Pluralism about practical reasons and reason explanations

ABSTRACT

This paper maintains that objectivism about practical reasons should be combined with pluralism both about the nature of practical reasons and about action explanations. We argue for an ‘expanding circle of practical reasons’, starting out from an open-minded monist objectivism. On this view, practical reasons are not limited to actual facts, but consist in states of affairs, possible facts that may or may not obtain. Going beyond such ‘that-ish’ reasons, we argue that goals are also bona fide practical reasons. This makes for a genuine pluralism about practical reasons. Furthermore, the facts or states of affairs that function as practical reasons are not exclusively natural or descriptive, but include normative facts. That normative facts can be reasons justifies a pluralism about reason explanations, one which allows for what we call enkratic explanations in addition to teleological ones.

KEYWORDS:

• Pluralism
• motivating reasons
• explanatory reasons
• action explanation
• folk psychology
• goals

1. Introduction

Although this is rarely appreciated, the last decades have seen parallel developments in theorizing about folk psychology within cognitive science and in theorizing about the nature of reasons for action within philosophy of action. Both fields come from a firm commitment to psychologism or subjectivism: they used to identify the explanantia of human action – the reasons for which people act – with mental states, typically belief-desire pairs (Davidson 1963/1980). Both have in recent years moved away from this commitment towards objectivism, a more world-involving picture, on which what explains or motivates actions are not mental states, but factors outside the mind. For instance, cognitive scientists who endorse ‘teleology’ hold that infants explain the behavior of others in terms of their goals (Gergely and Csibra 2003; Perner and Roessler 2010); and objectivists in the philosophy of action identify motivating reasons with ‘worldly’ facts or states of affairs (Dancy 2000; Alvarez 2010).

At the same time there has been a mismatch between developments in these fields. In discussions about the nature of folk psychology, cognitive scientists have moved towards pluralism. They have effectively kept mental states on board as part of one explanatory strategy, while accepting that people also employ various other strategies for understanding actions (cf. the forthcoming Synthese special issue Folk Psychology: Pluralistic Approaches). Among other things, cognitive scientists have argued that goals and normative facts may be among the reasons which explain one agent's action in the eyes of another (Gergely and Csibra 2003; Clément, Bernard, and Kaufmann 2011).

In the philosophy of action, the situation is more ambivalent. At a conceptual level, one finds a new pluralism. There is a widely accepted trichotomy between normative or justifying, motivating, and explanatory reasons (Alvarez 2016). Normative reasons are reasons to perform an action, considerations which count in favor of an agent acting this way and thereby contribute to the justification of her action. The fact that exercising is healthy is a normative reason for me to exercise. Motivating reasons are reasons for which an agent performs an action, considerations which make performing the action pro tanto attractive from her point of view. Explanatory reasons are reasons why the agent acts a certain way, that is reasons which explain her acting. These types of reasons can coincide, yet they can also come apart. For instance, there may be several justifying reasons for me to exercise, yet only some or perhaps none of them may motivate me; in fact, I might refrain from exercising altogether. And while all motivating reasons can contribute to explaining an action, explanatory reasons do not always motivate. If I punch Peter because I have learned that Peter betrayed me, the fact that I know this may explain my punching; yet it is a fact about Peter, that he betrayed me, which motivates me to punch him.

At the same time, at an ontological level a new monism holds sway. Many philosophers of action assume that it has to be either mental states (subjectivism) or facts/states of affairs (objectivism) that are the reasons which motivate an agent to act and thereby explain her action. Those who abandon subjectivism often combine a conceptual pluralism concerning the function of reasons with an ontological monism about the nature of reasons. By the lights of austerely monist objectivism (called ‘austere monism’ in the following), for instance, although the normative/motivating/explanatory trichotomy captures three distinct roles that reasons can play, all these roles are occupied by entities of the same category, namely facts (Alvarez 2010, 33–35). Not just normative reasons but motivating and explanatory reasons as well are all and sundry facts.

Against this backdrop, our paper addresses the question of whether and to what extent objectivists about reasons for action, or practical reasons, should adopt a pluralism about practical reasons and action explanation. The tenor of our answer: objectivism about practical reasons can and should be combined with pluralism. The pluralism we have in mind concerns both the nature of practical reasons and the structure of action explanations. We argue for an ‘expanding circle of practical reasons’. We start out from an open-minded monist objectivism (‘open-minded monism’ in what follows). At the very least, practical reasons are not limited to actual facts; they are states of affairs, i.e. possible facts that may or may not obtain. Next, going beyond such ‘that-ish’ reasons, we argue that goals are also bona fide practical reasons. This makes for a genuine pluralism about practical reasons. Finally, the facts or states of affairs that function as practical reasons are not exclusively natural or ‘descriptive’, they include normative facts. As we will argue, that normative facts can be reasons justifies a pluralism about reason explanations. Our arguments take motivating reasons as their starting point and then extend to normative reasons. But since both motivating and normative reasons are kinds of practical reasons, our results entail a pluralism about practical reasons in general.

We begin with an overview of competing pictures of action explanation and reasons for action relevant to cognitive science and philosophy of action and a discussion of monism and pluralism in both fields (section 2). Section 3 clarifies how the goals which teleologists accept as bona fide motivating/explanatory reasons fit into an objectivist picture; it also makes out a case for such a view, and thus for a genuine pluralism. In section 4, we turn to normative facts as reasons for action. We discuss how objectivists can make room for these and argue, against restrictive forms of teleology, that they call for a pluralism about reason explanations.

