Why should the UN intervene in this international crisis? Why did theAncient Egyptians mummify their dead? Should Huck Finn have helped Jimescape and, if so, why? Why is she selling her car? What shall we dothis evening? Questions like these that explicitly or implicitly askfor reasons, specifically reasons for action, are ubiquitous. Mostcontemporary philosophers who have sought to understand the nature ofreasons for acting start by distinguishing two kinds:“normative” reasons—that is, reasons which, veryroughly, favour an action; and “motivating”reasons—which, again roughly, are the reasons for which someoneacts. There are, in addition, reasons that explain an action withoutfavouring it or having motivated the agent.
A clear understanding of reasons for acting in their favouring,motivating and explanatory functions is of relevance to the philosophyof action, ethics, political philosophy and the philosophy of law. Theessential issues about these reasons: what they are, and how theyrelate to each other and to actions, are of wider concern.
This entry examines the various accounts that philosophers have givenof these different kinds of reasons and their interconnections, aswell as the disagreements among them about these matters. The focuswill be on reasons for acting—commonly called “practicalreasons”. For the most part, the entry will leave asidequestions about reasons for responses other than actions—forinstance, reasons for believing, wanting, and feeling. This is not todeny that a central question in the theory of reasons concerns thepossibility of a unified account of reasons that comprises reasons forany kind of response.
People engage in practical reasoning: they deliberate about what to doand how to do it. And they often act in light of or guided byreasons—which can then feature in an explanation of theiractions and, sometimes, justify them. These ideas go back to Plato(Protagoras andRepublic, Book 4) and Aristotle(Nicomachean Ethics, Bks VI and VII, andDe Anima,III.10; see Price 2011) and they have been a constant theme in thehistory of philosophy. In the eighteenth century, David Hume andImmanuel Kant offered radically different views on the role andimportance of Reason (the faculty of reason) in guiding and justifyingour behaviour. Their contributions remain influential today, but inthe past sixty years, the focus of discussion has shifted from thefaculty of reason toreasons: roughly, considerations thatguide or justify our actions. In the philosophy of mind, interest inreasons arose from the thought that intentional actions are done forreasons—a view connected to Elizabeth Anscombe’sIntention (1957) and explicitly defended by Donald Davidsonin his influential paper “Actions, Reasons, and Causes”(1963). In the field of ethics and normativity, broadly understood,interests in reasons arose especially from questions about theauthority of morality (as in, e.g., Nagel 1970; Foot 1972, Williams1979), as well as from connections between reasons and notions such asjustification, obligations, excuses, rationality, and moralworth—among others.
As mentioned in the introduction, contemporary authors tend todistinguish between “normative” and“motivating” reasons. Jonathan Dancy (2000), who discussesthe history of this distinction, notes that this contemporaryclassification doesn’t always map neatly onto their allegedancestors, for instance, Francis Hutcheson’s (1728). Whateverits history, the distinction is now accepted by most if not allcontemporary philosophers who write on this topic (representativeexamples include Raz 1975; Smith 1994; Scanlon 1998; Dancy 2000;Schroeder 2007; Alvarez 2010; Parfit 2011; Markovits 2014; Lord 2018;McHugh & Way 2022).
A normative reason is a reason (for someone) to act—in T.M.Scanlon’s phrase, “a consideration that counts in favourof” someone’s acting in a certain way (1998: 17). Amotivating reason is a reason for which someone acts—a reasonwhich guides their action. Since we can often explain what a persondoes by citing their reasons, motivating reasons are sometimes alsocalled “explanatory reasons”. However, recent authors tendto distinguish motivating and explanatory reasons and there are goodgrounds for doing so (seesection 3).
How should this distinction be understood? On an influential view(Baier 1958; Dancy 2000; Hieronymi 2011), this is just a distinctionconcerning different things reasons can do. Dancy writes:
When I call a reason “motivating”, all that I am doing isissuing a reminder that the focus of our attention is on matters ofmotivation, for the moment. When I call it “normative”,again all that I am doing is stressing that we are currently thinkingabout whether it is a good reason, one that favours acting in the wayproposed (Dancy 2000: 2–3).
