1 Introduction

What kind of activity is reasoning? The answer to this question is anything but obvious, if only because reasoning can take many different forms. Take the following examples:

  1. a.

    A gynaecologist calls her patient and says: ‘Given the result of your blood test, we think you should come in for an extra ultrasound, for we don’t want you to end up with an ectopic pregnancy’ (the life-threatening situation of an embryo becoming implanted in the ovarian tube). The patient concludes that this is probably what is happening, and starts to panic: she might be dying. Later, it turns out that her blood test only indicated a slight statistical risk, and that everything is in fact fine.

  2. b.

    I tell my daughter: ‘you should at least try one bite of this bread, otherwise you’ll never know whether you like it or not.’ She answers: ‘I tried it once and I didn’t like it!’ I try to convince her (without success) that tastes often change over time.

  3. c.

    While writing this paper, I turn over various arguments and counterarguments in my mind, assessing them in terms of their validity.

The gynaecologist is reasoning to clarify a decision to the patient, the patient subsequently uses this information to reason towards her own conclusion about the situation. I reason with my daughter to make her eat some bread; while writing a paper, I reason with myself in trying to develop a philosophically convincing analysis. Reasoning can thus be theoretical (aimed at theoretical knowledge) or practical (aimed at making someone do something). It can be done silently or aloud, in conversation with others or on one’s own, on topics ranging from the obstruse to the mundane. As point of departure I will adopt a very minimal working definition of reasoning, stating that what we are investigating is the activity of making inferences by attending to reasons (a definition taken from Mercier & Sperber 2017, p. 53). The definition adopted here implies that several other uses of the term ‘reasoning’ will be left aside: I will not be concerned with forms of cognition that are inferential but do not involve ‘attending to reasons’, such as the way machines make inferences or the way humans (according to certain theories) engage in non-conscious inferential processingFootnote 1. What I will be concerned with is thus making conscious, explicit inferences, either aloud or silently, and usually but possibly-not-exclusively by means of languageFootnote 2.

The general concern of this paper is the question how to understand this activity. More specifically, its aim is to show how it is a normative activity: in other words, how in reasoning we are inherently trying to get things right. This claim has become contested because our understanding of reasoning has recently undergone a quite radical transformation. Traditionally, philosophers took reasoning to be the capacity to apprehend the normative force of reasons: the capacity (uniquely ascribed to human beings) to see what is right and what is trueFootnote 3. From such a point of view, it is self-evident that reasoning as activity is inherently geared to getting things right. However, this picture of reasoning has come under pressure by what could be called the ‘social turn’ in the philosophy and psychology of reasoning, which has brought about a shift towards understanding reasoning first and foremost as a social activity. As I will show, the idea that reasoning should be understood as a social tool is often taken to imply that reasoning is thus not an activity geared to getting things right.

The claim I will defend in this paper is that even though the ‘social turn’ has greatly improved our understanding of reasoning, this turn does not warrant the conclusion that reasoning is not a normative activity. After giving a brief outline of the social turn in section two, section three will zoom in on Mercier and Sperber’s (2017) highly influential account. I will show how their interactionist view attempts to replace the idea that in reasoning we try to get things right with the idea that in reasoning we aim to justify ourselves and persuade others to accept our views. I will show that even though interactionism rightly points out that in reasoning we are guided by many other considerations than rationality, Mercier and Sperber fail to convince in their claim that reasoning is not a normative activity. In fact, I believe it is a mistake to try to throw the normative baby out with the intellectualist bathwater. As I will show in sections five and six, Gilbert Ryle’s conceptualization of reasoning as ‘didactic discourse’ provides a surprisingly relevant analysis which gives a fruitful example of how reasoning can be understood as an activity which is both social and normative.

2 The ‘social turn’ in the philosophy and psychology of reasoning

Many contemporary theories and empirical studies of reasoning now emphasize the fact that reasoning is a social activity. This general idea is usually spelled out in terms of two specific claims. The first claim is that reasoning is mostly and primarily done together with others, and only rarely, in certain specific circumstances, on one’s own. In other words, the physicist scribbling away on her or his blackboard, or my turning over philosophical arguments in my mind, should not be seen as core examples of reasoning but as pretty specific and derivative versions of the phenomenon (Laden, 2012; Dutilh Novaes, 2013, 2015).