2. Reason explanations and objectivism about practical reasons

Both the philosophy of action and the area of cognitive science concerned with folk psychology deal in action explanations, but there are also important differences between the two fields. Philosophers of action attempt to specify what it is that makes something an action rather than a mere happening or behavior. They ask whether the distinguishing feature of action is that it requires a particular kind of explanation, namely one that makes reference to the agent's reasons. Cognitive scientists, by contrast, are concerned with getting clear on the capacities of people – ‘the folk’ – to explain and predict others’ actions. While philosophers strive to get at the essential characteristics of acting through the peculiarities of action explanation, cognitive scientists seek to understand the explanatory strategies that ordinary humans at various stages of development actually employ to understand actions. Philosophers are often concerned with what action really consists in, whereas cognitive scientists can allow that the folk explain the behavior of others in ways that are only tenuously connected with what really accounts for that behavior. Still, there is much overlap in the kinds of action explanations explored by philosophers of action, on the one hand, and those attributed to the folk by cognitive scientists, on the other. This holds in particular for those philosophers who hold that the explanatory schemes ordinary folk employ in making sense of each other accurately capture the relevant aspects of agency.

There are multiple ways of explaining and predicting what an agent does. For instance, neural and physiological states can explain why an agent behaves in a certain way (‘Fiona started screaming because her C-fibers were misfiring’). Moreover, what an agent does can be explained by dispositions including habits (‘Kant will be here any minute, he walks by our house every day at this time’), character traits such as virtues and vices (‘He’ll help me, he's kind’ [Andrews 2008] or ‘As a punctual person, she showed up on time’ [Alvarez 2010]). Similarly, long-standing preferences can explain why a person acts in a certain way (‘He's eating chocolate because he likes chocolate’ [Kalish and Shiverick 2004]). Actions can also be explained by highlighting the agent's competences, e.g. her competence to do what her situation demands (Glock and Schmidt 2019).

While all these explanations specify reasons why an agent acted, they do not appeal to what motivated her, the reasons for which she acted. Such motivating reasons are considerations ‘in the light of which’ an agent acts (Dancy 2000), reasons that persuade her to act in a certain way. Philosophers tend to agree that the explanation of an action in terms of the agent's motivating reasons is a privileged kind of explanation: On the one hand, it displays what led to the action from the agent's point of view; on the other hand, it reveals the action as an action. ‘The characteristic way in which motivating reasons explain is thus one that distinguishes actions from mere behavior, i.e. from non-intentional movements and events.’ (Mantel 2018, 154; see also Davidson 1971/1980). Call this type of explanation the ‘reason explanation’ of an action (Alvarez 2010, 170). In keeping with this approach, theorists concerned with folk psychology also tend to describe humans as understanding others in terms of the reasons that moved them, be it their desires and beliefs, as the traditional picture has it (Fodor 1987), or their situations and goals, as teleologists suppose (Gergely and Csibra 2003; Perner and Roessler 2010).

In light of their central importance to both fields, our focus is on reason explanations. And our question is: Assuming objectivism, what are the prospects for pluralism about these explanations, and about the motivating/explanatory reasons invoked by them?

Traditional philosophy of action conceives of motivating reasons as the mental states or events that cause an action; it thus subscribes to subjectivism (Smith 1994). Relatedly, Humeanism holds that reason explanations are explanations in terms of belief-desire pairs which cause the agent to act (Davidson 1963/1980; Smith 2004). For instance, my putting the chocolate in my mouth is explained by, first, my desire to eat chocolate, and second, my belief that to eat the chocolate, I have to put it in my mouth. This view has it that reason explanations are both causal, since belief-desire pairs (or their ‘onslaughts’) cause the action, and rationalizing, since the action is shown to be rational in the light of these pairs. Correspondingly, the orthodox picture of folk psychology takes ordinary reasoners to explain others’ actions exclusively in terms of belief-desire pairs. Take ‘theory-theory’, which presents folk psychology as a theory that people use to explain and predict others’ actions (Baron-Cohen 1995; Gopnik and Wellman 1992). According to theory-theory, folk psychology invokes mental states, particularly beliefs and desires, as well as law-like regularities in accordance with which these mental states produce matching outward behaviors.

In the last two decades, philosophers of action have started to focus on reason explanations of a different, objectivist kind. These explanations do not appeal to mental states or events producing an action, but to facts or purported facts in whose light agents make up their minds to act (e.g. Dancy 2000; Parfit 2011). What motivates me to go to the kitchen cabinet, for instance, is not my desiring to eat chocolate or my believing that the chocolate is in the cabinet, but rather the fact that the chocolate is in the cabinet. Correspondingly, reason explanations use the (actual or purported) facts in the light of which the agent acts as explanantia of her action. On this view, what distinguishes actions from mere happenings or behavior is that actions are performed by the agent for a reason which she has, a consideration which casts a positive light on so acting.

In addition to facts or states of affairs, goals or good (desirable) outcomes that are reached by acting have been used as explanantia of action. This idea can be found in both theory of action and cognitive science (e.g. Anscombe 1962; Gergely and Csibra 2003; Perner and Roessler 2010). I open the cabinet in order to take the chocolate; the goal of my action is to take the chocolate; this is the intention with which I open the cabinet. Such explanations of action are teleological or purposive – they make the action intelligible by appeal to its point. They are distinct from Humean explanations: The latter make actions intelligible by appeal to mental states. Goals are not mental states, but rather something extra-mental, such as (possible) states of the world which agents aim to realize.¹ Cognitive scientists like Gergely and Csibra (2003) have claimed that ontogenetically, teleological understanding precedes belief-desire psychology. Infants begin by grasping what the goal of a behavior in a specific context is, and then form expectations of how the agent will act by calculating the most efficient means to the end in that context.

Last but not least, objectivists have proposed normative facts as explanantia of actions (Baillargeon et al. 2015; Clément, Bernard, and Kaufmann 2011). Someone's φ-ing may be explained by the fact that she ought to φ, by the fact that she has good reason to φ, or by the fact that φ-ing is permitted. The fact that the daycare center has the rule that all children put their shoes in the cabinet, for instance, helps young children predict where a girl will put her shoes. Just like non-normative facts, such normative facts can explain an agent's action only provided that she is aware of them, since only then can they motivate her to act in accordance with the norm. This suggests that normative facts can be among the agent's motivating reasons.