The thought here is that talk of different kinds of reasons can beunderstood in terms of the different roles that reasons can play.These different roles become apparent when we attend to differentquestions we can ask about reasons. For instance, we may ask whetherthere isa reason for the Chancellor to introduce a“sugar tax” for drinks and ask also whattheChancellor’s reason for introducing such a tax was. Thefirst is about reasons that favour actions (normative), the secondabout reasons that guide agents (motivating). Note that the samereason may answer both questions. For example,that the tax willreduce child obesity may be both a reason for the Chancellor tointroduce a “sugar tax” (normative reason) and theChancellor’s reason for doing so (motivating reason). But peopledon’t always act for the reasons that favour their doingsomething—even when they do what the reasons that apply to themfavour. A corrupt Chancellor may be motivated to introduce a sugar taxbecause it will benefit her husband’s low-sugar drinkscompany—but that is not a normative reason for her, asChancellor, to introduce that tax. If this way of understanding talkabout different kinds of reasons is right, perhaps the picture is morecomplex than the dichotomy of “normative vs. motivating”reasons suggests—as will be discussed insection 3.
A seemingly stronger, although consistent, interpretation of thedistinction treats normative and motivating/explanatory reasons asentities of different ontological kinds. Accordingly, normativereasons are objective and mind-independent: they are facts, or truths,or “worldly” things, such as states of affairs, etc.;motivating reasons, by contrast, are subjective and mind-dependent:mental entities such as mental states of agents—or theircontents. These issues will be explored further below.
The following section focuses on normative reasons.Section 3 focuses on motivating reasons andsection 4 discusses the relation between normative and motivating reasons.
A reason is said to be a “normative reason” for someone toact because it favours their so acting: it supports, or makes a casefor, or helps justify, that course of action. More can be said aboutwhat this amounts to by focusing on two roles normative reasons canplay.
First, normative reasons play adeontic role: they explain anaction’s deontic status for someone—that is, whether, allthings considered, she ought, must or may do that thing. If a reasonfavours my doing something, then I have apro tanto reason todo it: it is to that extent right for me to do it. But there may alsobe a reason against my doing it: apro tanto reason not to doit. The fact that a joke is funny may be a reason to tell it; but thatit will embarrass someone may be a reason against. So, I have a reasonto tell the joke and a different reason not to tell it. Whether, allthings considered, I ought to tell the joke depends on whether eitherof the reasons “outweighs” or otherwise“defeats” the other: if the reason for telling the joke isoutweighed by the reason against it, then I ought not tell the joke.If neither defeats the other, then, Imay tell it—butit’s not the case that I ought to tell it.
Second, normative reasons play adeliberative role: they arethings it’s appropriate to pay attention to in deliberation.When considering whether to tell the joke, I should take into accountboth that it’s funny, and that it’ll embarrass someone.And if the latter reason outweighs the former, and I deliberate well,I’ll decide not to tell the joke, precisely because it willembarrass someone. (In that case, this consideration will be mymotivating reason. Thus, the deliberative role of normativereasons makes them potential motivating reasons. More on this insection 3).
Can we say more about what exactly normative reasons are, given thesetwo roles they can play? This question raises two issues: (i) anontological one about what sort of entities can be normative reasonsfor action; and (ii) a constitutive issue about what it is for anentity of the relevant sort to favour a particular course of action.We address them in turn, before discussing some further issues aboutnormative reasons.
When it comes to ontology, ordinary language is liberal. We say thatyour upcoming birthday, or the weather, or your wish to see redsquirrels, is a reason to visit the Isle of Wight. This might suggestthat events, things, and states of mind, among other kinds, can all benormative reasons. But such reason claims can be rephrased using“that” clauses. For instance, that the weather will benice, that it’s your birthday, or that you want to see squirrelscan be a reason to visit the Isle of Wight (Schroeder 2007:20–21). Moreover, only what is the case can favour or justifyactions: if the weather won’t be nice on Sunday, then theweather doesn’t support visiting then. Consequently,philosophers tend to hold that normative reasons arefacts(Raz 1975; Scanlon 1998; contrast Fantl & McGrath 2009; Gibbons2010; Howard 2021).