The second claim is more radical: it is the idea that the functions of reasoning are social as well. Jonathan Haidt’s (2001) highly influential Social Interactionist Model (SIM) of moral judgement can be seen as an early version of this idea. Although the SIM is not a model of reasoning per se, the feature of his model that generated the most attention (see for example Saltzstein & Kasachkoff 2004, D’Cruz 2015, Summers 2017) is the role Haidt ascribes to moral reasoning. In the SIM, reasoning does not play a causal role: instead, its role is to provide post-hoc rationalizations. According to Haidt, the role of moral reasoning is not to test or correct our moral intuitions, but to justify them towards others (Haidt, 2001; Haidt & Bjorklund, 2008). In developing the SIM, Haidt thus also attempted to radically debunk the traditional understanding of reasoning by suggesting that reasoning might have a primarily social function.

Meanwhile, a parallel ‘debunking movement’ developed in the psychology of reasoning, predominantly in the huge body of work on the Wason Selection Task (the main paradigm for studying conditional reasoning, Wason 1966). The take home message of those studies was that people are notoriously bad at this kind of reasoning in laboratory settingsFootnote 4. Fierce debates followed (the so-called ‘rationality wars’, see Sturm 2012), which led to a rethinking of the role of normative systems such as logic, probability theory, and rational choice theory in explanations of reasoning (Chater & Oaksford, 2012; Holyoak et al., 1995). Importantly, the rationality wars also opened the discussion on the question whether reasoning is guided by norms at all (Stich, 1990; Gigerenzer, 2007; Elqayam & Evans, 2011). This debate inspired the hypothesis that reasoning might not be a capacity to access the True and the Good, but should instead be seen as a highly useful tool developed to facilitate social interaction.

Although different social theories of reasoning have been brought forward (Elqayam & Evans, 2011; Laden, 2012; Mercier & Sperber, 2011, 2017; Resnick et al., 1993), Hugo Mercier and Dan Sperber’s interactionist account (2017) is currently seen as the prime representative of the social approach. Their theory builds on both the debunking movement initiated by Haidt, and on doubts about the normativity of reasoning opened up by the rationality wars. In the next section I will summarize the main tenets of their view, and show how it rejects the traditional understanding of reasoning as an activity geared towards ‘getting things right’.

3 Mercier and Sperber’s interactionist account

As said, the core claims developed in the social turn are that we reason primarily with others, and that reasoning is done in order to achieve social aims. In their book The enigma of reason, Mercier and Sperber defend both claims. With regard to the first point, they make an effort to show that even famous solitary geniuses (such as Isaac Newton or Werner Heisenberg) were actually embedded in a community of peers who contributed and responded to their work (2017, p.320). However, at the core of their interactionist theory is the stronger claim: they argue that the aims of reasoning are social in nature, and that reasoning ‘is not properly geared to the pursuit of knowledge and good decision’ (2017, p.180). According to Mercier and Sperber, providing reasons to others has two main social functions: firstly we often give reasons to justify ourselves to others in order to enhance our reputation (p.186). This is similar to Haidt’s suggestion that in reasoning we aim to preserve or increase our social status within the group (Haidt & Bjorklund, 2008). Its second function, according to Mercier and Sperber, is persuasion. By providing reasons we try to actively win the trust of others, something which is necessary because humans evolved to be epistemically vigilant (2017, p.194).

Epistemic vigilance is ‘a suite of cognitive mechanisms […], targeted at the risk of being misinformed by others’ (Sperber et al., 2010, p359, see also Mercier 2020)Footnote 5. Epistemic vigilance mechanisms can be of two kinds: either directed at the source of the message, or the content of the message. Mechanisms of the first kind are geared to determining the reliability of one’s interlocutor (Origgi, 2019; Mercier, 2020): is this person a reliable/ informant? Is the person competent, benevolent, or does he or she have stakes in deceiving me? In describing mechanisms of the other kind, Mercier states that ‘some mechanisms examine whether a message is compatible with what we already believe to be true, and whether it is supported by good arguments’ (Mercier, 2020, p.xvi). This means that we also use reasoning to evaluate reasons provided by others (2017, p.7), in which case reason does seem to be ‘geared to the pursuit of knowledge and good decision’. Mercier and Sperber argue that reason indeed has ‘two faces’ (p.235): whereas in evaluating the reasons provided by others we are unbiased and demanding, we are lazy and biased when we are providing our own reasons.