The preceding paragraphs indicate that a variety of motivating/explanatory reasons can come into view from an objectivist perspective. We shall now investigate whether this puts pressure on objectivists to endorse a pluralist picture of practical reasons and reason explanations.

3. Pluralism about practical reasons: the case of goals

According to austere monism, all practical reasons, including motivating reasons, are facts. By contrast, we here start out from open-minded monism, which makes room for states of affairs as motivating reasons. It allows that, in error cases, motivating reasons are what the subject falsely believes, namely states of affairs, or potential facts (Dancy 2000; Glock 2019).² This preserves the unity of motivating reasons – and of practical reasons more generally – by treating them as all being of a general kind, viz. states of affairs, while allowing that some motivating reasons may not be facts. Take Parfit's well-known ‘hotel case’. A sees smoke coming from underneath the door of her hotel room and jumps out of the window, in the mistaken though reasonable assumption that there is a fire. According to austere monism, in such cases that there is a fire is no (or at best an apparent) reason for A to jump, since it is not a fact. By contrast, according to open-minded monism, that there is a fire, a non-obtaining state of affairs, is the motivating reason for which A jumps.

Our focus in this paper is not an assessment of who wins the debate between austere monism and open-minded monism, which has been extensively covered elsewhere (Alvarez 2010; Hornsby 2008; Schroeder 2008; Littlejohn 2012; Glock and Schmidt 2019). Instead, we turn to reasons for objectivists to go beyond open-minded monism, and to endorse a genuine pluralism about motivating reasons and, by extension, practical reasons generally. To this end, let us turn to goals as motivating reasons, thereby challenging the idea that all practical reasons are of a that-ish nature.

Nothing would appear more commonsensical than that agents are motivated by their objectives or goals. Moreover, behavior directed at achieving a goal of one's own qualifies as acting intentionally in one important sense: the action is performed with the intention of achieving the goal. There is also a connection between intentional action and responsiveness to reasons. Intentional actions are commonly equated with actions that the agent performs for a reason. Admittedly, there is a case for holding that A can φ ‘intentionally’ without φ-ing ‘for a reason’, in pursuit of a goal (Hacker 2007, 211–214). But that case revolves around understanding ‘φ-ing intentionally’ as φ-ing in a controlled manner, rather than as φ-ing with a specific intention. By contrast to the former, the latter is both equivalent to acting in pursuit of a goal and sufficient for acting for a reason. In a perfectly innocuous sense, acting in pursuit of a goal qualifies as acting for a reason. And the goal or intention with which an action is performed counts as a reason for which the action is performed. I can answer the question ‘Why are you φ-ing?’ by citing a goal of mine; the answer gives a reason for which I φ (Anscombe 1962, 9).

For instance, the question

(1) Why is A running to the bus-stop?

is standardly answered by

(2) In order to catch the bus.

And the question

(3) What is A's reason for running to the bus-stop?

can felicitously be answered by

(4) Catching the bus.

Infinitives as in (2) and progressives as in (4) are paradigmatic ways of signifying goals. Accordingly, our concepts and explanatory practices conform to a goals are reasons principle:

(GR) A φ-s in order to ψ / bring it about that p ⇒ A's (motivating) reason for φ-ing is to ψ / bring it about that p.

Now plausibly, the goal for which A φ-s is typically not already an obtaining fact at the time of A's φ-ing; if it were, A's φ-ing would be pointless, since its goal would already have been achieved. We should therefore distinguish, e.g. A's goal of climbing the Matterhorn from

i. the fact that A climbs the Matterhorn,

but also from

ii. the fact that A has the goal of climbing the Matterhorn or

iii. the fact that climbing the Matterhorn is good.

After all, A's goal is to climb the Matterhorn, not for herself to have the goal of climbing the Matterhorn – as in (ii). And the fact that climbing the Matterhorn is good – as in (iii) – may be why A has the goal of climbing it, but it is not itself her goal. These considerations further discredit austere monism, and appear to bolster open-minded monism: Austere monism is incompatible with acknowledging goals as motivating reasons, given that goals are not facts. Open-minded monists, on the other hand, can make room for goals as motivating reasons as follows. Goals are intended outcomes; these outcomes are states of affairs; and these states of affairs obtain if A actually reaches her goal. This response secures goals as motivating reasons by treating them as a sub-class of possible states of affairs. However, as observed above, goals are standardly specified through non-that-ish constructions. Many goals are themselves actions – I φ with a view to performing another action – to ψ. Here, what I aim for is not, at least in the first instance, that a state of affairs in which I ψ is realized, but simply to ψ or ψ-ing.

In response, monists can challenge, first, our claim that (some) goals are neither facts nor even states of affairs that may or may not obtain.³ They can also challenge our claim that goals can be motivating reasons.

Starting with the first challenge, one might use Thompson’s (2008) and Wiland’s (2013, 317) idea of ‘actions-in-progress’ to argue that goals already obtain while being pursued by an agent. One might grant that (i) – the fact that A climbs the Matterhorn – is not A's goal, while arguing that

(i*) the fact that A is climbing the Matterhorn

is A's goal, even as she is clambering up the mountainside. Actions typically take time and are composed of sub-actions that need to be completed to perform the overarching action. English grammar reflects this by way of the progressive form, the imperfective ‘is φ-ing’. The larger action that is in progress is typically a goal pursued by the agent in performing its sub-actions. Thus A pursues the goal of climbing the Matterhorn in clambering up the mountainside, at the time she is climbing up the Matterhorn.

