A Case for a Reasons-Based Theory of Argument

A basic and undeniable intuition is that argumentation is about giving reasons. This is recognized, for example, when it is stated that the object of study of argumentation theory is argumentative practices, which consist, in whole or in part, but, at least, to a significant extent, of asking for, giving and examining reasons. But this consensus does not translate into theory. Reasons occupy a modest place within argumentation theory, and especially in logical approaches. This fact may be obscured because “reason” is often used as a synonym for “premise”, although they are different things. Here is an example.

“Marijuana should not be legalized. That’s because sustained use of marijuana worsens a person’s memory, and nothing that adversely affects one’s mental abilities should be legalized.” In this argument, a claim is made that marijuana should not be legalized; that is the conclusion of the argument. And reasons for the claim are put forward; those are the premises of the argument. (Govier 2010, p.1).

In Trudy Govier’s example there are two premises or, depending on the model of argument chosen, a premise and a warrant, but not two reasons, if “A single reason is the smallest amount of information that by itself lends some measure of credence to a position” (Blair 2012, p. 148). Neither continued use of marijuana worsens a person’s memory nor nothing that adversely affects one’s mental abilities should be legalized individually lend any measure of credence to the position that marijuana should not be legalized. So neither of them works as a reason here. On the other hand, a premise is a statement that, by itself or in combination with others, expresses a reason. Given that the conjunction of “continued use of marijuana worsens a person’s memory” and “nothing that adversely affects one’s mental abilities should be legalized” supports the position that marijuana should not be legalized, both “continued use of marijuana worsens a person’s memory” and “nothing that adversely affects one’s mental abilities should be legalized” function here as premises (endorsing the premises-conclusion model and ignoring the complexities of the Toulmin model).

Francisca Snoeck Henkemans emphasizes the importance of not confusing premises and reasons when she writes:

The distinction between coordinative and multiple argumentation is therefore not that coordinative argumentation describes the relations between premises within one single argument and that multiple argumentation consists of a combination of single arguments, but that the relations between the single arguments that constitute these two types of complex argument are different.” (2000, p.460).

Connectors such as “in addition” or constructions such as “in the first place… in the second place…” provide a sort of linguistic test to distinguish between premises and reasons, because they introduce different reasons, not different premises within a single argument. So, the following sentence sounds weird: “Marijuana should not be legalized. That’s because, in the first place, sustained use of marijuana worsens a person’s memory, and, in the second place, nothing that adversely affects a one’s mental abilities should be legalized.” Compare this to: “I’m not going to tell you because, in the first place, it’s none of your business, and in the second place, you would tell everyone else.”

I will use “logic” to refer to the theory of argument. As is well known, Wenzel (2006 [1990]) distinguishes three main perspectives in argumentation theory according to their object: the rhetorical perspective, focusing on argumentative processes, the dialectical perspective, focusing on argumentative procedures, and the logical perspective, focusing on the products of argumentation (i.e., arguments). Ralph H. Johnson, furthermore, defines the theory of argument as the part of argumentation theory “devoted to the study of the product of the practice: the argument itself.” (2000, p.31). In this context, therefore, we can consider the logical perspective on argumentation and the theory of argument to be roughly equivalent.

The logical perspective studies the properties of arguments, among which are validity, cogency, and so on. The logical properties of an argument are those properties that are relevant to answering the question “Should we accept this claim on the basis of the reasons adduced in its support?” (Wenzel 2006 [1994], p.17), and which can be described without reference neither to the effects of the adduced consideration on the audience nor to the conventional rules of argumentative exchanges. In this way, the logical properties of arguments are defined in opposition to their rhetorical and dialectical properties. That does not mean, however, that the three are not interrelated. The audience’s beliefs about the validity or invalidity of an argument will undoubtedly influence its persuasive power, and any repertoire of dialectical rules will include rules that require participants to use logically sound arguments (or refrain from using logically unsound arguments).

Logical properties can be understood in terms of reasons or in terms of inferences, and, in that sense, we can contrast reasons-based theories of argument with inference-based theories of argument. We can summarize the contrast between these two approaches in a table:

In what follows, I will use the label “inferencism” for the position that argumentation theory should be inference-based, and the label “reasonism” for the position that it should be reasons-based.