To sum up: when we provide reasons to others, we are not concerned with getting things right, but with justifying our views to others in order to enhance our reputation, and with persuading our audience to accept our views. However, this threatens to come down to a pretty cynical view of what we do when we provide reasons: this seems merely a rhetorical exercise. Mercier and Sperber pay ample attention to this question−in fact they are adamant that they do not embrace a cynical view:

‘A first misunderstanding that we encountered again and again consists in attributing to us the view that argumentation is just a way to manipulate and deceive others and that it has no real intellectual merit. This very cynical view of reasoning and argumentation must have some appeal - possibly that of making one feel superior to naïve ordinary folks. To the risk of disappointing some of our readers, this is a view we do not hold and a cynicism we do not share.’ (2017, p331)

Instead, they argue that the activity of producing reasons still has intellectual merit, though admittedly it has so only in a derivative sense. Although producing reasons is not about getting things right, the reasons we come up with represent intuitions that are often right (2017, p6). Intuitions are the outcomes of well-adapted inferential processing modules, which ‘take as input representations of particular facts and use specialized procedures to draw conclusions from them’ (2017, p118, see also p198). The cognitive operations of such modules can be called rational in the minimal sense that they ‘contribute to humans’ cognitive efficiency’ (p144). Inferential processing thus tracks relations between facts, and spits out intuitions: and even though we can’t provide grounds for these intuitions (after all, we are not aware of the underlying processes) they will often turn out to be right, as they are produced by well-adapted inferential modules.

Mercier and Sperber thus argue that when we produce reasons, the reasons we come up with are representations of these intuitions. In some situations, we use these representations to consciously derive further conclusions by making explicit inferences on the basis of normative systems such as logic or probability theory: this is an activity they call reflection (2017, p150). However, they surmise that this kind of activity is very rare. As they note, psychologists and philosophers might have been misguided about the prevalence of reflection because engaging in such reflection is basically what they do for a living: researchers might ‘mistake their own professional twist of mind for a basic human trait’ (2017, p152). In contrast, Mercier and Sperber argue that for most people for most of the time, the reasons they give are not used for reflection, but for interpersonal justification or persuasion.

To summarize: according to the interactionist theory, the reasons we produce are representations of our intuitions. These intuitions are the outcome of inferential mechanisms which are efficient in tracking relations between facts, and only in this sense can our intuitions and the reasons that represent them, be seen as rational. When we subsequently communicate these reasons to others in our attempts to justify and persuade, we are not concerned with getting things right: in this stage reasoners can indeed be considered to be engaged in ‘mere rhetoric’Footnote 6. The exceptions are those rare cases where a reasoner is engaging in explicit reflection. So when Einstein is attempting to derive a mathematical conclusion, he is consciously applying rational norms such as the rules of logic or probability theory. But when the gynaecologist from the introduction explains why she considers it best to do another test, she is merely trying to convey her intuitions to the patient in a maximally persuasive way.

This is why Mercier and Sperber start their book with a rejection of what they call the dogma that reason is ‘the faculty that makes humans knowledgeable and wise’ (2017, p.1). In their view, the activity of producing reasons and communicating them to others is for most people most of the time not geared towards getting things right. However, they also emphasize that the second face of reasoning (the way we evaluate reasons produced by others) is geared towards getting things right. Evaluative reasoning, being one of the mechanisms realizing epistemic vigilance, thus could thus be seen as a corrective mechanism that ensures that pure rhetoric doesn’t win the day. Taken together, Mercier and Sperber argue that our cognitive apparatus enables us firstly to produce reasons that are grounded in intuitions that are usually pretty sound, and secondly to take a genuinely critical stance towards the reasons produced by others.

4 No getting rid of normativity

Even though Mercier and Sperber’s interactionist framework has been of crucial importance in bringing about the much needed ‘social turn’ in our thinking about reasoning, I do not think they successfully show that producing reasons is not a normative activity, in the sense of not being ‘properly geared to the pursuit of knowledge and good decision’ (2017, p.180), as they put it. Contrary to what Mercier and Sperber argue, I hold that both in evaluating and in producing reasons, we must at least to some extent be concerned with getting things right. To develop my objection, it is first important to further clarify what it means to say that reasoning is a normative activity.

I propose that we are performing a normative activity if, in performing this activity, we are concerned with norms or standards pertaining to knowledge and good decision (for example logical or epistemic norms). The normativity at stake is, to use terminology Mercier and Sperber also adopt, a personal-level concept (Mercier & Sperber, 2011). An activity is person-level normative if the person, in performing the activity, is necessarily to some extent concerned with following the relevant norms or standards (the meaning of ‘necessarily’ here will be clarified below). Activities like breathing or walking are not person-level normative: we do not need to be concerned with norms in any sense in order to breathe nor walk. The notion of person-level normativity is thus stronger than what one could call process-level normativity, or the idea that many activities are realized by rule-based processes or mechanisms (a notion of normativity that might also apply to activities like breathing or walking). However, even if person-level normativity is a stronger notion, it does not require that reasoners must continuously be aware of the relevant norms, nor that they apply such norms by explicit deliberation. Playing chess for example is a person-level normative activity, where the relevant norms are the rules of chess. However, players do not continuously have these rules in mind, nor do they make conscious decisions to for example not move the King more than one square. Players are concerned with the rules of chess in the sense that in the way they play, they aim to ‘get things right’ with respect to these rules (Achourioti et al., 2011).