We concede that some actions are indeed ‘in progress’. In such cases, the goal of an action can be given by a wider description (‘climbing the Matterhorn’) of what one is already doing under a narrower description (‘clambering up the mountainside’). We also concede that for some types of actions an agent can be φ-ing even if she doesn't φ. The progressive can be true of actions that aren't successfully completed (Thompson 2011). A can be climbing the Matterhorn even though she never reaches the summit. But not all progressive constructions signify a goal which obtains even while being pursued. There is an understanding of ‘A is ascending the Matterhorn at t₁′ on which it is false unless she reaches the summit at t₂. Even on the understanding favored by Thompson's approach one must distinguish between A's goal – reaching the summit of the Matterhorn by climbing – from the activity which is already taking place in pursuit of that goal. A has not reached that goal simply by pursuing it. In other cases, it is unequivocally precluded that the intended action is already taking place. During a stressful time at work, I may adopt the goal to relax once I’m on vacation. But unfortunately, I am not yet in the process of relaxing. Finally, there are goals that are states, not actions. A may pursue the goal of being filthy rich, or of being president, at a time at which she is neither. Consequently, some goals do not obtain at the time the agent pursues them.

Now, the last group of examples supports the open-minded monists’ claim that goals are at least potential facts or states of affairs. But their case is weakened rather than strengthened by the idea that goals are actions-in-progress. For actions take place or occur in space and time, and actions-in-progress even unfold in space and time. This does not hold for either facts or potential facts. Even if actions-in-progress support the idea that goals can in some sense be actual before being attained, they positively count against the idea that they are that-ish. Small surprise, then, that their champions are either silent on the question of whether goals can be reasons (Wiland 2013) – or they lend succor to pluralism. Thus Thompson (2008, ch. 8) holds that what we intend is primarily to do or accomplish something, i.e. something non-propositional. On the assumption defended above, namely that what an agent intends is her goal, this diagnosis extends immediately to goals.⁴

But what of a straightforward monistic argument which bypasses action-in-progress and argues straightforwardly that all goals must be analyzed in that-ish terms? For instance, one might insist that

(5) A intends to φ

is equivalent to

(6) A intends that A φ-s (should φ).

However, even if all infinitival and progressive constructions could be faithfully paraphrased by propositional constructions, it would not follow that the latter are more basic. For the possibility of paraphrase cuts both ways. And non that-ish constructions as in (5) are much more readily understood than that-ish ones. This strongly suggests that at the least some non-that-ish goals feature as reasons in our folk-psychological explanations of actions.

Furthermore, there are reasons for upholding that A's goal of φ-ing differs from A's goal that she φ (which she intends to bring about). We start from Baier’s (1970) distinction between A's intending to φ – as in (5) – and A's intending to bring it about that she φ-s, or to get herself to φ – as in (6). Imagine that Boris has difficulties saying ‘no’ whenever his friend Jack asks him a favor. Boris decides finally to stand up for himself, and so intends to get himself to say ‘no’ by giving himself a pep-talk. This is distinct from Boris intending to say ‘no’ to Jack. The two intentions differ with respect to their target, with respect to whether Boris aims to exert his causal powers on Jack or, respectively, on himself (Campbell 2019).

Analogously for goals. Assume that, unlike Boris, Dilara has no difficulties saying ‘no’ to Jack, and it is her goal to say ‘no’ to him the next time he asks her a favor. According to open-minded monism, Dilara's goal is the state of affairs that she will say ‘no’ to Jack, which she aims to actualize. But the open-minded monist's way of specifying Dilara's goal collapses the goal she actually has, to say ‘no’ to Jack, into a goal that she doesn't have, viz. getting herself to, or bringing it about that she says ‘no’ to him. One can apply Baier's distinction to goals only by taking non-that-ish ways of picking out goals at face value.

With this, we turn to the second challenge our opponents can put forth: Can goals be motivating reasons even if not included under the rubric of states of affairs?⁵ By way of tackling this challenge, note that, like states of affairs, goals satisfy the normative constraint (Dancy 2000, 103): it must be possible for an agent to be motivated by and to act for reasons actually favoring the action – motivating reasons must be capable of being normative reasons. And like states of affairs, goals are candidates for actually favoring the action. Admittedly, goals do not favor actions in the same way as that-ish reasons. But there is a tight analogy. The consideration that p can actually count in favor of φ-ing if it is a fact, but it need not. The goal to ψ or to bring it about that p can count in favor of φ-ing, but need not do so as a matter of instrumental fact: φ-ing may not be the optimal or even a feasible way of ψ-ing or bringing it about that p. The goal of catching the bus can favor running to the bus-stop; yet it need not do so, for instance if it is too late anyway or if there are better ways of catching the bus. At the same time, a goal that actually favors an action may nevertheless fail to motivate the agent. In both respects, goals are structurally similar to states of affairs.

Philosophers like Dancy (2000) have argued that what really does the favoring in such cases is not an agent's goals such as (4), but facts like (iii), for example,

(7) It is good for A to catch the bus.

In a similar vein, Wlodek Rabinowicz has suggested in discussion that goals as such are not motivating reasons. What features in an agent's reasoning, and thus is a reason, is rather instrumental facts of the form

(IR) φ-ing is a way of ψ-ing / bringing it about that p.

We concede that evaluative or instrumental facts must be in the background for a goal to actually favor a course of action, and that they are then also normative reasons for that action. But this doesn't entail that only facts can favor actions, to the exclusion of goals. It is widely accepted that whether a fact is a normative reason for an action depends on the agent's practical situation (e.g. Dancy 2004; Way 2017). Why then shouldn't one allow that whether a goal is a normative reason similarly depends the agent's practical situation, evaluative and instrumental facts included?

