiry from scientific inquiry. This is a formidable question whose answer is beyond the scope of this paper. However, we can identify some differences between philosophy and the sciences that mitigate against the effects of biased reasoning in the latter.

One notable difference is that for many philosophical questions, there are often serious disagreements among philosophers not only about the answers to those questions, but about what would constitute evidence for or against a particular answer. Take, for example, the intense disagreements among philosophers about the evidential role played by “intuitions” in philosophical inquiry (Bealer 1996, 1998; Goldman 2007; Nagel 2012; Stich 1988; Williamson 2000: Chapter 7, 2004). Lack of agreement about such a basic disciplinary norm as what kinds of propositions can count as evidence means that philosophy has few “consensus premises” – claims that are regarded by the majority of philosophers as undeniable (Chalmers 2015: 16–18). Without consensus premises, arguments that will persuade everyone (or nearly everyone) in the philosophical community are impossible to construct. Lack of substantial agreement on what counts as evidence introduces the possibility of selecting and interpreting evidence in a way that favors one's pre-existing commitments. Insofar as these kinds of fundamental disagreements about disciplinary norms are less prevalent in the sciences, there will be fewer opportunities for personal bias to enter the picture.

Another way in which philosophical questions differ from scientific questions is that the former are less frequently decidable through the collection of empirical data. The central role of empirical data in the sciences means that scientific consensus is often achieved by an accumulation of empirical evidence through ongoing experimentation, sometimes accompanied by an increase in precision in the tools used for measuring empirical phenomena. For example, phlogiston theory, the dominant theory of combustion in the eighteenth century, was eventually shown to be false through a series of increasingly sophisticated combustion experiments, many of which were devised and conducted by Antoine-Laurent Lavoisier in the 1770s. Although many proponents of phlogiston theory revised and elaborated the theory in an attempt to accommodate the new experimental findings, eventually the strain to incorporate new data became too great for the old theory, and phlogiston theory was dead by the early 1800s (Conant 1957). Similarly, disagreement in the early twentieth century over whether X-rays and gamma-rays are composed of particles or waves was eventually resolved through the accumulation of empirical data. There is now consensus among physicists that electromagnetic radiation is a form of light that exhibits the properties of both waves and particles (Wheaton 1981, 1991). In many cases, scientific consensus is achieved when the scientific community collectively responds to an accumulating preponderance of empirical evidence supporting the agreed-upon theory.

While empirical data are certainly relevant to philosophical inquiry, they are rarely if ever able to conclusively settle philosophical questions in the same way they often do in the sciences. This may be partly due to the fact that philosophical arguments frequently include as crucial premises definitions or value judgments which can more plausibly be denied than empirical observations.

All this is not to deny that scientists often disagree, nor that they sometimes consider philosophical questions or make assumptions that are not themselves empirically established in the course of their inquiries. We might think, for example, of cosmologists' informed speculations on the origins of the universe or of disputes over methods of data analysis in the social sciences. Insofar as these kinds of questions or assumptions cannot be settled by empirical means, those inquiring into them are more susceptible to the effects of motivated reasoning. However, the point is that these kinds of empirically intractable questions and assumptions in the sciences are rare when compared to philosophy. This, along with the methodological disputes discussed earlier, can render philosophical questions more susceptible to the influence of motivated reasoning than scientific questions.

Where does this leave us? With the conclusion that, for any given philosophical question, we don't have good reason to take agreement among philosophers on the answer to that question as a reliable indicator of the truth. We should be skeptical about the epistemic import of philosophical agreement.

6. Implications for philosophical progress

It might seem obvious that agreement among philosophers would provide a reason to believe the agreed-upon answer. Drawing on results from jury theorems, we have identified conditions under which agreement would be epistemically informative and argued that we lack good reason to think that the group comprised of philosophical inquirers meets one of the necessary conditions for agreement to be epistemically informative. We now turn to the question of what agreement (or the lack of it) can tell us about philosophical progress.