This apparent consensus is complicated by two issues. First, somephilosophers distinguish between “objective” and“subjective” (or sometimes “apparent”)normative reasons. Roughly, objective reasons are the facts that countfor or against acting: considerations an informed adviser would takeinto account. Subjective reasons are things you believe and that mightmake it reasonable to act. Since you might act reasonably on falsebeliefs, subjective reasons cannot be facts (Schroeder 2008; Parfit2011; see also Fogal & Worsnip 2021). This entry will continue touse “normative reason” to mean “objective normativereason” unless otherwise indicated.
The second complication is that there is disagreement about what factsare: are they concrete or abstract entities? Is a fact the same as thecorresponding true proposition, or is it the “truth-maker”of the proposition? Among those who hold that normative reasons arefacts, some identify reasons with true propositions (Darwall 1983;Smith 1994; Scanlon 1998; Alvarez 2010; Setiya 2014; Lord 2018; orwith “the truth of a proposition” (Hyman 2015). Othersreject this view; for example, Dancy (2000, 2018) and Mantel (2018) doso on the grounds that propositions are abstract and representational(they represent the way the world is) but, they say, reasons must beconcrete and non-representational (they are “ways the worldis”). These problems are complex and have many ramifications.For present purposes, this entry will ignore them and continue tospeak of normative reasons as facts.
The second issue is what it is for an entity of the relevantontological category to favour an action. Given our assumption thatnormative reasons are facts, the issue is thus what it is for a factto be a normative reason for a particular agent to undertake aparticular course of action? What is it for, say, the fact that thejoke is funny to “favour” my telling it? This questionasks for an account of the relation that obtains between a fact, a wayof acting, and an agent, when that fact is a reason for that agent toact in that way. (A relation that may have further relata—e.g.,time and circumstances [Skorupski 2010; Scanlon 2014]).
Some hold that no informative answer to this question is possible: theidea of a reason is primitive. Although we may offer metaphors, suchas that reasons “count in favour” of actions, thesemetaphors don’t explain reasons in more basic terms (Hampton1998; Scanlon 1998, 2014; Dancy 2004; Skorupski 2010; Parfit 2011).Others disagree. Very broadly, we can distinguish three (not mutuallyexclusive) groups of views among the latter.
The first group aims to characterise reasons in terms of their deonticrole, i.e., their capacity to determine the deontic status of actions.(This approach has roots in W.D. Ross’s characterisation ofprima facie duties—roughly, moral reasons—asfeatures that make an action obligatory in the absence of conflictingduties (1939: 19; for discussion see Dancy 2004: 18–20)). Themost prominent contemporary version of the view is due to John Broome(2013: ch. 4). Broome begins with the idea that a reason is anexplanation of why you ought to do something. However, this seems notto allow that reasons can be defeated (though see Nebel 2019). Broomethus introduces the idea of a “weighing explanation”: anexplanation of an act’s deontic status that appeals toconsiderations for or against the action. Broome proposes that areason to act is a consideration that plays the “for” rolein such an explanation (2013: 53; for discussion see Brunero 2013,2019).
The second group of views aims to characterise normative reasons interms of their deliberative role: to be a reason is to be a fact thatit is appropriate to consider in deliberation, or on the basis ofwhich it is right to decide what to do. Bernard Williams can be readas endorsing such a view in proposing that a subject has a reason toact only if he could reach the conclusion to so act “by a sounddeliberative route from the motivations he already has” (1989:35, see also Williams 1979). Although Williams developed this idea ina broadly “Humean” way (see below), the core idea ofunderstanding normative reasons in terms of their deliberative role isindependent of these commitments. For instance, one may hold that areason is a fact you would be motivated by in so far as you arerational (Korsgaard 1986), or one for which your rational counterpartwould want you to act (Smith 1994; Manne 2014) or a premise of goodreasoning (Setiya 2014; Way 2017; McHugh & Way 2022). For furthervariants, see, e.g., Street 2008; Markovits 2014). Kearns andStar’s (2008, 2009) proposal that a reason to act is evidencethat you ought to so act can also be understood in this way, if wetake practical deliberation to aim to figure out what you ought to do(Star 2015, 2018a; see also Raz 1978: 5ff; Whiting 2021: ch.2).