Overall, given the strong parallels between goals and standard normative reasons, there is a good case for regarding goals – even non-that-ish ones – as ‘favorers’ of actions, and thus, given the close connection between normative and motivating reasons, as motivating reasons. For one thing, our examples demonstrate that specifying a goal can be an answer to a why-question. For another, the dependency relation between goals and pertinent evaluative and instrumental facts runs in both directions: The invocation of that-ish reasons for its part presupposes a background of goals. Thus Alvarez (2010, 173) grants that ‘ … any explanation of an action performed for a reason involves explicit or implicit reference to the goal or purpose in pursuit of which the agent acted because the goal shows the connection, in the agent's eyes, between that reason (what the agent believed) and the action’. It is the relation between the fact that the bus is approaching and A's goal of catching it that makes running to the bus-stop attractive in A's eyes. But according to Alvarez, it is that fact rather than the goal that is A's motivating reason. At the same time, she rightly concedes that we can answer the question ‘Why are you φ-ing?’ by citing a goal. She nonetheless resists the suggestion that goals can themselves be motivating reasons, on the following grounds:

[a] If goals were reasons, it would follow that any animal that acts in pursuit of a goal would also act for a reason. [b] But whereas it is fairly uncontroversial that a dog who digs in order to find a bone acts in pursuit of a goal, [c] it is much more controversial to say that the dog acts for a reason: [d] that what motivates the dog to dig is a reason, namely that digging is a means to find the bone, and that the dog acts guided by that reason. (Alvarez 2010, 98; our numbering)

[b] acknowledges that the dog digs in order to achieve a goal. But against the ‘goals are reasons’ principle, [a] takes for granted that reasons should not be construed as something for which animals can act. Do [c] or [d] vindicate that assumption? It is arguable that there are perfectly legitimate attributions of that-ish reasons to animals (Glock 2019). Admittedly, the current issue is not whether animals can act for reasons. But it is whether the implication that they can counts against the idea that goals can be reasons. If animals can in principle be credited with doing things for reasons of any kind, the fact that treating goals as reasons implies that animals can act for reasons no longer constitutes a reductio ad absurdum. Furthermore, characterizing the behavior of higher animals as goal-directed is uncontroversial. Specifying a goal like ‘Finding the bone’ is a legitimate answer to the question ‘What is A's reason for digging?’, irrespective of whether A is a human or an intelligent animal. In summary, Alvarez's denial that goals are motivating reasons and her denial that animals can act for reasons are mutually supportive. However, folk-psychological practice counts against the first claim, as Alvarez acknowledges; in conjunction with cognitive ethology, it also counts against the second.

Admittedly, Alvarez (2010, 40) also provides an independent argument for rejecting goals as reasons. Reasons are the premises of good reasoning; furthermore, by contrast to facts like (IR), goals do not feature as premises in practical reasoning; therefore, they are not reasons. However, the assumption that reasons simply are premises is itself contested (Logins 2020; Schmidt 2020). In any event, our examples show that goals can fulfill one of the central functions of reasons, namely answering why-questions, even if they cannot fulfill the function of featuring in practical syllogisms.

This leaves the austere monist with a final reservation: a reason must be an explanans (see Hyman 2015, ch. 5). Goals, not being facts, cannot satisfy this requirement. The pluralist has two responses. The first is that goals, unlike, e.g. non-obtaining states of affairs, can be explanantia, provided that it is a goal the agent in fact has. The traveler's goal of catching the bus explains her starting to run; the hotel guest's goal of avoiding calamity explains her jumping, etc. Both explanations are factive, irrespective of whether the goal actually favors the action, as in the first example, or not, as in the second.

Monists would retort that the explanans in such cases is the fact that the agent has a certain goal, rather than the goal itself. Even if they were right, however, this would still leave a second response. By contrast to explanatory reasons, motivating reasons per se need not explain anything. Rather, they need to weigh with the agent. Unlike their explanatory role, that motivating role can be fulfilled by things other than facts – states of affairs that are believed to obtain without doing so, or goals that are not (yet) reached.

We should therefore endorse a pluralism about practical reasons: Not only states of affairs but also goals, conceived as distinct from – obtaining or not (yet) obtaining – states of affairs, can play the role of motivating reasons. In this regard we defend teleologists like Perner and Roessler (2010) or Gergely and Csibra (2003) against monist objections. Furthermore, our argument also shows that goals can play the role of normative reasons.

4. Pluralism about reason explanations: the case of normative facts

We now proceed to arguing for more expansive forms of pluralism. We start with a further variety of practical reasons: the normative facts that ordinary folk appeal to in reason explanations. We conclude that normative facts should be taken seriously as motivating reasons. We then use the corresponding reason explanations to make a case for a pluralism about reason explanations; the latter encompass both teleological and ‘enkratic’ explanations.

Julie puts her shoes into the shoe cabinet. What moves her to do so is the fact that, at her daycare, it is a rule that children store their shoes in the cabinet; it is that – as she knows – she ought to put her shoes in the cabinet. So, the fact that she ought to put her shoes in the cabinet (or that there is a rule to do so) is the reason for which she puts her shoes there. A bystander can explain her action by appeal to the fact that the daycare has the rule that children put their shoes in the cabinet, and he can successfully predict where she will put her shoes on the basis of the fact that she ought to put them there.