The labels “inferencism” and “reasonism” and Table 1 need to be clarified in order to understand them properly. The first point to clarify is that the term “inference” as used in argumentation theory (and in philosophy in general) is ambiguous. In any textbook of logic, for example, “to infer from” refers to a relationship between semantic contents (propositions), while in epistemology and the psychology of reasoning it refers to the psychological action or process of drawing a conclusion from a set of data or evidence. ^{Footnote 1} In Table 1, “inferable” should be interpreted in its logical sense, allowing “inferable” to cover different modalities, such as deductively inferable, inductively inferable, and so on.

The second point is that, depending on the moment and the purpose of the exchange, a good reason can be a pro tanto reason or a weighty reason. I borrow the category of pro tanto reasons from Shelly Kagan, who distinguishes between prima facie and pro tanto reasons:

The term ‘pro tanto’ may be somewhat unfamiliar (partially, perhaps, because my use of it is also slightly ungrammatical), so a word of explanation may be in order. A pro tanto reason has genuine weight but nonetheless may be outweighed by other considerations. Thus, calling a reason a pro tanto reason is to be distinguished from calling it a prima facie reason, which I take to involve an epistemological qualification: a prima facie reason appears to be a reason, but may actually not be a reason at all, or may not have weight in all cases it appears to. In contrast, a pro tanto reason is a genuine reason—with actual weight—but it may not be a decisive one in various cases.” (Kagan 1989, p. 17).

A pro tanto reason, therefore, is a reason that is relevant in the critical examination of an issue. I coined the term “weighty reasons” inspired by the words of John Dewey: “Considerations which have weight in reaching the conclusion as to what is to be done, or which are employed to justify it when it is questioned, are called ‘reasons.’ (1924, p.17). Thus, a weighty reason is a reason that, by itself or in conjunction with others, leads to a conclusion on an issue. Note that an argument can give a weighty reason without the conclusion being inferable from its premises. This happens when that reason leads to a conclusion only in conjunction with other reasons.

The third and final point of clarification is that a distinction can be made between three kinds of reasons, or rather, uses of reasons. Reasons fulfill three main functions: they may be invoked to support claims about what someone ought to do, believe, feel, etc., they can be cited in an explanation, and they can move someone to act or feel in a certain way, to believe something, etc. We therefore speak of justificatory or normative reasons, explanatory reasons, and motivational reasons (see Álvarez 2010). A reasons-based theory of argument is primarily interested in justificatory or normative reasons.

I will use “follows from” to name the relationship between the premises and the conclusion that holds whenever the conclusion is inferable from the premises. In the theory of reasons (a branch of metaethics) the term “favoring” is used to name the relation between a reason and that for which it is a reason.

… the favouring relation, the relation in which features of the situation stand to action or to belief when they are reasons for doing one thing rather than another or for believing one thing rather than another (Dancy 2004, p.79).

A reason is a consideration that favors, or calls for, a certain response (Dancy 2007, p.20).

Hence, the relation of favouring or being a reason for occurs between a set of considerations and an agent’s response. It is, therefore, a ternary relationship between what is favoured (e.g. acting in a certain way, believing something or having some emotion), an agent and the relevant consideration (whatever does the favouring) (Dancy 2007, pp.35–36).

We might say that while the fundamental relation in inference-based approaches is following from, in reasons-based approaches it is favoring. When we need to refer indistinctly to the relationship between the premises and the conclusion, we will use supporting. Thus, for some argument theorists “P supports C” means “C follows or can be inferred from P” and for others “P favors C”.

The proposed division may suggest a false parity. Even if one can conceive of the theory of argument in terms of inferences or in terms of reasons, in practice the predominance of inference-based approaches is overwhelming.^{Footnote 2} This makes it difficult to formulate a reasons-based alternative, because the basic and seemingly innocent vocabulary (argument, premise, conclusion, etc.) is theoretically loaded.

In this paper I aim, first, to show that the distinction between following from and favoring is robust, and that there is a genuine difference between logical perspectives based on inferences and those based on reasons. I will then argue that, as far as argumentation is concerned, a logical approach based on reasons is preferable to a logical approach based on inferences. Furthermore, the development of a reasons-based account of argumentation can help clarify the concept of reasons, which occupies a prominent place in metaethics and epistemology. However, this clarification is beyond the scope of this article and will be left for a future occasion.