As Achourioti et al. argue, playing chess is person-level normative in a constitutive sense: the rules make it the game that it is, and someone who doesn’t care about these rules at all just isn’t playing chess but is doing something else. It is in this sense that chess players are necessarily concerned with the rules of chess. In a similar vein, my claim is that both in evaluating and in producing reasons we are necessarily concerned with certain norms pertaining to knoweledge and good decision: without any such concern, we would not be reasoning but doing something else. Of course this is not yet an account: in the next section I will argue in more detail why I take reasoning to be normative in this constitutive sense.

As described in Sect. 3, Mercier and Sperber claim that whereas evaluating the reasons of others is normative activity, producing one’s own reasons is not. In contrast, I will argue that both faces of reasoning involve normative activity, and that this follows even from Mercier and Sperber’s own conceptualization of reasoning as social activity. My main argument for this is as follows: even if in producing reasons we are geared towards justification and persuasion, this seems to require that in producing these reasons we are at least somewhat concerned with getting things right. After all, the audience is supposedly equipped with a critical faculty of epistemic vigilance, and is thus ‘demanding so as not to be deceived by poor or fallacious arguments into accepting false ideas, objective so as to be ready to revise our ideas when presented with good reasons why we should’ (2017, p332). So if Mercier and Sperber are right in claiming that reasoning is a social tool geared to persuasion and argumentation, the question is how our reasoning manages to achieve such aims. How do we manage to successfully overcome the epistemic vigilance of our audience by providing them with reasons? My claim is that in order to function as the social tool that it is taken to be, reasoners should be able to assess their reasons in terms of the same standards the audience will use to assess them. Importantly, the process-level normativity Mercier and Sperber ascribe to ‘well-adapted inferential modules’ clearly will not do the trick here: our intuitions might often be in correspondence with rational norms, but in Mercier and Sperber’s view, reasoners do not (except in rare cases of ‘professional’ reflection) critically assess their own intuitions in terms of such norms (this is their main point).

So, my objection is that in order to be the social tool Mercier and Sperber take it to be, producing reasons necessarily seems to involve a critical evaluation of reasons as convincing or unconvincing: thus, some kind of activity that is not merely often in line with certain norms or standards, but that we (on the personal-level) actually engage in in order to follow certain norms or standards.

Adherents of the interactionist account might object that understanding the production of reasons as a normative activity in this sense is actually not necessary at all: instead, they might argue, the mechanisms of epistemic vigilance ‘hone’ our reasoning skills from the outside. On such a view, reasoners will find out that some reasons are more readily accepted by their audience than others, and will keep adapting their strategies accordingly. On such an account, the normative sensitivity of reasoning is nothing more than having well-adapted intuitions (process-level normativity) sharpened by a critical audience. I think such a strategy fails, because sharpening one’s reasons on the basis of social feedback still requires a form of critical evaluation: in order to sharpen the reasons we provide, we need to understand why they are rejected by our audience and thus in what sense these reasons fail as means towards justification or persuasion. And this, I argue, precisely requires engaging in the kind of critical evaluation which, according to the interactionist theory, has no place in everyday reasoning.

So far, I have argued that Mercier and Sperber’s social account of producing reasons does not show that it is not a normative activity. In the remainder of the paper, I want to show that it is in fact perfectly feasible to understand both faces of reasoning in terms of activity which is both social and normative. More specifically, I will propose that Gilbert Ryle’s later work on reasoning offers such an understanding, and that as such his ideas could be of great value for the social turn in our thinking about reasoning. I will present a Rylean perspective as an alternative social account: it offers an account of reasoning that is clearly different from Mercier and Sperber’sFootnote 7, and also stands out from other social accounts such as those developed by Elqayam & Evans (2011) or Resnick and colleagues (1993), in putting normativity central stage. However, the Rylean alternative is still grounded in the same basic tenets shared by all social accounts: reasoning is primarily done with others, and its aims are social.

5 Reasoning as giving-and-taking-lessons

Although Gilbert Ryle is most known for the core ideas developed in his early book The concept of mindFootnote 8, in later years he also developed an understanding of reasoning as activity that is refreshingly relevant for our present concerns (Ryle, 1971, 1979, 1993)Footnote 9. In outlining his views on reasoning, my aim is not to argue that this is the only viable way to think about reasoning, but merely that Ryle’s conceptualization offers valuable leads on how to combine the idea that reasoning is social with the idea that reasoning is normative activity. Ryle’s general concern in his later work is a much broader question, namely the question what is thinking? The proposal he develops, is that there is not one answer to this question: thinking is a polymorphous concept (1971, p261). One specific (and pretty advanced) form of thinking is reasoning or, as he describes it, making inferences by ‘operating with propositions’ (1993, p.68).