Candidates for normative facts that can motivate agents and be used to explain or predict their actions include: facts involving norms or rules (‘It is a rule that children put their shoes in the cabinet’); ‘ought’ facts (‘Children ought to put their shoes in the cabinet’); facts about permissions (‘It is permissible to drop one's shoes where one stands’); facts about right and wrong (‘It is wrong to hit children’); or facts about what there is reason to do (‘There is a good reason to put one's shoes in the cabinet’).⁶

Do motivating reasons include such normative facts? When we consider cases of practical reasoning realistically and without axes to grind, it seems that they do. For one thing, it is perfectly ordinary for parents to say, for instance, that they really wanted to hit their daughter because of her terrible behavior, but that what persuaded them not to do so was that hitting children is morally impermissible/wrong. Or an agent might be in a situation in which, because of limited access to her environment, she knows no facts favoring her φ-ing, except that she ought to φ – in such a situation, she can plausibly act appropriately by φ-ing, and the only motivating reason available for her is the fact that she ought to φ (cf. Schmidt 2017). For another, subjects can be more or less ‘enkratic’, i.e. motivated by facts about what one ought to do (see Kiesewetter 2017). If an enkratic subject is aware of the fact that she ought to φ, this will move her to (intend to) φ. Third, studies by Clément, Bernard, and Kaufmann (2011) indicate that three-year-old children actually do better at predicting the behavior of others on the basis of normative facts than on the basis of facts about another’s mental states. Similarly, Baillargeon et al. (2015) argue that sociomoral principles such as reciprocity and fairness are used even by very young children to predict the behavior of others. Accordingly, including normative facts as motivating/explanatory reasons is in line with linguistic practice, common sense, and empirical findings. It is even accepted by some austere monists (Alvarez 2016). For all that, it faces an objection from the thought that only descriptive or natural facts can be normative reasons and thus also motivating reasons.

The objection starts out from the claim that normative reasons cannot themselves consist in normative facts. Imagine that we ask Julie, ‘what reason is there to put your shoes in the cabinet?’, and she answers

(8) In the daycare it is a rule that we put our shoes there.

(8) allows of two interpretations, both of which appear to militate against normative facts as reasons. (8) can function as what von Wright (1963, viii, 108) calls a ‘norm-proposition’, to the effect that a community follows a norm. Such facts about norm-guided behavior certainly exist, but they are not themselves normative facts. They concern people's behavior and mental lives, and that they can function as normative reasons is beside the current issue. Alternatively, (8) might express a rule with a normative import, prescribing a certain behavior. In that case, the objection runs, (8) does not specify a reason that speaks in favor of putting one's shoes away, like the fact that otherwise children will trip over them or that offenders are liable to get into trouble. The latter are natural or descriptive facts that favor putting one's shoes in the cabinet. By contrast, the fact that there is a rule is not something that putting one's shoes away has going for it. At most, that fact indicates that there is something else to be said for putting one's shoes in the cabinet. Mutatis mutandis for alleged normative reasons such as that one ought to put one's shoes in the cabinet or that there is a good reason to put one's shoes in the cabinet.⁷ So normative facts cannot be normative reasons.

The monist's second step is a reductio ad absurdum which extends the problem to motivating reasons. Assume that some motivating reasons indeed consist in normative facts. If the normative constraint is correct and motivating reasons must be capable of being normative reasons, then some normative facts would have to be capable of being normative reasons. But the first step showed that no normative fact can qualify as a normative reason; so, it seems that no such fact can qualify as a motivating reason either. If the folk explain actions as motivated by normative facts, then they employ an inappropriate kind of explanation.

In response, we reject the first step of the argument. It is incorrect that only natural or descriptive facts can be normative reasons. In established discourse, whether everyday or specialized (e.g. jurisprudence or pedagogy), it is spectacularly unproblematic to say that the fact that the agent ought to φ speaks in favor of or is something to be said for her φ-ing. We suspect that the philosophical scruples are based on a failure to distinguish the norm itself and the reason it provides for a subject to behave in a certain manner from the reasons for or against adopting a norm. Even if all the reasons for adopting a norm were descriptive and natural (e.g. concerned the preferences of potential subjects of the norm), the norm itself would retain a special status. The formulation of a norm does not state that there are reasons for adopting or following the norm, let alone list these reasons. Rather, it enunciates such a norm. This is obvious in the case of legal norms, the codified statements of which do not include the case for adopting the norm (which will include reasons of various types, material, procedural, etc.). Nevertheless, the norm enunciated can be a pro tanto normative reason for acts or omissions, one which is obviously distinct from the reasons for instigating or accepting the norm. We must distinguish between norms as reasons and reasons for norms.

Let us grant that there must be an explanation of why a norm (to φ or not to φ) holds, and that such an explanation appeals to facts that speak for or against a certain type of action (i.e. φ-ing). For instance, it is the facts that being hit hurts children, and that it teaches them that violence is acceptable, which speak against hitting children. It is in virtue of these facts that it is wrong to hit children and that one ought not to do so. But even if all such facts, prima facie evaluative ones included, were ultimately natural facts, the following point would stand. A particular case of refraining from hitting one's daughter is justified because it is wrong to hit children. That normative fact constitutes the proximal justification of the parent's refraining from hitting their daughter, whereas the natural facts backing it provide the distal justification (similarly Schroeder 2009). The normative fact is a normative reason for parents to refrain from hitting their daughter, and one that can motivate them, even if they do not accept or are unaware of the distal justification of the norm.

We conclude that there is no principled problem with taking normative facts to be motivating reasons; the fact that it is wrong to hit children may be the reason for which the parents refrain from hitting their daughter. First, this calls for another way in which monists should be open-minded – by accepting normative facts as practical reasons. Second, it can be used to motivate a pluralism about reason explanations, one which makes room for both teleological and enkratic explanations.

According to teleology, actions are understood as goal-directed; they are explained by appeal to the outcome they aim at bringing about:

[C]hildren find actions intelligible in terms of fully objective reasons … They conceive of Maxi's reasons in terms such as these: ‘Maxi needs his chocolate. (Or: It is important, or desirable, that Maxi obtain his chocolate.) The way to get it is to look in the blue cupboard. So he should look in the blue cupboard.’ (Perner and Roessler 2010, 205)

In going to the blue cupboard Maxi aims at getting the chocolate. This valuable outcome is his goal; going there is an effective means of achieving it. That it is good to get the chocolate is a normative reason which at the same time motivates Maxi to go to the cupboard. This gives rise to the teleological scheme of action explanation:

(TE) A φ-s for the reason that it is good to ψ and φ-ing is a means to ψ. It is good to ψ and φ-ing is a means to ψ-ing, so A φ-s.