To say that C follows from P is to say that P allows the conclusion C to be drawn, even if only tentatively and revisably. In their most promising version for the study of arguments, inference-based logics contextualize the notion of inference. Stephen Toulmin’s conditions of exception or rebuttal are an example of such contextualization:

Modal qualifiers (Q) and conditions of exception or rebuttal (R) are distinct both from data and from warrants, and need to be given separate places in our layout. Just as a warrant (W) is itself neither a datum (D) nor a claim (C), something about both D and C—namely, that the step from the one to the other is legitimate; so, in turn, Q and R are themselves distinct from W, since they comment implicitly on the bearing of W on this step— qualifiers (Q) indicating the strength conferred by the warrant on this step, conditions of rebuttal (R) indicating circumstances in which the general authority of the warrant would have to be set aside. (Toulmin 2003, pp.93–94).

The point is that conditions of exception or rebuttal are not parts of the argument (as premises are), but contextual factors relevant to the logical evaluation of the argument. Thus, with the incorporation of conditions, we move from atomism — the thesis that the logical properties of arguments are intrinsic and context-independent — to holism — the thesis that logical properties are extrinsic and context-dependent.

Although Toulmin’s quote mentions both modal qualifiers and exception conditions, note that I am only claiming that the latter are contextual factors. So, I am not denying that qualifiers can be part of the argument. However, since Toulmin states that the modal qualifier is “built into” the claim (Toulmin, Rieke and Janik, 1984, p. 339), it seems that in any case the modal qualifier would be part of a part of an argument.

I take the parts of an argument to be those elements that differentiate an argument from another (Marraud 2024b, p.626). In denying that conditions of exception or rebuttal are part of the argument, I claim that.

• I ate at La Paloma last Saturday, and the food was great, so, La Paloma Mexican food is great.

is the same argument as.

• I ate at La Paloma last Saturday, and the food was great, so, La Paloma Mexican food is great, unless I happened to hit an exceptional meal.

The only difference between the two is that the second formulation mentions a contextual factor that is relevant to the evaluation of the strength of the argument. Although this point deserves a more detailed discussion (see Marraud 2024b), I will limit myself here to making two observations that support my interpretation of the conditions of exception. The first is that Toulmin, Rieke, and Janik do not include the conditions of exception among the elements of the basic pattern of argumentation that appear in every argument (Toulmin, Rieke, and Janik 2023, p. 271). The second is that they point out that in fact it is impossible to list all the exceptions to which an argument is subject. Instead, conditions of exception are stated explicitly only when there are good practical reasons for doing so; otherwise, exceptions are set aside to be dealt with separately as rebuttals when and only when they arise. (Ibid., pp.100–101). The identity of the argument would therefore be blurred if the conditions of the exception were part of the argument.

The defining feature of holism is that it differentiates what is relevant for evaluating an argument from what is relevant for knowing what the argument is. Let’s imagine that someone proposes to visit today the Machado de Castro National Museum because it is one of the best sculpture museums in Europe. It happens however that today is Monday, and therefore the Museum is closed. The offered argument is clear:

The Machado de Castro National Museum is one of the best sculpture museums in Europe, so, we should visit it today.

That today is Monday is a condition of exception or rebuttal, in Toulmin’s sense, which is relevant to evaluate the argument, although it is not part of it (nor is its denial “today is not Monday”). This means that if it were Monday, the fact that the Machado de Castro National Museum is one of the best sculpture museums in Europe would not be a reason to visit it today. Allowing conditions, therefore, means that logical properties are made context dependent.

The way in which a condition is relevant to the evaluation of an argument is that if an exception occurs, what appeared to be, and in principle would be, a reason ceases to be one. Counterfactually, the quality of the sculpture collection at the Machado de Castro Museum would be a reason for a visit if today were not Monday. To prevent confusion: it is not that, being Monday, the fact that the Machado de Castro Museum is one of the best sculpture museums in Europe is an insufficient reason to visit it, but that it is not a reason at all. For it to make sense to discuss whether something is a sufficient reason or not, it must first be a reason at all. The fact that Madrid is 541 km away from Coimbra is not a sufficient reason for me not to return to Madrid flying like a bird. Since I cannot fly, there are no reasons either for or against my returning to Madrid in this way.

