Reasons

Abstract

The temptation to look for the “purely normative essence” of argument stems from the understandable ambition to distinguish rational persuasion from mere persuasion. But in seeking a purely normative notion of argument it is easy to overlook—or actually deny—that rational persuasion is a kind of persuasion. The burden of this essay is to show that the concept of reason from which our interest in argument derives can only exist and have normative force as a kind of persuasion, that is, as something (also) causal.

1 Introduction

In certain important contexts the words ‘because,’ ‘so,’ ‘reason,’ and ‘consequence’ allow systematic interparaphrases of sentences in which they function as connectives. A because B will say the same thing as B so A, and either may be rephrased as B is the reason for A and A is a consequence of B. This cluster of paraphrases represents what might be called the “structural vocabulary” of two distinguishable activities: reasoning on the one hand and explaining of a sort that has been dubbed “broadly causal”¹ on the other. In any of these formulations B may be being offered as an argument for A or, alternatively, as a (broadly) causal explanation of A. In the idiom of reasons, the contrast is between reasons for A and reasons that A.

This grammatical distinction suggests, and most of us naturally suppose, that these are two different kinds of reason: that the structural notions are merely homologous, not identical. And it is worth spelling out in some detail the sense in which this is right. Reasoning and (causal) explaining are essentially distinct activities: the relation of “reason” alleged between A and B in one is of a different kind from that alleged in the other. This is most easily seen in the fact that we can give arguments for conclusions that may not have causal explanations. We may conclude that the planet Vulcan never existed, that the square root of 2 is an irrational number, or that a story is implausible, while not making any sense of them as causal explananda. But even when B is an ordinary event, something that would make a fine explanandum, the difference stands out. For when we offer B as an argument for A, we address the question whether A happened, recommending, more or less strongly, the answer ‘yes.’ By contrast, when we offer B as a (causal) explanation of A, we address the question why A happened by giving the answer ‘B’: such a question is not grammatically susceptible of the answer ‘yes’ (or ‘no’).

This shows up in the cause/indicator distinction: a gauge reading can tell us whether the boiler pressure is too high while not being the right sort of thing to address why. And its lack of explanatory credentials weighs not at all against its role as a reason in the argumentative sense. So even when a single item plays both roles—appeal to a termite tent can both explain why and argue for the conclusion that the neighbors are not at home—these are still clearly two different jobs for that single item to perform.

Philosophers sometimes capture this by distinguishing a normative use of ‘reason’ from a descriptive one. Arguments are justifications: they recommend B as a conclusion, underwrite its truth as a proposition. By contrast, when we explain B, say as an event, we simply describe what was involved in bringing it about. Whether it came about is a separate issue, something usually taken for granted. So again, even if a single item can do both jobs, they are clearly different jobs to do.

Nevertheless, a deep ambivalence about this distinction, about whether and how the two are connected, pervades the recent philosophical record. Attempts to simply assimilate one notion to the other are easy to find in different literatures, although there is no consensus about just how the reduction should go: all three possible “directions” are well represented. In the philosophy of science, one strand of the literature on scientific explanation takes causal explanation to simply be a kind of argument (the strand initiated by Hempel and Oppenheim (1948), but one with many subsequent ramifications as spelled out in Salmon (2006). On the other hand, part of the “analytic” literature on action and agency reverses this priority and takes argument (justification) to be a kind of explanation (see, e.g., Broome (2004) for illustration). And finally, a theme in the Informal Logic literature finds the two notions essentially identical, from a logical or conceptual point of view, differing if at all only in “extra-logical features” of the contexts in which they occur “...such as their functions, social purposes, normal contexts, and the beliefs, interests, intentions, and goals of the speaker...” (Thomas 1981, p 147) Very similar views may be found in Govier (1987), p. 173 and Walton (1997), p. 612.

Considered in isolation, each of these assimilations can seem wrongheaded or wholly unhelpful. The first has been subject to perhaps the most sustained and withering criticism that any analysis of a concept has ever received [summarized in Salmon (2006)]; the second concerns essentially normative conclusions, not propositions more broadly; while the third simply ignores the why/whether distinction, making no attempt whatever to address the apparent conflict. Despite these reservations, it seems unwise to ignore altogether the positive side of this ambivalence. That so many thoughtful philosophers have been tempted to find a deep connection between the two notions should give us pause in declaring it a simple ambiguity in the word ‘reason.’ Neither should the striking structural homologies be dismissed out of hand as an etymological accident.