To the extent that teleologists claim that every proper reason explanation is teleological, they are committed to the monist claim that every reason explanation conforms to this scheme.⁸

Normative facts appear to defy the teleological scheme, since they naturally fit into an enkratic explanatory scheme. Enkratic reason explanations have the form

(EE) A φ-s for the reason that she ought to φ (that it is right to φ, that it is a rule that one φ, etc.). A ought to φ (it is right to φ, it is a rule that one φ, etc.), so A φ-s.

Julie puts her shoes in the shoe cabinet for the reason that it's a rule at her daycare. She does not desire to do so; if anything, she desires to drop her shoes where she stands. Nor is the fact that it's a rule that children put their shoes in the cabinet a means for satisfying a separate desire she has. The reason explanation instead has an enkratic structure: It is a rule that all children put their shoes in the cabinet. So Julie puts her shoes in the cabinet.

In response, the teleologist can insist that it is possible to fit reason explanations appealing to normative facts into the teleological structure – the supposed problem for teleological explanations from normative facts is based on a literal, everyday conception of ‘desire’ or ‘goal’. Even an agent who does not desire to follow the rule (in an everyday sense of that expression) can have a mere pro-attitude towards it. Alvarez (2010, 55) helpfully distinguishes between a ‘full’ and a ‘thin’ sense of ‘want’. The full sense is ‘associated with preference and likes, and contrasted with, for example, duty or requirement’. Call this want_f. By contrast, the thin sense ‘includes anything that is willed’ – call this want_t. Even when normative facts conflict with an agent's preferences – and so with what she wants_f – they may still be in line with what she wants_t. For this it suffices that she knowingly conforms to a norm in φ-ing, as when Julie willingly puts her shoes into the cabinet so as to do what the daycare's rule prescribes.

Accepting this thin sense allows the teleologist to claim that all reason explanations presuppose that the agent pursues some goal in action, or wants_t something and acts in order to attain it (see Alvarez 2010, 173 quoted in sct. 3). The suggestion is that when a subject φ-s because she ought to φ, she φ-s in order to conform with a norm, or with the goal of doing what she ought to do. There is something the agent aims at – or wants_t – in acting. However, as this is only wanting_t, we can at the same time allow that she has no preference for φ-ing and actually acts against her goals or wants_f.

This response fits acting on a normative fact into the teleological explanatory scheme: explaining or predicting an action by appeal to a normative fact is not different from explaining it by appeal to any other goal or desirable outcome – any such explanation relies on the same means-end structures. Where in one case, what the agent aims at is to acquire the chocolate, and the efficient means is to go to the cabinet, in another case, what the agent aims at is conforming to the daycare's rules, and the means is to put her shoes in the cabinet. This option accommodates acting on normative facts within the teleological scheme of action explanation; it thereby avoids a pluralist commitment to enkratic reason explanations in addition to teleological ones.

However, a ‘one size fits all’ model of reason explanations is problematic. First, enkratic reason explanations (taken as a distinct kind of explanation) reveal the action as an action rather than a mere behavior no less than teleological explanations do. Further, the developmental evidence suggests that humans pick up many different strategies for explaining and predicting action (including habits, competences, goals, norms, desires, beliefs and others, for references see above). From the perspective of empirical research, there is no premium on trying to force all these strategies into the teleological mold; so why should it be important to do so with explanations appealing to normative facts? In this vein, Clément, Bernard, and Kaufmann (2011) argue that it is less cognitively demanding to explain or predict someone's action by appeal to normative facts than by appeal to her mental states; their studies indicate that children are able to predict an agent's behavior on the basis of rules before they possess a full-fledged concept of belief. And a study by Kalish and Shiverick’s (2004) shows that five-year-olds are more willing to predict another's action on the basis of rules than on the basis of her preferences.

To be sure, even if children don't explicitly refer to mental states in explaining or predicting an agent's behavior, they may still tacitly presuppose that she is aware of the rule and wants_t to conform with it (Kalish 2006; Kalish and Shiverick 2004). That the agent is aware of the normative fact has to be assumed anyway since we are concerned with reason explanations, which appeal to considerations in the light of which she acted.

However, this is no grist to the mill of monistic teleology. Au contraire! Consider the reality of practical deliberation. The agent's considerations – her motivating reasons – are phrased in terms of normative facts alone. They are not phrased in terms of the agent wanting_t to abide by the norm. And the agent would positively disavow that her reason is that she wants_f to abide by the norm.

When Julie puts her shoes in the cabinet because the rule is that children put their shoes there, that rule itself is what persuades her to do so; it is what shows up in her deliberation. But according to monistic teleology, the deliberation should be something like ‘It's desirable/it's my goal to conform to the daycare's rule to put one's shoes in the cabinet. The way for me to reach that goal is to put my shoes in the cabinet.’ (Cf. the quote from Perner and Roessler [2010] above.) In this deliberation, the fact that it is a rule that children put their shoes in the cabinet does not show up as a motivating reason! But this conflicts with what motivates the agent, from her own perspective. It is the normative fact that children ought to put their shoes away which makes doing so attractive in Julie's eyes – on the assumption of certain goals of hers, to be sure, yet that does not distinguish the case from descriptive motivating reasons. This removes the rationale for insisting that the only correct way to explain her action is by appeal to her (alleged) goal of conforming with the rule, to which she chooses the corresponding means of putting her shoes in the cabinet. Rather, in some cases the reason for which an agent φ-s is simply that it is a rule that she φ-s, or that she ought to φ.