It should be emphasized that exceptions are not a special type of negative premises. If they were, their inclusion would not solve the problems arising from the assumption that all information relevant to establish whether the conclusion follows from the premises of an argument refers to the properties of its parts. This assumption leads to the dilemma of either multiplying the hidden premises indefinitely (today is not Monday, the museum is not closed for construction, streets are not impassable, etc.), or including the logical minimum (if The Machado de Castro National Museum is one of the best sculpture museums in Europe, then we should visit it today) among the hidden premises, interpreting conditions as considerations that are relevant to judge the truth of the logical minimum. The first option blurs the identity of an argument, while the second one renders logical evaluation useless, because then all arguments become logically valid (see Marraud 2024a).

The presence (or absence) of exceptions does not differentiate reasons-based accounts from inference-based accounts. The main difference between inferences and reasons is that the latter are weighed and the former are not. Reasons are better or worse, but inferences are not. This may be overlooked because of the equivocity of the term “inference”. “Inference” can refer to the process of drawing of a conclusion from a set of data or evidence, or to a logical inference, a relationship between statements or propositions. The purpose of the figure.

3. Bart will love cheese

3.1 Bart is Dutch

3.2 If Bart is Dutch, then Bart loves cheese is not to represent a reasoning process, but rather to show that the conclusion “Bart will love cheese” follows from the premises “Bart is Dutch” and “If Bart is Dutch, then Bart loves cheese”. That Bart is Dutch may be a better or worse reason to believe that he loves cheese, but “Bart will love cheese” is either inferable from the premises “Bart is Dutch” and “If Bart is Dutch, then Bart loves cheese” or not inferable. The recognition of exceptions does not change this; it simply obliges us to specify that this conclusion is inferable from those premises, all things being equal. But in any particular situation, “Bart loves cheese” is inferable from “Bart is Dutch” and “If Bart is Dutch, then Bart loves cheese” or it is not.

Consequently, in reasons-based theories of argument weighing occupies a central place. Reasons-based models define validity (used here as a general and undefined term for logical evaluation, meaning “logically good”) in terms of weighing reasons, not of inferences from premises to conclusion. Horty (2012, p. 9), e.g., gives the following definition of validity (which I call “reasonist validity”):

An argument P₁ and… and P_n, therefore C is valid if and only if (1) it has been established that P₁ and… and P_n, so that we are presented with a reason for C, (2) the default rule that provides P₁ and… and P_n as a reason for C has not been excluded from consideration, and (3) we have not endorsed another reason, at least as strong as than P₁ and… and P_n, that supports a conflicting conclusion.

Horty’s definition is holistic in a double sense, since clause (2) makes the validity of the argument depend on contextual factors that determine the applicability of a default rule (op. cit., p.124), and clause (3) on its comparison with other arguments, which obviously are not part of the argument P₁ and… and P_n, therefore C. But what I want to emphasize now is that the validity of an argument depends on how it compares to other arguments, and validity thus becomes a comparative rather than a qualitative concept.

To sum up, there are two distinct ways in which a contextual factor can affect a prima facie reason (i.e., a consideration presented as a reason, although it may later turn out not to be reasons at all). One is by affecting its relevance to the claim for which it is adduced as a reason, and the other by affecting its relative strength. In the first case it is a condition for it to function as a pro tanto reason (i.e., as a consideration that must be taken into account in the examination of an issue, although it may later be outweighed by other reasons), and in the second a modifier (see Bader 2016, p.28). Since reasons are weighted, unlike inferences, the recognition of modifiers, as distinct from conditions, is a characteristic feature of reasons-based approaches.

Conditions and modifiers are associated with two different kinds of counterarguments: rebuttal and refutation, respectively. A rebuttal is an argument that a prima facie reason is not a pro tanto reason, while a refutation is an argument that a pro tanto reason is outweighed by some contrary reason (Leal and Marraud 2022, pp.317–323). Thus, a proper rebuttal shows that something is not a pro tanto reason, and a proper refutation shows that it is not a weighty reason (i.e., a pro tanto reason that has weight in reaching the conclusion). Refutation differs from rebuttal in that it involves the weighing of reasons, and is, therefore, the form of counterargument that distinguishes reasons-based accounts from inference-based accounts.