2 The Relation Between Reasons and Causes

These preliminaries still allow some choice of starting point, and as always the best way into a problem can be determined only by trial and error. But the grammatical complication that drives the two notions of reason apart must be at the center of any attempt to bring them back together, to find insight in their interconnection. We began with the observation that the two kinds of reason address different questions: Whether and Why. An argument for A concerns whether A is the case, addressing the question “Is C true?”, which allows A to range over propositions in general. Whereas a (causal) explanation of A concerns why A is the case (addressing Why A?), which restricts A’s range to propositions that make sense as explananda. A moment’s reflection reveals that not only do these different sorts of reason address different questions, they do so in grammatically quite different ways. In responding to a Why? question, an explanation (explanans) actually answers that question. When it explains the fire, a short circuit answers the question Why did my house burn down? Similarly, a dead battery answers Why won’t the car start? By contrast, arguments, reasons, are of the wrong grammatical form to answer Whether questions. These have only two proper answers: yes and no. Arguments address Whether questions by recommending an answer, characteristically yes, not by being one.²

The value of framing the distinction in this way is that the strength of an argument appears in it as the strength of this recommendation. A good argument just is one that strongly recommends its conclusion, a weaker one will recommend it with less ardor, a bad one with little or none. So the normativity distinctive of this notion of reason is entirely captured by the “recommends” made explicit in this formulation. Since our reservations about assimilating the two notions rest largely on this interrogative disparity, it is promising to find the key feature distinguishing them showing up in the structure of its grammar: in the difference between simply answering a question and recommending or evaluating an answer. It turns out, as perhaps could not have been anticipated, that examining this complication, how recommending relates to simply answering, does reveal a systematic connection between the two notions. It also exposes a budget of issues that must be addressed to appreciate the deep ramifications of that connection.

2.1 Giving Reasons

So let us take a look at these recommendations. Where do they come from? Why do we care about them? Well, the ones that matter to us occur in our everyday practice of giving reasons; and we care about them because that practice plays a substantial role in our lives. We regularly ask for reasons, give them unsolicited, complain of their lack or about their quality, and weigh them in making up our minds. What we think and do is often determined by engaging in this practice. We have high expectations of it. This of course presumes that we, ordinary human beings, commonly have access to the resources required to evaluate its recommendations as well as the competence to exploit them. Criticizing an argument, rejecting a reason, would be a pointless waste of time without a good grip on what would count as a satisfactory one in the circumstances. If, as philosophers sometimes suggest, we seldom or never actually meet this condition, that would not be a criticism of our arguments so much as a repudiation of the value we place on institution of giving them: our high expectations would be groundless.

Sometimes this skepticism stems from a forgivable preoccupation with ambitious examples, ones that draw our attention precisely because they don’t go smoothly. The endless controversies over nuclear power and legalizing drugs, or even free will and determinism, do suggest that we lack full command of the resources required to deal with them. But contentious matters of culture and policy, or ones deep in the abstractions of philosophy, on which even the great thinkers of our tradition were not of a single mind, cannot be the paradigms that ground our interest in giving reasons (see Fogelin 1985 for much more on this). What we value in this practice, and in the concept of argument deriving from it, must be found in the plethora of mundane applications that do not stand out like this precisely because they are so effortlessly satisfactory. Exemplars such as these:

Proposition

The neighbors are away.

Reason

Yellowing newspapers have accumulated in their driveway.

Proposition

We should take the coast route.

Reason

It’s snowing in the mountains.

Proposition

We’ve had a power failure.

Reason

The digital clocks are flashing.

Proposition

The car’s out of gas.

Reason

The motor just quit and the fuel gauge reads empty.

Each of these might be schematized in canonical argument form: support on top, conclusion below. e.g.:

S: The car just stopped on its own.

The fuel gauge reads empty

------------------------------------

C: We are out of gas.

To see what it is we value in these appeals and to understand the resources we deploy in evaluating them, requires that we place them, or imagine placing them, in familiar contexts in which they would naturally arise. This is accomplished for the schematized argument by imagining my car drifting to an unscheduled stop on a drive to the supermarket. The flashing diodes might draw attention when I arrive home expecting a roast to be done; the newspapers come up naturally during a party when the noise reaches a level that might disturb the neighbors; appeal to snow becomes relevant in discussing the location of a breakfast stop on our trip north: the eatery I recommend being on a route that is longer and generally more disagreeable than the usual one. In each of these cases we would easily understand, and typically recognize, the myriad details required to make the support offered a good reason: reason, that is, to accept the proposition, assumption, or recommendation in question. And even before we try to say anything general about what we accomplish in exchanges of this sort, there is no doubt that they play a valuable role in human commerce.

The objectivity of what we evaluate here is manifest in our ability to recognize circumstances in which the reasons first mooted are less than satisfactory.³ In such cases we also normally have some idea how to pursue further relevant considerations if matters: that is, to know when the inadequacy has been addressed, and when to abandon the inquiry in spite of its persistence.⁴ Our ability to multiply examples like this more or less indefinitely testifies both to the utility of the practice of giving reasons like this and to the range of our competence in it.