This essay began with the observation that, in discussions of philosophical progress, lack of agreement is often acknowledged to be a legitimate if not insurmountable problem for the discipline of philosophy. It is not unusual for contributions to this literature to begin with a lamentation that philosophers have not been able to agree on the answers to the big questions of their discipline. Just over a century ago, Arthur Lovejoy identified agreement with progress in his 1916 presidential address to the American Philosophical Association: “In our own subject [of philosophy] … if we fail to achieve a measurable amount of agreement and a consecutive and cumulative progress there, we fail altogether. … The fact [that we don't] remains, then, a standing scandal to philosophy, bringing just discredit upon the entire business in which we are professionally engaged” (Lovejoy 1917: 129–30). And in the introduction to a recent anthology on progress in philosophy, Russell Blackford identifies lack of agreement as a real problem for the discipline: “Philosophy proceeds, supposedly, by way of rational inquiry and argument, yet, as Jonathan Glover has written, ‘philosophers persistently disagree’ to such an extent that the ‘apparent lack of clear progress or a body of established results is an embarrassment’” (Blackford 2017: 1). These passages suggest the following argument against philosophical progress:

Argument against Philosophical Progress

(1) If philosophers haven't arrived at a consensus about the answers to any philosophical questions, then philosophy hasn't made any progress.

(2) Philosophers haven't arrived at a consensus about the answers to any philosophical questions.

Conclusion: Therefore, philosophy hasn't made any progress.

If the argument is sound, philosophy has not progressed. This would undoubtedly be a regrettable conclusion for philosophy. However, what the foregoing discussion about the epistemic informativeness of philosophical agreement suggests is that there may not be as tight a link between agreement and progress as this argument assumes. Premise (1) claims that consensus is a necessary condition for progress. This assumption makes sense if we think of consensus as a proxy for truth or knowledge. But if agreement (and a fortiori, consensus) supplies no reason to think that the agreed-upon answer is correct, it's unclear why we should take agreement as necessary for progress.

The correct response to the above argument is to reject the assumption that consensus is a necessary condition for philosophical progress. This is already a familiar move within the literature on philosophical progress. As others have pointed out in response to the line of reasoning advanced in the above argument, convergence on the truth is one way of measuring progress in a discipline, but it is not the only way. Bertrand Russell famously argued that the value of philosophy “must not depend upon any supposed body of definitely ascertainable knowledge to be acquired by those who study it”; rather, the value of philosophy is to be measured by its success in enlarging our understanding of the possible answers to philosophical questions.¹⁹ Other philosophers have followed Russell in this, arguing that philosophy as a discipline progresses by making connections between disparate areas of inquiry (Jackson 2017); by evoking and refining conceptual landscapes (Pigliucci 2017); by clarifying concepts, problems, definitions, and theories (Kamber 2017); and by developing better and better models (Williamson 2017). All these achievements of philosophical inquiry contribute to our understanding of the world and our place in it, and should be counted as progress, even if philosophers are unable to deliver definite answers to philosophical questions that result in consensus.

It's interesting to note that many philosophers who believe that philosophy's primary value is not in securing convergence on the truth often seem very committed to their own philosophical views (e.g., think of Russell's political and anti-religious writings). It might be thought that this reveals a kind of incoherence in their overall views: they think philosophy can't secure truth or knowledge through agreement, yet they themselves proceed to argue in such a way that appears to indicate that they believe it's possible to make progress toward truth in these areas.

However, philosophers who recognize philosophy's limited ability to secure agreement can consistently argue sincerely for their own philosophical views. Recognizing that there is considerable disagreement concerning some matter and that one's own reasoning may be susceptible to error is compatible with doing one's best to let one's inquiry be guided and motivated by a desire to discover the truth, and with continuing to believe and argue for one's conclusions. Presumably, this is what many philosophers who are skeptical of the epistemic value of convergence of opinion take themselves to be doing.

If philosophy's primary value is in its ability to clarify and refine concepts, to help us better understand the possible answers to philosophical questions, or to make connections between different areas of inquiry, this isn't merely a consolation prize for having missed out on securing the greater good of converging on truth. These things are genuinely valuable. How can we make informed and principled decisions about, for example, how best to protect democracy, if we don't get clear on what we are trying to protect, and identify the range of possible justifications for doing so? This is philosophical work, and it is important work, even if it doesn't result in consensus.

The charge that the discipline of philosophy is foolish or broken because philosophers can't agree is based on a misconception: the misconception that the main point of philosophical inquiry is to converge on the right answers to life's big questions. While this is certainly one way that philosophical inquiry could be helpful, it is not the only way. Given what we know about the nature of philosophical inquiry and the nature of human psychology, we should not expect to see much agreement among philosophers. But neither should we view this as a problem for philosophy, since the true value of philosophy does not depend on its ability to secure consensus.

To be sure, converging on the truth is a great good. And if we could be reasonably sure that agreement among philosophers was a reliable guide to the truth, that would be an excellent thing. But it's misguided to expect that philosophy alone will give it to us; philosophy was never meant for that job.²⁰