To rebut an argument, one can point to some exceptional circumstance that prevents what has been presented as a reason from really being one. Thus, the argument that we should visit the Machado de Castro National Museum today because it is one of the best sculpture museums in Europe is rebutted by the assertion that it is Monday. The association of modifiers with refutation is less direct. To refute an argument A is to oppose it with a stronger contrary argument A’, with the double commitment that A’ is valid and as strong or stronger than A. Modifiers can be used to justify or question the greater relative weight given to A’ compared to A. I will talk more about the role of modifiers when I discuss particular weighting.

To avoid being misled by the metaphor of the weight of reasons and arguments, we must bear in mind that weight is always relative weight. That is, weightier-than is prior to weight. That a modifier increases or decreases the weight of an argument only makes sense in the context of implicit or explicit argument weighing. Thus the strength or weight of a reason depends on the terms of comparison that vary from situation to situation. But in addition, the strength of a pro tanto reason varies according to contextual factors called “modifiers”, even when compared to the same set of arguments. Let’s go back to the Machado de Castro Museum. That it is one of Europe’s finest sculpture museums is a reason to visit it today (unless it is Monday), which is intensified if today is the last day of our stay in Coimbra, and is attenuated if we are going to stay in the city for a week. These circumstances are, respectively, an intensifier and an attenuator of the reason for visiting the Machado de Castro Museum, namely that it is one of the best sculpture museums in Europe.

Nevertheless, it could be argued that weighing can be simulated in a holistic inference-based model by treating refutation as a form of rebuttal to avoid weighing — that is, treating modifiers as conditions of exception or rebuttal. Were refutation a special kind of rebuttal, weighing would be avoided, and inferencism — the thesis that in a logically good argument the conclusion is inferable from the premises — could be preserved.

Let us consider an example. That the Machado de Castro Museum is one of the best sculpture museums in Europe is a reason to visit it on Friday, and that we must attend the sessions of the International Colloquium on Argumentation a reason not to. But having two opposing arguments is not having a compound argument with a conclusion. For that to happen, you have to integrate those two arguments into a single argumentation. Connectors like “but” or “although” do that job, joining two arguments to form a compound argument with a definite conclusion (Table 2).

Argument composed with “but”

Full size image

The two adduced reasons are good reasons, but one of them is or is held to be stronger or weightier than the other. We can say that both are pro tanto reasons but that only our having to attend the sessions of the International Colloquium on Friday is here a weighty reason. The point is that here the conclusion is not the conclusion of a set of premises, but of a set, or better, a network of arguments.

I can think of only two strategies for avoiding weighing to save inferencism: (1) claiming that the premises of the “stronger” argument work at the same time an exception to the weaker argument, and (2) postulating hierarchies of types of arguments or reasons. I will consider these strategies in turn.

3.1 The Dual-Role Explanation

To save inferencism a double role in weighing can be attributed to the premises of the stronger argument. On the one hand, they express a reason and, on the other hand, an exception that invalidates the contrary argument. I will use the expression “If A then B” to mean that A is a reason for B, and “unless” to introduce an exception, as does Toulmin (2003, p.94). The above argument, according to the dual-role explanation, would rest on two conditions:

C1. If the Machado de Castro National Museum is one of the best sculpture museums in Europe, then we should visit the Museum on Friday, unless we must attend the sessions of the International Colloquium on Argumentation.

C2. If on Friday we must attend the sessions of the International Colloquium on Argumentation, then we should not visit the Machado de Castro National Museum on Friday.