It is thus relevant to note that philosophical attempts to model the goodness of reasons, deductively or inductively, have failed to provide a standard that can be applied by ordinary humans in contexts like these: the ones giving rise to our interest in and high expectations of the normative notion of reason. In brief the judgment of the literatures that have concerned themselves with this issue has been (a) that few of our genuinely good appeals to reasons are (or can plausibly be reconstructed as) deductive,⁵ and (b) that even if the paradoxes generated by the induction literature may be set aside, the formal discussion of probabilities is rarely helpful in the substantive cases we wish to comprehend.⁶

This failure has had a melancholy effect on the one literature most directly concerned with practice, and from which we should expect clearest insight into the source of our interest and expectations. Whole swaths of Informal Logic have abandoned the philosophic project altogether. One reaction there has been to deny that philosophy has anything useful to say about substantive reasoning, and that our instruction in such matters must rest with the various disciplines.⁷ What reasoning looks like in physics must be learned from the physicists, that in the kitchen from cooks. A more popular response has been to abandon altogether the normative picture of argument and look instead to certain patterns in conversation:⁸ a model would then consist in rules governing moves in a forensic language-game subserving dialectical goals, most notably agreement.

Although both of these radical reactions contain genuine insight,⁹ and each displays a healthy desire to distance itself from epistemic skepticism, they abandon the traditional questions of justification without taking a serious look at what practice actually does accomplish, and whether that achievement might contain resources adequate to the philosophical project. Adequate, that is, to setting out a general account of argument that accommodates both its normativity and our actual deployment of it. And when one examines with this ambition the paradigms of practice—examples of giving reasons in which they evidently achieve whatever it is we value in them—we do find resources to exploit in that project, although their inventory requires abandoning some preconceptions about what they might include. It also reveals that an intimate interconnection between this notion of reason and the causal one is essential to the work it does for us. So a closer look at those paradigms should also give us a better grip on the temptation to run the two notions together.

2.2 Conditions for the Possibility of a Reason

To see how reasoning practice can be of any value at all, however, we must surface an important insight in the pragma-dialecticians’ preoccupation with conversation. In fleshing out the contexts required to make sense of our paradigms, we naturally involve another person to whom we give the reason. It is easy to dismiss this as a dispensable convenience; but the dialectical occasion that prompts us to offer a reason is actually an indispensable logical feature of the concept of argument implicit in this practice. That becomes apparent if we look more closely at the conditions reasons must deal with in order to be actually given. Without an interlocutor, for instance, were my car to sputter to a stop on the way to the store, I might not, in the event, say anything at all, much less give a reason for a diagnosis. I might simply dial triple A, or a friend, or get out and start walking to a gas station. And even if I mumble to myself as I did any of this, we do not yet have enough context to make sense of the mumble as a reason for anything, much less for one of my current convictions.

Talk of the gauge reading and unrequested nature of the stop are made apposite by the concern of an interlocutor, say, seeking reassurance on the need for a walk.¹⁰ Similarly, I mention the newspapers only when a guest worries that our noise will disturb the neighbors. This is essential to the value of practice because I inevitably have at the time—at any time—literally countless convictions about my circumstance that are formally susceptible of explicit support. As I coast to a stop I am sure the fuel will not replenish itself on its own, that the engine won’t run unless I refuel, that the gas station will have gas if I walk to it, that it won’t have moved since I last drove past, that I can walk that far, that to do so I have to open the car’s door, that I can do that by tugging on a particular handle, that I can manage the tugging and will know what to do to get out once the door is unlatched, that I will find my way back to the car afterward, that I won’t starve to death on the trip, and so on. An interlocutor’s intervention selects from this long list a proposition for my reasoned attention.

But the constraint is actually more severe than this suggests. What we understand and do are not intrinsically propositional: we don’t really start with a list. We may break what we think and do into propositions for particular purposes, but this may be done in indefinitely many ways for any given chunk of life. An evening at a baseball game yields hits and runs and errors, double plays and ground outs, pitches good and bad, inspired plays and mere competence. Propositions correspond to each of these, which, with average attention, I reasonably accept as a result of my observations. But each one of them may be resolved into further propositions I reasonably accept also as a result of simply following the course of the game: things I think over and above the coarse characterizations.

One out, I observed, resulted from the team’s best batter taking a called strike three with a man on third. He was fooled by the pitch. He was looking for something inside. The pitcher cleverly set him up with an unusual sequence of pitches. He was chagrinned at letting himself be fooled in this way. The strikeout changed the course of the game. It clearly demoralized lesser batters on the team. And the crowd was slightly stunned too. I think all of these things and more about a scintillation, and I am sure I have good reason to think them all. But were I to actually stop and set out reasons for each of those propositions, and others like them from that single episode, it would use up my life. The batter in question is overweight for an athlete. Fans and the press hassle him about it when he’s not doing well. This may be running through his head as he strolls back to the dugout. It might explain his pace. Others in the stadium surely think these thoughts too. Some with sympathy, others, aching for the team’s prospects, indignantly. And so on.