Teleologists might complain that we have saddled them with a picture of practical deliberation that they do not, as a matter of fact, endorse. Their picture of teleological reasoning is Anscombe's. It starts with a premise detailing a respect in which an object is actually desirable for the reasoner, but without stating that the object is wanted or that attaining it is the reasoner's goal. The wanting is instead a background condition that enables the reasoning to terminate in action (Anscombe 1962, 66).⁹ Regarding the Julie case, her premise ‘It's a rule that children put their shoes in the cabinet’ gives the thing wanted (children putting their shoes in the cabinet) under a desirability characterization (that it's a rule). That Julie actually wants_t to put her shoes in the cabinet, under that description, is what moves her to do so. On this reading, Julie's goal is not to conform with the norm as we suggested earlier. It is (for the children, including her) to perform the action. In Anscombe's terms, what motivates Julie is a forward-looking rather than a backward-looking motive. Perhaps we should think of Julie as ‘attracted by something ostensibly good that is seen to require implementation’ rather than ‘prompted by an ostensible fact that is seen to call for a response’ (Müller 2011, 245, fn. 5).

Is this convincing, however? In our view, it is infelicitous to describe Julie as attracted by something good that she aims to implement – viz. the children, including her, putting their shoes in the cabinet – when she puts her shoes in the cabinet in response to the rule. The more natural description is: Julie responds to a normative fact that calls for her to put her shoes in the cabinet; she is an enkratic agent. In other words, where the rule motivates and thus explains her action, what explains it is not that she puts her shoes in the cabinet in order to do something that attracts her. Instead, she puts her shoes in the cabinet in response to the daycare's rule.¹⁰

This suffices to motivate pluralism about reason explanations. Not all such explanations fit the teleological model. Reason explanations from normative facts are better captured by the enkratic model, which depicts agents as responding immediately to normative facts.

5. Concluding remarks

We have argued that, with respect to motivating reasons and the action explanations based on them, a monistic version of objectivism is less attractive than a pluralist one. States of affairs, (non that-ish) goals, and normative facts should be acknowledged as motivating reasons to which subjects can appeal in explaining the actions of others. This leads objectivism away from austere monism towards open-minded monism and then on to pluralism about practical reasons. Moreover, since some reason explanations fit an enkratic rather than a teleological model, we are driven towards a pluralism about reason explanations. This chimes with the current trend towards pluralism in cognitive science with respect to folk psychology.

Vice versa, the lesson to be drawn for cognitive science from the theory of action is twofold. On the one hand, objectivist interpretations of folk psychology, such as teleology, should be regarded as concerning reason explanations of action. This leaves room for the possibility that there are other strategies for explaining and predicting the behavior of others, including those employed by neuroscience. On the other hand, we recommend accepting that there are folk psychological reason explanations that go beyond teleology, thus making the case for a more expansive pluralism about folk psychology.

Acknowledgements

We are grateful for feedback from an anonymous referee and for questions and comments from audiences in Belgrade, Fribourg, Oslo, Zurich, at ECAP10, and at the Zoom Epistemology Group.

Disclosure statement

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

Additional information

Funding

This research was supported by the Volkswagen Foundation grant Az 98510 (‘Explainable Intelligent Systems’), Swiss National Science Foundation grant #5100019E_177630 (D-A-CH ‘The Nature and Development of our Understanding of Actions and Reasons’), and Swiss National Science Foundation Grant #51NF40_180888 (NCCR ‘Evolving Language’); Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung.

Notes on contributors

Hans-Johann Glock

Hans-Johann Glock is Professor of Philosophy and Head of Department at the University of Zurich (Switzerland). He works in the philosophy of language, the philosophy of mind, the history of analytic philosophy and the theory of action. His publications include What is Analytic Philosophy and The Blackwell Companion to Wittgenstein, co-edited with John Hyman.

Eva Schmidt

Eva Schmidt is Assistant Professor of Theoretical Philosophy at TU Dortmund. She works in epistemology, the philosophy of action, and philosophy of mind. Her main research interests are epistemic reasons and reasons for action, explainable artificial intelligence, and the epistemology and nature of perception. She is the author of Modest Nonconceptualism: Epistemology, Phenomenology, and Content.

Notes

1 Both ‘intention’ and ‘goal’ can signify a mental state of, respectively, intending or aiming at something, or what the mental state is directed at. We are concerned with the second reading.

2 This understanding of a state of affairs is inspired by Wittgenstein's Tractatus, while dropping the metaphysical baggage of logical atomism. States of affairs are simply potential facts signified by meaningful noun-clauses of the form ‘that p’. If it is true that p, the possible state of affairs that p obtains; if it is false that p, it does not.

3 We thank an anonymous referee for this journal for pressing us on this point.

4 By the same token, the arguments in Fara (2013), Thompson (2008), Hornsby (2016), and Campbell (2019) for non-that-ish desires and intentions can be extended to non-that-ish goals.

5 Thus Alvarez (2010, 40) accepts that goals differ from both facts and states of affairs, while insisting that they cannot be reasons.

6 There are important differences between these normative phenomena (von Wright 1963; Wedgwood 2007). We shall disregard them, except where they impinge directly on our discussion.

7 This and related worries are discussed by Chappell (2019). They were also raised in discussion by Wlodek Rabinowicz.

8 This position reflects the Humean/Davidsonian idea that acting on mere (beliefs about the) facts is impossible; there needs to be something valued by the agent which motivates her. Perner and Roessler (2010) and Perner, Priewasser, and Roessler (2018) are naturally read as endorsing that all proper reason explanations are teleological. The same goes for Gergely and Csibra (2003), who suggest that Humean reason explanations are nothing but a psychologistic extension of prior objectivist teleological explanations.

9 We thank an anonymous referee for raising this worry.

10 Interestingly, Anscombe (1981, viii) abandoned the ‘struggle to treat all deliberate action as a matter of acting on a calculation how to obtain one's ends’, with explicit reference to actions that are prima facie motivated by rules. Our claim that the reason for an agent's action can be a norm also has affinities with Thompson's idea that internalized rules of a social practice can shape dispositions for individual action (2008, chs. 9–11).