Thus, the speaker would first claim that on Friday we must attend the sessions of the International Colloquium on Argumentation to show that, given C1, the fact that the Machado de Castro National Museum is one of the best sculpture museums in Europe is not a reason to visit it on Friday. She would then repeat the same consideration, now to conclude that, given C2, we should not visit the Museum on Friday. In this way, the conflict of reasons is avoided, although the maneuver has an intense circular flavor. The reason for asserting that we should not visit the Museum on Friday is at the same time the exception that invalidate the reason to visit it on Friday, which only makes sense if one accepts that the former is stronger than the latter, an assumption that does not even make sense in this inference-based theoretical framework. Even worse, this move turns something that is relative and contextual — the priority of attending the Colloquium over visiting the Museum — into something absolute and fixed.

Weighing is incompatible with the claim that in a logically good argument the conclusion is directly inferable from the premises. Weighing would be avoided if refutation were a special kind of rebuttal. But, as we have seen, such a reduction is unfounded unless some appropriate explanation is given of how the same consideration can be both a reason for affirming something and an exception for rejecting a contrary argument.

Jan Albert Van Laar has developed a sophisticated version of the thesis of the dual role of the premises of the strongest argument, well-illustrated by his analysis of a dialogue constructed from an example of Carl Wellman’s third pattern of conductive argument.^{Footnote 3} The dialogue is as follows (van Laar 2014, pp. 268–270).

Proponent: You ought to take your son to the movies (T).

Opponent: Why? For all you’ve shown, my lawn needs cutting! (C)

Proponent: Okay. But though your lawn might need cutting, you ought to go.

Opponent: Why?

Proponent Because the picture is ideal for children. (S)

Opponent: How would that support that I ought to go even if my lawn needs cutting?

Proponent: It will be gone by tomorrow. (V)

According to van Laar’s analysis, the proponent argues that the opponent should take his son to the movies because the picture is ideal for children (S so T). The opponent counter-argues that he has to cut his lawn, and therefore cannot take his son to the movies (C so not T). In his second intervention, the proposer rejects the need to cut the lawn as a reason for not taking the child to the movies, arguing, when the opponent asks him to justify his rejection, that the picture will be gone by tomorrow (V so it is not the case that if C, then not T).^{Footnote 4} In this way, S plays a dual role: (1) it is a second-order reason for rejecting the need to cut the lawn as a reason for not taking the child to the movies, and (2) it is a first-order reason for taking the child to the movies.

Van Laar’s solution is ingenious and sophisticated, but unnecessarily complicated and not very effective. My main argument against his proposal is that it cannot easily account for the concurrence of opposing modifiers. Let’s imagine that the dialogue continues, and that the opponent argues that today is the only day he can mow the lawn (D). How would van Laar interpret D? There are many possibilities:

• D is second-order reason for denying that if the picture is ideal for children, then the opponent should take his son to the movies.

• D is a third-order reason for denying that if the picture is ideal for children, then it is not the case that if the opponent’s lawn needs mowing, then he should not take his son to the movies.

• D is a fourth-order reason to deny that if the picture will no longer be available tomorrow, then it is not the case that if the picture is ideal for children, then it is not the case that if the opponent’s lawn needs mowing, then he should not take his son to the movies.

• D is simultaneously a second, third and fourth order reason.

On the contrary, according to a reasons-based analysis, what we have here is a reason to take the child to the movies (the picture is ideal for children), a reason not to take the child to the movies (the opponent’s lawn needs mowing), and a modifier that intensifies the relative weight of the first reason (the picture will be gone by tomorrow), tipping the balance in favor of the first reason. In the dialogue extension, a new intensifier is added, now for the second reason (tomorrow is the only day the opponent can mow the lawn), which presumably restores the balance.

3.2 Priority Rules

There is still a second way to rescue inferencism by appealing to priority rules or maxims that hierarchise the types of reasons (e.g. lex posterior derogat (legi) priori). I call this “general weighing”.

Weighing maxims are a kind of commonly used priority rules. These maxims establish that arguments of such type generally outweigh arguments of such other type. The proverb “the end does not justify the means” provides an example of such a weighing maxim. Taken literally, the proverb does not seem to make much sense. Arguments from ends and means are used very often; indeed, according to Fairclough (2018), deliberation is characterized by the intensive use of arguments from ends and means and arguments from consequences. But what the proverb really means is that arguments based on values in principle outweigh arguments from ends and means and arguments from consequences.

Let me now analyze an example of general weighting, first from a reasonist point of view, and then from an inferencist point of view.