Furthermore, the activities of choosing conclusions to support, and support to offer, also generate propositions that may be the conclusions of arguments. I think I should provide support for this conclusion rather than that, or this before that, or that the order doesn’t matter, or that it does matter, but only in certain respects. Further, I think this is the right sort of thing to adduce in support here, but it would not be were the occasion to change. And of course some of this organizes itself into a regress. Because deciding to think about what to offer reasons for generates another level of propositions to support: I think I should stop now and think about what to give reasons for, rather than earlier or later, and so on.

This is doubtless part of what Quine had in mind in insisting that we do not confront the world one proposition at a time. We take things in (the activity of the game) and decide to do things based on our settled convictions (walk to the filling station, consider deliberating) in a continuous sweep that is variously and endlessly articulable. We cannot conceivably make explicit arguments for even a small fraction of our distinguishable beliefs, even as we accumulate them. And this does not remotely mean that most—or even any—of what we think is unreasonable. It is simply to point out the huge difference between what we have reason to think and what we can give reasons for thinking.

But it does mean that whatever it is we value in the reasons we give depends on their being called upon to underwrite a proposition only rarely, with a frequency almost vanishing relative to its formal possibility. The reasonability of most of what we think must go without saying if explicit argument is to play a useful role in our lives. We belabor this point because it is immensely consequential. This “condition for the possibility” of our reasons is, as Kant would have it, in part constitutive of them. An important part of the concept of argument is to be found in the characteristic that distinguishes, in the welter of propositions we might reflect on, conclusions profitably given explicit support.

This is easy to miss because, in our paradigms, the device that privileges a proposition in this welter as worthy of support is so mundane: someone expresses a doubt about something I think. In each example my interlocutor does not share my conviction about something, at least in degree, perhaps not at all, and this is what they express in requesting a reason. I am sure, my interlocutor is not, that we’re out of fuel, the neighbors are away, and the rest. And it is this that locates the proposition to support: the conclusion of the argument, what goes below the line in its schematized form.

Its greater importance, however, lies in the way the particular doubt being expressed on such an occasion determines the rest of the argument: the reason that goes above the line. For even when someone’s request does manage to privilege a particular proposition as requiring support, the doubt about it that needs to be addressed can be one of any number. This is a second continuum our reasoning confronts. My diagnosis of fuel exhaustion, for example, depends on a vast and complex understanding of how the world works, about which countless particular doubts might arise. These may be articulated in distinct possibilities I reasonably ignore in addressing the doubt I actually do confront in the imagined context. I ignore all manner of weird coincidences (fuel filter plugged as the gauge expired), practical jokes (friends rigged up a timer to cut off the fuel and disable the gauge), condescending cars (designed to quit periodically to encourage exercise), all the way to extraterrestrial plots. Every facet of our grasp of matter and motion, mechanisms and liquids, even the motivation and talents of creatures and gods is relevant to the diagnosis,¹¹ and each is the locus of possible doubts to address by argument.

Unless something confines a doubt about a proposition to one small part of this spectrum of relevant considerations, all I could say in its defense is that it fits congenially into the intricate congeries of interconnections relevant to it. Every request would elicit the same response. The practice of giving reasons could serve no purpose—our familiar notion of a reason would simply not exist—were this all we could offer.

This is Wittgenstein’s point, early in On Certainty (Wittgenstein 1969), in discussing the sort of proposition Moore had insisted that he (and implicitly, everybody) knows without question. In our terms, Moore’s propositions are examples of those we would undoubtedly want to call reasonable, well-grounded. I have “all the reason in the world” to think I was at one time a baby and grew gradually into the adult I now am; that I have spent my entire life on or near the surface of the earth; that the mountain range I see from my window has been there a long time, certainly longer than I have been alive, and so on. But, Wittgenstein observes, if I actually tried to set out reasons in support of these propositions, I would not know where to begin. I could of course offer something that has the form of a reason, such as that I have pictures of myself as a baby, or that there are recorded accounts of people traversing those mountains going back centuries. But these would be, even collectively, no better grounded than my simple confidence in the stability of mountain ranges, and in any case they represent only be a tiny fraction of what I know to be relevant. The least misleading thing to actually say in their defense would be a version of Wittgenstein’s offering in § 89: “Everything speaks for, and nothing against [them].” Or: they fit congenially with everything else I know that is relevant. Again, this is not to give a reason, it is one way to reject the request for one.

Moore’s propositions do not differ from those for which we ordinarily do give reasons in being grounded in an inexhaustible web of interconnections. What distinguishes them is the obdurate way they resist attempts to find a context in which the question of their reasonability might arise, that is, a context in which a doubt about them is specific enough to be assuaged by a reason. They were chosen by Moore precisely because they are so deeply part of our general grip on experience that such an occasion seems beyond imagining, certainly beyond anything that is part of our ordinary commerce with each other, within which our competence with reasons is lodged.