A third group of views understands normative reasons in terms of arelation between reasons and certain types ofends.“Desire-based” versions of this view, which takeinspiration from Hume’s remarks about reason and the passions,take the relevant ends to be the intentional objects of desires. On asimple desire-based view, a reason for you to act is a fact that helpsexplain why acting that way would achieve something you desire. Forinstance, the fact that there will be dancing at the party is a reasonfor Ronnie to go to the party, because it helps explain why going tothe party will help Ronnie to do something he wants todo—namely, dancing (Schroeder 2007; see also Williams 1979,1989; Alan Goldman 2009; Markovits 2014; Manne 2014, 2016).“Value-based” versions understand reasons in terms of arelation to good, or valuable, ends. This view is associated withAristotle who links the right in practical reason with what isconducive to the good (Nicomachean Ethics) and was prevalentamong medieval philosophers (e.g., Thomas Aquinas,SummaTheologiae, 1a, q.82). A simple formulation takes reasons to befacts that explain why acting in a certain way would achieve somevaluable end (Finlay 2014; Maguire 2016). Other versions invoke otherrelations to the valuable, such as respect for it (see also Anderson1993; Raz 1999, 2001, 2011; Wedgwood 2017; Sylvan 2021. For a recentdiscussion see Kiesewetter 2022).
These different approaches to characterising normative reasons can beassessed in various ways: for instance, with reference to how wellthey accommodate the deontic and deliberative roles of normativereasons; or to their implications concerning what reasons we have inparticular cases. Each approach faces challenges on these and otherfronts, some of which are outlined below.
One very general issue worth noting is that an account of “thereason relation” should arguably be general: it should apply notjust to reasons to act but also to reasons to believe, want, and feel(e.g., Kearns & Star 2009: 219–22; Gibbons 2010; Way 2017:255). There doesn’t seem to be a difference in therelation we are invoking when we say, for instance, that thefact that there’s going to be an election is a reason both toregister to vote and to fear things are going to get worse. Rather,one fact stands in the same relation—of favouring orsupporting—to both an action and an emotion. So, the differentaccounts should be assessed according to how plausible they are, notjust as accounts of reasons to act, but of reasons for these othertypes of response too.
However we characterise normative reasons, they raise a host ofinteresting and important issues, which intersect with other areas ofpractical philosophy. This section will briefly outline three suchissues, and then say a little more about a fourth.
First, there are questions about the relationship between reasons andmorality. A traditionally central question is whether we all have goodreason to do as morality requires. This question is especiallypressing for desire-based accounts of reasons and for certain versionsof the view that reasons should be understood in terms of theirdeliberative role. For it certainly seems possible that doing whatmorality requires may not serve any desire a person has and may not beanything to which there is “a sound deliberative route from themotivations he already has” (Williams 1989:35; see entry onreasons for action: internal vs. external). Beyond this issue, there are important questions about what, ifanything, is distinctive about moral reasons (see, e.g., Scanlon 1998;Darwall 2006; Southwood 2011). And there are questions about whetherthe theory of normative reasons has any implications for normativeethics (see, e.g., entry onmoral particularism and moral generalism).
Second, there are questions that intersect with discussions inmetaethics about how normativity fits with the non-normative world.For instance, can the relation of being a normative reason be analysedin naturalistic or descriptive terms? Do normative reasons depend onus in some way? Are judgements about reasons even in the business ofascribing properties at all? While questions of this sort havetraditionally been raised about moral properties—e.g., moralrightness and wrongness—many recent philosophers have come tosee them as general questions about normative facts. Accordingly, muchrecent metaethical work has focused not on morality but onnormativity, and sometimes on normative reasons in particular (seeentry onnormativity in metaethics).