Like Mercier and Sperber, Ryle argues that reasoning has often been put on a pedestal as a human superpower (1993, p66). He locates the origins of this mistake in classical Greek philosophy: the Greeks invented the skill of constructing systems of propositions, and because of the huge intellectual benefits this skill provided, they came to believe that this was what distinguished them from barbarians and non-human animals (1993, p67). To explain the mistake involved in doing so, Ryle asks us to imagine a society which invents the game of bridge, and is so proud of this invention that it comes to believe that the ability to play bridge is the essential capacity that distinguishes them from lower creatures. They come to define most other skills and activities in terms of playing bridge: ‘Even the Bridge-players have to look after their houses, gardens and families. But it has been found easy to say that these activities are simply Practical Bridge — paying the grocer is simply a special sort of Following Suit, and being kind to the children is simply a special sort of Supporting your Partner’ (1993, p69). Ryle argues that the same unfortunate thing has happened to our understanding of thinking: mistakenly, we have come to see all forms of human thinking as forms of reasoning (1971, p423).

So what, according to Ryle, is the actual role of reasoning within the wider concept of human thinking? He develops the idea that we produce reasons mostly to report on insights regarding what-follows-from-what. This means that making inferences, according to Ryle, is not the activity by which we find things out. Here he makes an interesting distinction between path-making (finding things out) and path-following (stating or explaining what we found out). Reasoning is mostly path-following activity: think of a detective writing up a report, or a scientist writing down his or her theory. Both are reporting on what they found out, which means that they must already, at least in vague outline, know what they found out: ‘far from it being true that ‘inference’ denotes an operation in which a discovery is made, an operation, therefore, which could not be repeated, we mean by ‘inference’ an operation which the thinker must be able to repeat’ (1949, p283).

Path-making on the other hand is a much broader and vaguer phenomenon. Like Mercier and Sperber, Ryle emphasizes that we often don’t know where our insights come from (1979, p65). And even though he does provide some interesting suggestions (Ryle, 1993) Ryle does not give a systematic account of the kinds of thinking that are involved in the making of paths. However, here our concern is with his account of reasoning. Reasoning, according to Ryle, must thus be understood as path-following activity or the reporting on inferences.

From a contemporary point of view, this raises the question: how can reasoning be a report on our insights regarding what-follows-from-what, given that we by now know that reasoning is often ‘just’ post-hoc rationalization (Haidt, 2001)? Hasn’t the psychology of reasoning shown us precisely that the reasons we provide are grounded in intuitions and not in intellectual insight? The crux here is that reasoning should not be seen as a report on cognitive processes. As Ryle’s account of reasoning as path-following shows, in reasoning we report on the outcome of a cognitive process. Think back to his analogy that reasoning is like writing up a scientific theory or a detective’s report: a theory is not a report on all the work, intuitions and coincidences that contributed to the scientist coming up with the theory. Similarly, a detective’s report is a report on who committed the murder and on the basis of what evidence this claim could hold in court, and not a report on all the detective’s activities that went into finding this out. In a similar manner, reasoning is a report on what-follows-from-what, not on the cognitive processes that contributed to us now making the inferences we make. This suggests that the structure of reasoning as activity is a normative structure, not a causal one: our insights on what follows-from-what express our adherence to certain norms. I will further elaborate on this point in the next section.

So what is path-following for—what is the aim of such reporting on inferences?

Ryle’s answer is, again much like Mercier and Sperber’s, that reasoning is a social tool: its aim is communication. However, Ryle has a very different interpretation of the kind of communication involved. According to Ryle reasoning is a didactic tool: in reasoning we are concerned with giving and taking lessons in the broadest sense of the term (1949, p.293). In providing reasons to our interlocutors, we give them reports on our insights regarding what-follows-from-what. In parallel, in evaluating our reasons our audience is interested in learning a lesson on what-follows-what. This can be illustrated by looking back at the examples from the beginning of the paper. On a Rylean interpretation, when I reason with my daughter, I try to teach her the lesson that tastes change over time and therefore that she should taste the bread. The gynaecologist teaches the patient a lesson about the existence of a certain risk, trying to show her that because of this risk it might be wise to come in for another test. The patient learns another lesson (a faulty one as it turns out later), namely that she is in great danger. In developing the idea of reasoning as didactic discourse, Ryle thus provides a wholly different interpretation of what ‘reason’ is: ‘Reason is not an internal source but the sediment of what one has been taught (the ‘voice of reason’ is oneself rehearsing one’s teachers’ teachings)’ (1949, p.297).