This of course is the contrast. In cases like our exemplars, we not only can imagine such occasions, they are a familiar part of our conversational lives. These are the occasions already examined on which the concern of an interlocutor isolates a particular proposition for support by a reason. And a reason is adequate in these contexts because they do, also involve a very specific doubt about that proposition, one that may be adequately addressed by a tiny slice of the relevance spectrum. The reassurance sought by my passenger in the silent car does not concern every aspect of my understanding of the world relevant to the diagnosis of fuel exhaustion. Mentioning most of it would be not just boring but also condescending: we both know better. The question is: how do I know what particular doubt to address?

2.3 The Connection

We have approached this point with such hyperbolic diffidence because we must now confront directly that feature of practice most responsible for the conflations and confusions we have been trying to sort out. For the standard way for my interlocutor to express her doubt about C, and the one that allows me to grasp it in all its particularity, is for her simply to ask me why I think it. Why do you think we’re out of gas? Why do you think we should take the coast route? Why do you think the neighbors are away? This is usually all I need to respond with a reason: the silence and gauge reading; the snow in the mountains, and the rest. This brings two things into focus. First, this same interrogative may be used -- typically is used -- indifferently, no matter what particular doubt someone has about C.¹² So if knowing the right doubt to address is what allows me (any participant in this practice) to pick the proper item from the clamor of relevance, then that doubt must be available elsewhere in the context.

Second, this is quite explicitly a Why? question, not a Whether question. Yet in these contexts it obviously requests a reason, an argument, something essentially normative. It has thus been tempting in our tradition to think that this question is not really causal in spite of its surface. But this is a temptation we must resist if we are to understand the concept of a reason, the concept of argument that can have any value for us. For although this question requests something essentially normative (an argument) its having an answer at all requires that it be also causal. For in practice, what allows me to pick an item to give from the continuum of possibilities is the causal role it played in my coming to think C. Of all the things that could, logically, make a difference to the reasonability of my thinking C in these exemplars, one item (or a short list of them) actually did play an immediate causal role in my new conviction: made the difference to my thinking it. And it is this that the hybrid question requests.

A common variation makes this hard to miss: my interlocutor might well have asked, What makes you think C? What makes you think we’re out of gas? What my passenger in the silent car wants to know is what persuaded me, and this inter alia requests something causal, the item that led me to think.... Furthermore, this is what I do provide in response. The flashing diodes are what led me to think—what occasioned my judgment—that the power had failed; the snow caused me to alter the itinerary. The silence and gauge-reading were literally what made me think the fuel was exhausted: they explain my conviction. This causal role is all that distinguishes the reason I give from the welter of considerations that are just as crucial to the truth of my judgment as they are; and it is, furthermore, just what my interlocutor has explicitly requested. In practice, this completes the argument: tells us what goes above the line in the schematic form.

But if this is right, you may wonder what happened to the particular doubts we began with. Weren’t they supposed to do the sorting? It makes some sense to say that different contexts require different reasons because they raise distinct doubts to address. But if what actually selects the reason we give is its causal role in producing conviction doesn’t that run roughshod over the why/whether distinction? Isn’t that the wrong kind of answer?

This is actually a good example of the way our natural analytic vocabulary can generate philosophical puzzles in its struggle to articulate an abstract insight. For in rightly noting that doubts can be distinct, it too hastily implies that we have (or should have) a dedicated doubt-vocabulary. But although different doubts are always at work in selecting different reasons, they are not, and need not be, made explicit in this way. The doubt at work is always a “middle term” that drops out in the exchange: it was the doubt removed by the item responsible for my altered conviction.¹³ In the run of such exchanges, our only access to a doubt is through the reason that assuaged it; and the particularity of the doubt is measured by the distinctness of that reason in the continuum of possibilities.¹⁴

This is of course a very special context, because we often, perhaps usually, have no clear sense of exactly what changed our minds about something. But it is in the special context in which we do that the institution of giving reasons grows up. And it grows up here partly because I can so often take it for granted that my interlocutor’s grasp of local circumstance is enough like mine that her doubt about C may be addressed by the same consideration that relieved me of my own. The nearly universal use of second-person form to request a reason—Why do you think C?—testifies to the commonness of this presumption. And we care about the reasons this form elicits only because we are so frequently right in making it.

Giving reasons is thus possible in part because the occasions on which a doubt is expressed are explanatory occasions in this sense. A reason can recommend an answer to a Whether question, provide support for a conclusion, only by actually answering a quite specific Why? question.¹⁵ The possibility of giving reasons in support of our convictions depends on the ability of an item distinguished from the welter of relevance by its role in the second of these to also and thereby address the first. Of course, connecting the two notions of reason in this way inevitably involves us in some of the more enduring controversies of our tradition,¹⁶ and this may stand in the way of properly appreciating that intersection. But we are now in a position to see how the practice responsible for our interest in reasons negotiates those controversies and how the concept of argument deriving from that practice can be untroubled by them.