The [Supreme] court’s ideological divisions were evident in the recent 5 − 4 decision in favor of same-sex marriage and 6 − 3 for Obamacare subsidies. In each instance, the outcome benefited the country. But the end does not justify the means. The role of the court is to evaluate and decide, not to legislate. (Statesman Journal Editorial Board, “Where did our Constitution go?”. Statesman Journal, July 7, 2015. https://eu.statesmanjournal.com/story/opinion/editorials/2015/07/04/constitution-go/29727587/).

In the first two lines it is argued that the Supreme Court has become a bastion of competing ideologies (using the words in the previous paragraph of the article, not quoted), based on the results of votes on same-sex marriage and Obamacare subsidies. In the following two lines this ideological division of the Supreme Court is assessed, considering reasons for and against, as indicated by the use of the conjunction “but”. The positive evaluation of the ideological division of the Supreme Court is based on the fact that (1) the two decisions of the Supreme Court that prove its ideological division have benefited the country, and the negative evaluation is based on the fact that (2) in making these decisions the Supreme Court has exceeded its responsibilities. Just after the “but” connector comes a general statement that does not mention the Supreme Court and a statement about the role of the Supreme Court. The adage “the end does not justify the means” is used when a pragmatic argument is weighed against an argument based on values to point out that arguments of the second type generally carry more weight than those of the first. That is, its function is to weigh opposing arguments. The following diagram represents this reasons-based analysis:

Statesman editorial board argumentation

Full size image

(The thicker lines delimit two rectangles of the same width containing the premise, “the end does not justify the means”, and the conclusion, that the Supreme Court legislated carries more weight (as indicated by “but”) than the outcome of its decisions, of a (meta)weighing argument).

From a reasons-based point of view, what Statesman editorialists do can be described as follows:

1. (1)

   They accept that the fact that the outcome of the decisions benefited the country is a reason to approve the decisions of the Supreme Court.

2. (2)

   They accept that the fact that the Supreme Court exceeded its responsibilities is a reason to disapprove decisions of the Supreme Court.

3. (3)

   They claim that (2) is a stronger reason than (1), because the end does not justify the means.

4. (4)

   Accordingly, they conclude that the Supreme Court did not act appropriately on the basis that it did more than evaluate and decide.

Some may find Table 3 difficult to understand, and may think it would be clearer to see the Statesman editorial board’s argumentation as a meta-argument, in the sense of Finocchiaro: “an argument about one or more arguments or about argumentation in general” (Finocchiaro 2013, p.1).

Table 4 shows more clearly than Table 3 the points (1)-(3) above, but leaves out (4), and therefore that the Statesman Editorial Board concludes, based on this argument, that the Supreme Court did not act appropriately. In other words, Table 4 represents an argument about other arguments, while Table 4 shows a complex argument with two sub-arguments appearing in the big box at the bottom.

Statesman editorial board Meta-argument

Full size image

In contrast, to explain general weighing by means of conditions of exception, it is posited that the argument “the outcome of the decisions benefited the country” is subject to the exception “unless the means chosen (to benefit the country) are contrary to some value or principle”. When such a circumstance occurs, the rule “the one who wants the end will also want the means” is not applicable. Thus the argument that the outcome of the decisions benefited the country is invalidated, and there is no place to compare its probative force with that of the argument that the Supreme Court exceeded its attributions, which remains as the only reason to decide the matter. Therefore, the conflict of arguments or reasons would be only apparent, and general weighing, so understood, would be compatible with an inference-based logical perspective.

Thus, from an inference-based point of view, Statesman editorialists follow a different path than the one attributed to them from a reasons-based point of view to reach the same conclusion.

1. (1)

   They accept that the fact that the Supreme Court has exceeded its authority is a reason to disapprove of the decisions of the Supreme Court.

2. (2)

   They reject that the outcome of the decisions benefited the country is a reason to approve of the decisions of the Supreme Court, because the end does not justify the means.

3. (3)

   Consequently, they conclude that the Supreme Court did not act appropriately on the grounds that it did more than evaluate and decide.