The general concept of a lesson does not specify in what sense the insights provided or acquired are supposedly valuable: lessons can be of all kinds, about all topics from physiology to coffee making to love. Ryle states that whereas in reasoning we always express our adherence to certain norms, the scope of possible norms is very wideFootnote 10. For example, one can reason ‘from’ norms of etiquette, for example when a parent says to a child ‘We don’t eat with our hands in this house, so use your fork!’. Here, the proposition ‘We don’t eat with our hands in this house’ is offered as a reason for the child to use his fork: this case of reasoning is governed by cultural norms of etiquette in that the second proposition only follows from the first in so far as one accepts the cultural norm. However, even in cases when reasoners refer to norms that do not hold universally or unconditionally, they bring forward such norms as relevant for knowledge and good decision. The norms that are at stake in a specific case of reasoning thus structure the reasoning in the sense that they show why a certain conclusion follows.

It is important to clarify the conceptual relation between reasoning on the one hand and ‘giving and taking lessons’ on the other. The point is not that giving and taking lessons is always reasoning: we also give and take lessons in many other ways (such as memorizing, doing exercises, or transmitting factual knowledge). So Ryle’s point is not that everything teachers and pupils for example do in a school, for example, comes down to reasoning. His claim is that in exchanging reasons we are giving and taking specific kinds of lessons, namely lessons on what-follows-from-what. However, this also doesn’t mean that reasoning is always like being in school. It only implies that in reasoning we are always and necessarily engaged with the communication of insights, even when we are reasoning for ourselves in an inner monologue. Importantly, this does not imply that in reasoning we are exclusively concerned with communicating insights: often we will also be guided by other considerations and motives (like the various biases and processes known from research on motivated reasoning, see for example Kunda 1990; Epley & Gilovich, 2016). However, the main claim defended by the Rylean perspective is that in so far as we are reasoning, we are always at least also concerned with the question what follows-from-what.

In the remainder of this section I will contrast Ryle’s conceptualization of reasoning with Mercier and Sperber’s interactionist account, and show how I think Ryle’s ideas could form the basis of an alternative social account of reasoning, which explicitly embraces the idea that reasoning is a normative activity in which we are geared to ‘getting things right’. Like Mercier and Sperber’s theory, the alternative social account I develop emphasizes that the aims of reasoning are social: we reason in order to facilitate social interaction. Also, it emphasizes that we mostly reason with others. True, we do sometimes reason with ourselves (as in the third example I gave in the introduction); in the Rylean framework this can take two different forms. Often solitary reasoning is ‘following one’s own paths’, in other words, silently telling oneself things one already knows (for example in preparing for a conversation). But in some cases we are also assembling propositions in our head in order to find things out (like Einstein working out a mathematical problem). However, we only know how to do this because we have experience with reasoning with others. Ryle gives the following description of a solitary reasoner: ‘now he experimentally applies to himself, just in case they may turn out to be effective, operations of types that are often or sometimes employed by live teachers upon live pupils’ (1979, p. 75).

A Rylean social account also shares with Mercier and Sperber’s the idea that in reasoning most people most of the time are not directly engaged with ‘the pursuit of knowledge and good decision’ (Mercier & Sperber, 2017, p180). In reasoning we are not communicating with the True and the Good: we are communicating with other people, and sometimes with ourselves. However, in contrast to Mercier and Sperber’s interactionist account, the Rylean alternative view does rely on the idea that reasoning as social activity is nevertheless person-level normative activity, thus in the sense that a reasoner is necessarily at least somewhat concerned with the pursuit of knowledge and good decision. Even though the aim of teaching and learning is communication of lessons, that aim can only be achieved in so far as one is at least somewhat interested in ‘getting things right’Footnote 11. To further explain why this is so will be the focus of the final section.

6 The normativity of reasoning as activity

In what sense is the activity of reasoning understood as lesson-giving-and-taking, constitutively normative? In this section I want to deepen the conceptual analysis, and spell out two normative assumptions I think are contained in the concept of a lesson. This is not an analysis Ryle himself developed, even though he did offer some suggestions pointing in a similar direction.