2.4 The Ramifications of Practice

The deepest and most pressing issue is of course normativity: how does a cause manage to be a justification? How is it that just knowing what changed my mind can count as underwriting that change? We might be tempted by the trajectory of our discussion to respond that the cause is selected from the relevance continuum and so is born normative: its credentials are built into its provenance. But that continuum is not a datum: our grip on it is only through our competence in the practice of giving reasons. It is generated by—and its significance is exhausted in¹⁷—our sense that there is no limit to the list of items which, had they been different, would affect the reasonability of a proposition. So the normativity of the spectrum rests at bottom on our competence in judging individual reasons. And about this we can say something helpful and revealing.

Practice achieves normativity quite straightforwardly by screening antecedents: our competence consists in recognizing that not just anything that causes me to think C will also underwrite it. The request for a reason would not be satisfied with stories that appealed to ancient toilet training, a trick of optics, or a recent bonk on the head, for instance. The practice of giving reasons exercises our ability to distinguish good and bad ways to be affected. So when I offer the silence and gauge reading as what made me think the fuel has been exhausted, I am alleging not only that they caused me to think this, but that this is a good way for me to have arrived at this thought. In doing so I claim, inter alia, an ability to distinguish the present circumstances from others in which noting these facts would not be a good way to come to think this,¹⁸ and that that competence was (causally) involved in my conviction.¹⁹ This is a paradigm of reasons-giving in part because we are so familiar with situations in which these claims would be boringly uncontroversial and in which we would easily recognize myriad flags that our confidence was misplaced.

Even given all this, however, such an exchange is successful only if the normative judgment is one that my interlocutor can also make. This is something presupposed in the second-person form of the typical request. For though she asks “Why do you think C?”, I understand her interest is primarily not in an effect on me, but one on herself: she is wondering whether what made me think C might do the same for her, and in a way that she can judge to underwrite the proposition. This observation has misled many students of argument into thinking that all the second-person and causal features of the paradigms are an irrelevant distraction and the question my interlocutor really wants an answer to is always Why should I think C? and that this is a purely normative query.

This suggestion has been so destructive because it is not entirely wrong and what is wrong about it is the subtlety we have spent so much time teasing out above. So we may gain a usefully different perspective on that subtlety by noting what happens when we take the suggestion seriously. As a matter of diction, we could always request a reason in this form: Why should I think C? It can succeed, however, that is, yield a giveable reason, only when it is hedged about with enough context to isolate a tractably finite doubt about C, and in a way that allows an interlocutor to identify it. And since, as we have seen, the only access we have to those doubts is through the reasons that assuage them, what practice requires is a way to identify in advance items that will remove those doubts in the proper way. In the abstract, this is a daunting challenge. So it is no surprise that practice begins with, and is anchored most securely in, cases in which interlocutors recognize that they share a comprehensive background relevant to C and in which one of them has just had a doubt about it removed by learning²⁰ something that may be captured propositionally. For such a context contains, perhaps like no other, resources capable of meeting this extremely difficult precondition. It also makes clear why the second-person, and explicitly causal, form of the query is canonical: it’s the natural request for the effective item in these contexts.²¹

The temptation to look for the “purely normative essence” of argument stems from the understandable ambition to distinguish rational persuasion from mere persuasion. Much of the Critical Thinking movement has been devoted to therapeutically identifying “fallacious” reasoning: bad ways to become persuaded of something. But in seeking a purely normative notion of argument it is easy to overlook—or actually deny—that rational persuasion is a kind of persuasion. Rational considerations are good ways to come to think something in contrast to bad or fallacious ones. The burden of this essay has been to show that the concept of reason from which our interest in argument derives can only exist and have normative force as a kind of persuasion, that is, as something (also) causal. Its causal role is all that distinguishes a particular item from the relevance spectrum as something that may be given; and a sharable competence at distinguishing good and bad ways to come to think something is the source of its normative force.

If any mystery remains about either the nature of the competence this implicates or its function in our lives it may perhaps be allayed by noting that it is already at work in the way our everyday conversational interactions constantly facilitate our negotiations with each other and with the world. We routinely draw each other’s attention to items in our environment that are responsible for our thinking something, confident that it will both produce and underwrite the same conviction in them.²² Some are directly perceptual (look, the Kim’s house is on fire), others more inferential, (oh there’s the bell, we are going to be late; look at those newspapers, the Robinsons must be on vacation).

Exchanges like this become routine of course only (when and) because in simply growing up in a conversing community we all naturally acquire just that understanding and competence required for the flourishing of reasoning practice. That is, they require that we sometimes know just what instilled a thought, recognize it as a good way to come to think it, and know that being made aware of it would engage an interlocutor’s competence in the same way it did our own. So it is only a very short step from the inferential instances of this routine to the indirection of the reasoning practice: when my mind changes in a consequential way without explicit citation. An interlocutor may then inquire about the provenance of my new conviction, hoping to assimilate the case to the more general practice.