Third, there are questions about how normative reasons play theirdeliberative role, such as: What makes for good deliberation, orreasoning? What is involved in responding to a normative reason? Whatkinds of evaluations turn on the normative reasons we act for? Thesequestions take us into moral psychology and action theory. Some ofthem are explored insection 3 andsection 4.
Fourth, there are questions about how normative reasons play theirdeontic role, that intersect with questions in normative ethics. Howdo the various reasons for and against a course of action combine todetermine whether someone ought or may undertake it? A simple proposalis that whether someone ought to act in a certain way depends on howall the reasons for and against so acting weigh up in comparison toall the reasons for and against the alternatives (see, e.g., Berker2007; Schroeder 2007). But this already raises a host ofquestions.
To begin with, is it really the case thatall the reasonsbearing on an act are relevant to determining its deontic status?Above we said that normative reasons are facts. The combination ofthese two ideas leads to some surprising results. Suppose, to take awell-known example of Judith Thomson’s (1990: 299), that due toan extraordinary and unforeseeable coincidence, flipping the lightswitch when you get home will cause a small lightning flash in yourneighbour’s home, badly burning them. This fact seems tostrongly count against flipping the switch. So, all else being equal,you ought not flip the switch. However, since the harm to yourneighbour is entirely unforeseeable, many people (though notably, notThomson; see also Graham 2011) find this conclusion hard toaccept.
In light of such concerns, some philosophers endorse an epistemicconstraint, either on normative reasons (Dancy 2000: 56–9;Kiesewetter 2017; Markovits 2010: 219), or on the reasons relevant todeontic status (Setiya 2014; Lord 2018). On this“perspectivist” view, a fact must fall within yourepistemic perspective—you must know it, or perhaps be in aposition to know it, or reasonably believe it, etc.—if it is tobear on what you ought to do. Others distinguish between types ofdeontic status. In Thomson’s example, it seems plausible that,relative to all the facts, you ought to refrain from flipping theswitch; but, relative to the facts you know, or to your beliefs, youmay flip the switch. These different verdicts play different roles inour thought and practices: for instance, what we ought to do relativeto all the facts may be what we aim to discover through deliberationand advice, while what someone ought to do relative to what they know,or perhaps believe, is more closely tied to assessments ofrationality, and merit and culpability (Schroeder 2008). Given thesedistinctions, the deontic role that reasons play should be refined:objective reasons determine deontic status in the formercontexts, whilesubjective reasons determine it in the second(Schroeder 2007; Parfit 2011; Fogal & Worsnip 2021).
However, these issues are settled, reasons’ deontic role raisesfurther, more fundamental, questions. How exactly do the relevantreasons determine deontic status? In many cases, it is natural toappeal to metaphors of weight and strength: my reasons to complete myreport are stronger than my reasons to take the day off and enjoy thesunshine, so I ought to keep working. The amusement the joke willbring is outweighed by the embarrassment it will cause, and so I oughtnot tell the joke. What should we make of these metaphors? They seemto have limitations. For instance, reasons are claimed not to combinein a simple additive way, or to vary with context, or to be sometimesincomparable (see, e.g., Dancy 2004, 2018; Bader 2016; Cullity 2018;and the entry onincommensurable values). An important question is thus how an account of reasons’weights can make sense of those observations (Schroeder 2007: ch.7;Horty 2012; Nair 2021; McHugh & Way 2022: ch.5).
Another worry is that the metaphor of weight and strength isinappropriate in portraying the relation between some reasons. Forinstance, a longstanding idea is that some moral reasons—say,that so acting would be taking an innocentlife—“silence”, rather than “outweigh”competing reasons (McDowell 1978; for a related idea see Raz 1975,1989, and Raz 2021 on “exclusionary reasons” in the OtherInternet Resources). More generally, it is not clear how the proposalencompasses the full range of considerations that can bear on deonticstatus—for instance, obligations to others, rights, commitments,ideals. Finally, the proposal builds in “maximisation”: weought to follow the weightiest reasons. But very often it seems okayto do less than the best—for example, when the best would besupererogatory, or an alternative would be good enough. Nor is itclear how to accommodate a distinction between what we ought to do andwhat we must do. Such concerns have led some to refine the simpleproposal, or pursue alternatives (Raz 1986; Gert 2004, 2007; McElwee2011; Lord & Maguire 2016a; Snedegar 2016, 2018; Maguire &Snedegar 2021; Tucker 2022; Whiting 2021: ch. 3).