Firstly, the concept of a lesson necessarily involves the assumption that there is something supposedly worth learning. As Bakhurst (2020) argues in a recent essay about the nature of teaching: ‘The currency of teaching is knowledge, even though there can be counterfeits in circulation’ (2020, p. 306). So even if not everything given or taken as a lesson is necessarily worth learning, the conceptualization of reasoning as lesson-giving-and-taking, entails the assumption that it is an activity meant to convey insights that are genuinely valuable.Footnote 12 This clarifies in what sense reasoning is a normative activity in the sense outlined in section four: in reasoning we aim to get things right in the sense that the ‘currency’ (to use Bakhursts’ term) of reasoning is ‘insights worth having’. I believe that this gives a plausible social turn to the traditional idea that reasoning is “the faculty that makes humans knowledgeable and wise” (Mercier & Sperber, 2017, p.7). In contrast with more traditional interpretations, the Rylean perspective states that this is a communicative faculty that, even if it is indispensable for human functioning, doesn’t offer any guarantee for success. In other words, the account leaves ample room for either the provider or the receiver of reasons being misguided (the insight might actually not be valuable) and also for providers of reasons being deceptive or manipulative (and thus only pretending to be reasoning). Imagine a case where someone does not actually believe in the reasons they produce, but merely wants to manipulate the audience. From the Rylean perspective, such a trick can work precisely because by bringing forth reasons, the speaker suggests that there is genuine reasoning going on: that there is a valuable insight to be gained. Someone intent on manipulating his audience thus piggybacks on the ‘normal’ role of reasoning in communication: manipulation succeeds when the audience comes to think that the sender is right. Ryle himself notes that giving someone a lesson is fundamentally different from persuading someone (one of the basic notions used by Mercier & Sperber to characterize reasoning):

‘In contrast with the electioneer, the will-be lecturer, at least if he cares about his subjects and about his students, intends not to persuade them of anything, but to instruct them. The last thing that he wants is that his hearers should vote for his doctrine without having thought it through. He wants them to accept it for its merits, or even to doubt or reject it for its demerits’ (1971, p. 491).

Thus according to Ryle, whereas the activity of persuading only requires the ‘sender’ to maximize the chance that the ‘receiver’ will end up agreeing with him or her (by rational or non-rational means), the activity of lesson-giving requires more than that: in lesson-giving the ‘sender’ aims to maximize the chance that the ‘receiver’ will get things right, and this requires the sender to at least have some concern with getting things right.

So a reasoner must necessarily assume that there might be something to be learned in the exchange. This provides some normativity to the activity, but not much. In order to get a fuller picture of the normativity of reasoning as activity, I think we should acknowledge a second normative feature: the idea of giving and taking lessons entails that one must actually to some extent be interested in the communication of insights (Bakhurst, 2020). When I am just thinking ‘in my mind’ about how my children should improve their eating habits, this cannot count as a lesson because I am not even trying to bring it about that my children acquire my supposedly valuable insight. For the activity of reasoning, this imposes certain constraints: I am only reasoning with someone in so far as I am trying to actually communicate my insights: I should be concerned with expressing my insights in such a way that they can at least potentially be picked up by others. This implies that a reasoner must be ‘necessarily at least slightly concerned to think properly’ (Ryle, 1971, p.428)Footnote 13. The suggestion Ryle seems to offer here is that even though reasoning is governed by a wide variety of norms, reasoning is always also governed by certain basic inference rules (1971, p.431). After all, without at least some respect for something basic like the principle of non-contradiction it is difficult to convey any kind of insight. Again, this doesn’t mean that someone who happens to violate the law of non-contradiction is thereby not reasoning: there is clearly conceptual room for fallacious instances of reasoning. However, it does imply that the notion of a lesson brings with it a substantial sense in which reasoners should care about, for example, the law of non-contradiction: care for such norms is entailed by the concept of lesson-giving-and-taking. This thus shows how reasoning is a normative activity not only in the sense that it is an activity geared to communicating valuable insights, but also in the sense that the possibility of such communication depends on reasoners being concerned with certain basic norms of inference.

This makes clear that the Rylean position defended here is a form of what Elqayam & Evans (2011) refer to as normativism: the claim that our reasoning is grounded in a normative system such as logic or probability theory. Like Mercier and Sperber, Elqayam and Evans are suspicious of the idea that normative features would be needed to explain reasoning: as they argue, it is misguided to think that we reason in the way we ought to reason according to the rules of logic or probability: ‘theories of higher mental processing would be better off freed from normative considerations’ (2017, p233). This is in line with Mercier and Sperber’s suggestion that the activity of providing reasons is not guided by rational norms except on a minimal, instrumental sense of rationality: the sense that our reasoning processes help us to attain our (social) goals. However, the Rylean perspective developed here is very different from the type of normativism rejected by these authors. Firstly, on the didactic conceptualization, reasoning is not governed by one normative system such as logic or probability theory, but by a wide and varied array of normative considerations. Some of these norms (like inference rules) are indeed rational norms - but other kinds of norms (such as cultural norms) are not. This means that whereas the Rylean agrees with Mercier and Sperber’s starting point that reasoning is guided by other considerations than rationality, I disagree with their conclusion that in reasoning we are almost never concerned with any kind of norms, let alone with rational ones. On the Rylean conceptualization, when we are reasoning we are always and necessarily guided by normative considerations, some of which are rational.