This assimilation may make a bit less mysterious (though only that) the extreme selectivity of our appeal to reasons, compared to the continuum of thoughts we have good reason to think. For in our quotidian transactions, we simply accept most of what people think, as revealed in what they say and do, and, for the rest, we get on perfectly well keeping the bulk of our reservations to ourselves. It will require a longer essay to display the deep and subtle role this possibility plays in giving significance to what we do say in these exchanges. We mention it here only to allow the following observation. The relatively rare conditions of our exemplars are distinguished from the run of our interactions in two ways: I ask for a reason when something in the context makes it worth addressing my reservations about someone’s conviction that I do not share²³ and I have reason to suspect that their holding it is traceable to an item they can provide and that would properly engage my competence.²⁴ As a statistical matter it is not clear how often this latter suspicion pans out; but it does so often enough to make the institution of providing such items worth philosophical attention: it makes clear why we should care about the concept of argument.

2.5 Derivative Applications

The paradigms anchoring a practice like this typically lie at the center of a cluster of occasions that are continuous with them in functional ways. So even when a competence is shared, for instance, it may bear differently on a proposition depending on accidents of personal history. Thus I can sometimes give someone a reason—good reason—to think C even when I have no idea howIcame to think it, or how to characterize that history if I do. I might produce a map to argue that the Bosporus is in Turkey or an alibi as reason to think I’m innocent of a crime. But neither the map nor the alibi account for my thinking these things. The etiology of my thinking them may be unreconstructably lost in the mists of the past.

Such cases are continuous with the paradigms because even there, when Why do you think C? to requests an argument, my interlocutor seeks, and will only be satisfied with, a good way for her to come to think C. It’s just that in those exemplars, this would be the same as the way the person being asked (already) came to think it: the reason is whatever made them think C. So even when there is no useful answer to the question concerning how I came to think C, I can sometimes proceed as before and not notice the difference. For I often do know of something that would constitute a good way for me to come to think C if I did not already, and I know enough about my interlocutor to know that it would complement her competence in the same way it would have mine. I can then draw her attention to it and it would have the same credentials as does a reason in the simpler case.

This could be what actually happened in the noisy party exchange. I may have known about the Robinson’s trip to Bermuda for some time, have been feeding their goldfish, etc.; but had I not, I certainly could have reasonably inferred their decamping from the accumulating newspapers, given modest familiarity with local patterns. So simply recognizing my interlocutor to be in that position—sharing that aspect of my local competence but ignorant of the Bermuda antecedents—the pile of newsprint would work here in the same way it did in the simpler case. It would provide a good way for her to come to think they are not home. Similarly for a map of Eurasia and my alibi.

As we move further from the conditions anchoring our paradigms, continuity becomes less and less robust, but with care is still adequate to generate further variations on its pattern that we naturally exploit to expand the utility of practice. This pragmatism is a hazard for analysis, however, because the effortlessness of the central cases obviates the need to develop much descriptive vocabulary for them, in contrast to the more distant derivatives whose continuity with the anchors sometimes calls for direct attention. Our resulting ability to say more about cases as they depart from those anchors can make them appear to philosophical reflection to be better models than the paradigms themselves. And this will generate puzzles that can even seem to subvert the practice, producing a familiar skepticism. With that in mind, let us examine some further variations that thrive on especially congenial circumstances encountered as we move further from the core of practice.

First, there is at least one straightforward way in which someone can answer Why do you think C? directly, about themselves, even when C is a long held conviction of unclear provenance. But as we should now expect, it works because the familiar explanatory pattern can be marshalled in a different way. It occurs when the conviction is confronted with a specific objection, call it “S₁.” Someone may ask me why I (continue to) think the neighbors are away in spite of the sudden illumination in an upstairs room (S₁)? My long-held conviction may then be buttressed by a reason in the usual fashion when an articulable item²⁵ will explain why S₁ is not a good way to come to think not-C (to stop thinking C) in the circumstances. In this case the reason could be that the light in that room is on a timer: it comes on at the same time every night. Given this, S₁ is not a good way to come to think the Robinsons have returned. A particular doubt has been identified and dealt with in the standard way, the difference being that the doubt was about a settled perception that did not change as a result of the reasoning. The lack of change was the effect of the reason: what the reason underwrote by explaining.

A second variation is more complex. Although simply taking someone’s word is the contrast that defines this practice, there is a hybrid case worth noting. Sometimes we may give a reason to think C not by addressing the substance of C, but by giving the credentials of the source of C, that is, by giving reason to take their word for it. Here we treat another agent as we would a gauge or meter, and accept their dispensation even though we don’t know how they “work.” This can give me access to propositions beyond my understanding: ones for which no giveable item would supplement it adequately to allow me to judge C on my own. Perhaps some training would bring me up to speed, but it may just be beyond my competence entirely. In this way a mechanic’s reputation or track record can be an argument for his diagnosis—can be a good way to come to think the motor has a bad valve; just as an autistic savant’s success rate can be good reason to think the dancer just took exactly 879 steps.