The difficulty with this inference-based analysis is that it is not true that an argument based on values always carries more weight than an argument from consquences. In fact, the balance depends on the importance of the consequences of the action being evaluated and the degree to which it affects the principle involved (unless, of course, one is willing to subscribe to the maxim Fiat iustitia, et pereat mundus). But any reformulation of the alleged exception to the argument that the outcome of the decisions benefited the country reintroduces weighing — e.g., “unless the means chosen undermine some value or principle to a greater extent than the importance of the good they promote”.

The alternative to hierarchies of types of reasons is particular weighing. Particular weighing consists in alleging circumstances C that increase the relative weight of a reason A or decrease the relative weight of another, opposite reason B, in order to give more weight to A than to B in a particular case. Note that in neither case does C strip B of its condition of reason, so that here there is a genuine conflict of reasons. This means that in certain circumstances argument A has more weight than argument B, and in others it is the other way around or they are on a par. For instance, the fact that Friday afternoon Colloquium lectures are of little interest to us would intensify the weight of the reasons for visiting the Museum that day; and again, the fact that we leave Coimbra on Sunday would intensify the weight of the reasons for attending the Colloquium, since then we could visit the Museum on Saturday.

This is a weighing meta-argument, whose conclusion is that one reason (right) outweighs another reason (left), and whose premise is a modifier (Table 5).

A weighing (meta)argument

Full size image

The use of a modifier suggests that the speaker believes that if we were interested in the lectures on Friday afternoon, the fact that the Machado de Castro Museum is one of the best sculpture museums in Europe would not be a weighty reason to visit it on Friday.

The above argument, and the role of the modifier, is therefore easily explained in the context of balancing reasons, but not when the modifiers are to be interpreted as conditions.

To interpret the modifier as a condition of rebuttal, that the Friday afternoon lectures are of little interest to us would be an exception that makes the obligation to attend the sessions of the International Colloquium no longer a reason not to visit the museum on Friday. In other words, the speaker would accept the conditional:

If on Friday we must attend the sessions of the International Colloquium on Argumentation, then we should not visit the Museum on Friday, unless the Friday afternoon lectures are of little interest to us.

The inferencist strategy for dealing with particular weighing is to reduce modifiers to exceptions — or refutation to rebuttal —, so that what appeared to be a conflict of reasons turns out to be only apparent. An exception is a consideration that shows that what was initially taken to be a reason really is not. Thus, when an exception is rightly claimed, the conflict between reasons is solved because one of the conflicting reasons turns out to be only apparent. However, this distorts the debate.

A key difference between conditions of exception and modifiers (rebuttals and refutations) is that the former affect the relation between premises and conclusion of a single argument, while the latter affect the relationship between two arguments. As Horty points out: “it is a consequence of our definition that an intensifier must make a reason stronger than some other reason — it cannot simply make a reason stronger per se” (Horty 2012, p.145). This makes it debatable whether the alleged fact that the Friday afternoon lectures are of little interest to us is a genuine exception to the argument that we should not visit the Museum on Friday since we have to attend the sessions of the International Colloquium on Argumentation. It seems that there is no direct relationship between the fact, if any, that the Friday afternoon lectures are of little interest to us with the argument that The Machado de Castro Museum is one of the best sculpture museums in Europe. This relationship exists because of the fact that the Friday afternoon lectures are of little interest to us exempts us from attending those sessions, given our interest in visiting the museum. That is, the supposed exception is only understandable in the context of a comparison of (the strength of) two arguments.

So far, I have argued that there are genuine differences between inference-based and argument-based theories of argument (or logics). In an inference-based theory of argument, a good argument logico sensu authorizes the assertion of its conclusion, whereas in a reasons-based theory it affects the overall orientation of the argumentation in which it occurs. The logical properties of an argument depend on two types of contextual factors: conditions and modifiers. Inferencism is compatible with the recognition of the former, but not of the latter, since the modifiers of a reason cannot be reduced to conditions of exception or refutation. Moreover, I have shown that reasons-based theories can account for argumentative phenomena, related to weighing, that resist inference-based explanations. In a reasons-based approach to argumentation, reasons are not something given, but something constructed in argumentative exchanges. Thus, the development of a theory of argumentation based on reasons can help to clarify the concept of reason, which occupies a prominent place in metaethics and epistemology. But, as Kipling said, that is another story.