However, the normativism defended here certainly does not state that people always reason in the way they ought to reason. In fact, Ryle clearly shares Mercier and Sperber’s assessment that humans are flawed reasoners:

“just as it would strike us as absurd to say that all men are far-sighted or all men are strong or loyal, it strikes us as absurd to say that all men are rational. All men cannot excel. [.] Moreover, for quite other reasons, we have lost optimism about the prevalence in the world and the influence of this special excellence of rationality. Apart from the fact that it cannot be deserved by all of us, the testimonial seems to be undeserved by any of us” (Ryle, 1971, p.41).

The fact that in order to evaluate or produce reasons you must have some regard for getting things right, leaves ample room for all kinds of mistakes. Beginning chess players usually have great concern for the rules, but it does nevertheless not surprise us when such players make frequent mistakes. Therefore, there is no reason to think that the claim about the constitutive normativity of reasoning as activity is in conflict with the huge literature on reasoning fallacies and biases Tversky & Kahneman 1983ab, Tversky and Kahneman 1983b; Stenning & Lambalgen, 2012>; Stanovich et al., 2013). Whereas the rules of chess are easy to learn, the norms guiding our reasoning are complex, often only implicitly taught and widely varied in nature. Therefore frequent mistakes are to be expected: one could say that as reasoners we are ‘perpetual beginners’.

However, one might think this picture is still too optimistic in that it wrongly suggests that reasoners are at least always trying hard to get things right. But as Mercier and Sperber rightly point out, there is a large body of empirical evidence showing that reasoners are often lazy, and also very much concerned with other aims than getting things right. This is addressed in the literature on motivated reasoning, which discusses how for example wishful thinking, motives relates to social status, or partisan goals affect our reasoning (Kunda, 1990; Taber & Lodge, 2006; Epley & Gilovich, 2016). Also, myside bias is a good example that is extensively discussed by Mercier and Sperber. As they point out, reasoners generally dislike changing their mind and resist doing so whenever possible: “reasoning [.] systematically works to find reasons for our ideas and against ideas we oppose. It always takes our side” (Mercier & Sperber, 2017, p.218). Here again, my response is that the Rylean approach does not in any sense dispute these important psychological insights. In reasoning we are necessarily concerned with getting things right, but this leaves ample room for the insight that we are most of the time also concerned with other aims, and driven by a multitude of other motives. Moreover, I hold that the literature on motivated reasoning might actually provide us with important psychological explanations for the observed differences between the two faces of reasoning that Mercier and Sperber emphasize. As they state: “whereas the interactionist approach predicts people to be biased and lazy in producing arguments, it also predicts that people are demanding and objective in the evaluation of the arguments of others’ (2017, p332). My point here is that the Rylean account of reasoning as social normative activity, would reach the same predictions, but not for conceptual but for psychological reasons. After all, psychologically one would indeed expect that motives such as wishful thinking or regard for social status are most likely to play a role when we are producing our own reasons, and less so when we evaluate the reasons of others. However, the Rylean point here is that such psychological differences in themselves do not give us reason to conceptualize the two faces of reasoning differently: it is possible to do complete justice to our psychological reality while maintaining that both in producing and in evaluating reasons a reasoner must be at least somewhat concerned with getting things right.

Obviously, much more work would need to be done in order to show how a conceptualization of reasoning as didactic discourse relates to the vast body of empirical findings about reasoning. What the Rylean perspective offers is a conceptualization: an alternative answer to the question what reasoning is. As such, it does not deliver its own psychological explanations of how and under what conditions biases, successful instances of reasoning and fallacies occur. However, I hope to have made clear that at least at first sight, the Rylean account does not seem to go against core psychological insights with regard to biases, fallacies and phenomena related to motivated reasoning. Hopefully, the conceptualization developed here offers fruitful leads for further discussion on how to integrate a normative understanding of reasoning with such psychological insights. For this paper, my aim has been to show that Ryle’s later work on thinking could be a highly valuable contribution to the social turn in our understanding of reasoning, by showing that reasoning could be seen as an activity which is both normative and social.