It is precisely this explicit appeal, however, that makes these cases derivative outliers of practice rather than paradigms. The track-records and reputations we deploy here are topic-specific: it does not impugn a mechanic’s judgment about valves to note that the he is no good with dance steps. Such a record is always a ratio between successes and total cases of a certain kind. And for the run of things people think, and might give reasons for, we would have no idea how to determine its kind in this sense. The other cases it should be grouped with in order to make such a ratio would be simply indeterminate. That this is not a lack we can ordinarily make up by collecting more data raises the question of the default status of candor in the possibility of communication, something beyond the scope of this essay.²⁶ For now, it is enough to note the epistemic peculiarity of these cases.

Another constraint that may be relaxed in the proper context is the dialectical immediacy of the paradigms. One way to put the role of explanation in those (paradigmatic) cases is that a reason is always a reason for somebody. But as a conceptual matter, this simply points out the need for an interrogative context: the details of an inquiring understanding are required to sort through the continuum of relevance. But having recognized this, we may allow that those details can do this job indirectly. We thus may, and occasionally do, offer some item S_n as “the (objective) reason to think C”, without having any particular person in mind; and we can offer S_n to an audience for whom it is not currently compelling: not a recognizably good way to come to think C. The significance of calling S_n the reason to think C can be to say that the competence for which S_nis compelling is accessible to the intended audience. I can appreciate S_n as the reason to think C, even if I do not know enough at the moment to judge its force, because I take it that I (anybody in this conversation) could easily be brought up to speed if it mattered. We may in this way speak objectively of the reason to think Stonehenge has astronomical significance, or that Bruno Hauptmann was innocent. This still implicates a very particular level of skill, just not one that must be present in an interlocutor.²⁷

This possibility is easily misunderstood, however, and in a way that obscures its continuity with the practice we care about. For that continuity presumes the audience can accumulate whatever is required to evaluate S_n without already settling the question of C in the process. This possibility is part of the significance of picking out an item, something manageably finite, as the reason to think C.²⁸ Imagining such tuition is fairly unproblematic in everyday cases like Hauptmann’s innocence, as well as in more specifically disciplinary contexts in which the speaker has travelled the suggested path herself. This history is what licenses saying things like the endothermic nature of fusion at atomic numbers above iron and nickel is the reason to think those heavier elements in the earth’s crust are the product of stellar explosions. Making sense of such a claim may even organize a curriculum (as such a path). But Wittgenstein’s issue with Moore shows the conceptual importance of having such a path worked out in advance. For we can scarcely imagine a context in which acquiring the sophistication needed to participate in the practice of giving reasons might leave his propositions open questions at all, much less ones that may be neatly addressed by a propositional item. And without this, selecting such an item from the welter cannot give it the significance of a reason.

As a practical matter, this point is directly relevant when we are tempted, as we sometimes are, to casually offer a good or striking consequence of C as an argument for it, as the (objective) reason to think it, when C is part of our common understanding of things. We do this for instance when we appeal to the plummeting apple as an argument for gravity or to the shape of the earth’s shadow on the moon as reason to think the earth is a sphere. But these things can be genuine reasons at all only for an understanding that (a) does not already include the conclusion in question and (b) for which its having that consequence constitutes a recognizably good way to incorporate it. And in cases like this, such an understanding would be so distant from our own that to imagine “bringing it up to speed” at all, much less in a way that leaves C unsettled, and, further, unsettled in a way that allows just this consequence to be a good way to adopt it, is well beyond anything we’re prepared for, or even competent to do, in ordinary contexts. It may be possible for someone’s understanding of the local universe to develop far enough to make the moon’s darkening relevant to the earth’s shape and in the process not settling the issue of sphericity. But being able to imagine this is certainly not part of our casual appeal to such consequences; that appeal is thus not even a derivative part of the practice²⁹ underwriting our high expectations of argument.

3 Conclusion

If the temptation to assimilate the normative and causal notions of a reason is traceable to the structural homologies we began with, we can now better understand the difficulty our tradition has had in deciding just how that assimilation should go. For those homologies do reflect a deep connection between the two notions, but one that is tricky to appreciate. It is relatively uncontentious to say that in providing support for a conclusion, C, an argument, S, always answers a question of the form Why should I think C? with the proper pronoun and referent depending on contextual detail. But our exploration has disclosed, more contentiously, that answers to this question are recognizable as arguments for C, only when the ‘why’ is given a causal interpretation. In the practice within which the normative notion of reason derives its significance, an argument is always a particular kind of causal explanation.

To insist on this inevitably entangles the discussion of reasons in knotty issues that have divided great minds for millennia, issues which a short essay cannot begin to address. This is disappointing, however, only if we can separate our interest in reasons from the practice from which our expectations of them derive. For by connecting our civilian notion of a reason in just this way to neighboring denizens of the conceptual landscape, such explanation and proposition, competence and purposiveness, it can guide a novel attack on those famously difficult issues and one that may have an exciting payoff.³⁰ It should also aid in the development of an argument model that is adequate to the complexity of everyday reasoning.³¹