Motivating reasons are the reasons for which we act—the reasonsthat motivate us. Motivating reasons can thereby explain our actions.However, motivating reasons should not simply be equated with thereasons that explain our actions.
Consider the behaviour of Othello in Shakespeare’s play of thesame name. Othello kills Desdemona guided by the belief, induced byjealous Iago, that she has been unfaithful to him. The tragedy of theplay lies in the fact that Desdemona is innocent: she loves Othelloand is faithful to him. Clearly—and putting aside whetherinfidelity could ever favour murder—there is inOthellono reason that favours the murder: no normative reason for Othello tokill Desdemona, since she is not unfaithful. Still, there are twothings that seem true about Othello’s killing of Desdemona. Oneis that he is motivated to kill Desdemona by the (putative) fact thatshe has been unfaithful. The other is that his action can be explainedby citing the fact that he believes that Desdemona has beenunfaithful. So here we seem to have two quite different reasons: onethat motivates Othello—the (putative) fact that Desdemona hasbeen unfaithful; and one that explains his action—the (actual)fact thathe believes that she has been unfaithful. We candistinguish, that is, between the reason that explains Othello’saction (the fact that he believes that Desdemona has been unfaithful)and the reason, or apparent reason, that motivates him to act(Desdemona’s alleged infidelity itself).
Accordingly, we should distinguish in general (at least) three kindsof reasons: normative, motivating and explanatory, corresponding towhether a reason favours, motivates, or explains an action. This sitswell with the suggestion (insection 1) that the distinction between normative and motivating reasons isdrawn on the basis of the different questions that reasons can answer.For there seem to be at least three distinct questions about therelation between reasons and actions. There are questions aboutwhether there is a reason that favours someone’s action(normative); questions about the considerations—(putative)facts—in light of which someone act (motivating); and alsoquestions about what reasons explain an action (explanatory). Thisthreefold classification is explicitly accepted and defended byvarious authors (Baier 1958; Alvarez 2007, 2009, 2010; Hieronymi2011); and it is implicit, sometimes using different terminology, indiscussions by others (Smith 1994; Scanlon 1998; Mantel 2014).
This use of “the motivating reason” for the consideration“in light of which” someone acts is of course somewhatstipulative.
First, talk of “the agent’s motivating reason”typically involves some simplification. For one thing, a considerationin light of which one acts will figure in deliberation in combinationwith other considerations. For example, that I won’t have timeto hoover later may motivate me to hoover early in the morning only incombination with other considerations, for instance, that the houseneeds hoovering today. My reason then is, arguably, a combination ofat least two considerations: that the house needs hoovering today andthat I won’t be able to do it later. This point is relevant toquestions about the relation between normative and motivating reasons(seesection 4). Moreover, an agent may be motivated to act by more than oneconsideration in a different sense: I may hoover the house early inthe morning both because I won’t have time to do it laterand also because it will annoy my inconsiderate neighbour.Alternatively, I might take myself to have those two reasons to hooverearly but act motivated only by one of them (this point is central toDavidson’s (1963) influential argument that motivating reasonsare causes of intentional actions). Finally, I may consider somethingthat counts against acting, for instance, that hoovering early willalso disturb my other, very considerate neighbour. If I still decideto hoover early, I do not act for that “con reason” but,arguably, I am still guided by it in some sense, if I give it someweight in my deliberation (Ruben 2009).
Talk of motivating reasons is stipulative also in excluding somethings that might reasonably be assumed to be motivating reasons. Forinstance, someone’s goal or intention in acting (to growvegetables, to become rich) are motivating factors but, because theyare not considerations in light of which one acts, do not fall underthe category “motivating reasons” as here understood (butsee Audi 1993; Howard 2021). Similarly, a state of desiring (such aswanting to have one’s revenge), or a motive or emotion (forinstance, jealousy) can be states “that encompassmotivation”, to use Mele’s phrase (2003). Nevertheless,they are not motivating reasons as understood here. Rather, theconsiderations (if any) in response to which the person desires, orfeels the emotion, will thereby be the person’s motivatingreasons. To continue with our example, Othello’s jealousy anddesire to kill Desdemona is based on the thought that she isunfaithful to him. This consideration is Othello’s reason bothfor wanting to kill her and his reason for doing so.
But what exactly is a motivating reason, thus understood? What kind ofentities can be considerations in light of which someone acts? Andwhat kind of relation must hold between an agent and an entity of therelevant kind for it to be the reason for which she acts?Section 3.1 addresses the former question, whilesection 3.2 examines the latter.
In the literature on the theory of action until the turn of thetwentieth century, the dominant view was that motivating reasons aremental states of the agent’s—a view sometimes called“psychologism”. Davidson 1963 is often cited as thelocus classicus of psychologism, and the view was alsoinfluentially defended in Smith 1994: ch. 4; see also Wallace 2003,Pryor 2007, Turri 2009, Gibbons 2010. Davidson characterises anintentional action as an event—a bodily movement—caused“in the right way” by a combination of two mental states:a “pro-attitude” (roughly, a desire) and a belief.Davidson called this combination of mental states the agent’s“primary reason” (1963: 687).
Psychologism is appealing. First, when an agent acts for a reason, heacts motivated by some end towards which he has a pro-attitude andguided by a belief about how to achieve that end. To return to ourexample, Othello murdered Desdemona motivated by the desire to defendhis honour and guided by his belief that the latter demandedDesdemona’s death, given her unfaithfulness. Second, for areason to motivate, you must have the right sort of “epistemiccontact” with that reason: you must know or believe theconsideration that constitutes that reason. Both these points appearto support the view that the reasons that motivate are agents’mental states.
However, these arguments don’t suffice to establishpsychologism. In fact, there are forceful arguments against theposition. A simple point, highlighted in the Othello example, is thatreflection on the considerations that agents take as reasons foracting, and on what they typically give and accept as their reasons,counts against psychologism. As Othello deliberates about what to do,even while in the grip of jealousy, his reasoning does not includeconsiderations about whether he believes this or that but rather aboutwhat Desdemona has or has not done. The things that Othello considersand responds to are not his mental states but rather facts, or allegedfacts, about the world around him, in particular about Desdemona.
Another important argument turns on the idea that any account ofmotivating reasons must meet what Dancy (1995; 2000) calls “thenormative constraint”:
This [normative constraint] requires that a motivating reason, that inthe light of which one acts, must be the sort of thing that is capableof being among the reasons in favour of so acting; it must, in thissense, be possible to act for a good reason (2000: 103; see also Raz2011: 27).
Dancy’s point is that psychologism fails to meet thisconstraint—in fact, he says, psychologism has the consequencethat “the reasons why we act can never be among the reasons infavour of acting” (2000: 105). Why? In order to act for a goodreason (in Dancy’s sense), we need to act for a reason that isor could be a fact—for Dancy, something that is or could be thecase. But psychologism does not meet this constraint because a mentalstate is not, not could it be, a fact. In short, Dancy thinks that youact for a good reason only if your motivating reason is a normativereason that favours your action, or it is at least a considerationwhich, if true, would or could favour it (see Heuer 2004 for a helpfulexplanation).
A variant of psychologism avoids this objection by claiming thatmotivating reasons are “psychological facts”—thatis, facts about the agent’s mental states. For instance, on thisview, Jo’s reason for running is the (psychological) fact thatshe believes that she’s late. This position appears to meetDancy’s constraint, at least formally: psychological factsare facts. But the view is unsatisfactory because, althoughwe are occasionally motivated by facts about our psychology—aswhen someone decides to see his doctor because he is convincedhe’s being pursued by the Security Services (Hyman1999)—most motivating reasons are not psychological facts: theyare (putative) facts about all sorts of things.
Alternatively, it may be argued that psychologism meets Dancy’snormative constraint because it holds that motivating reasons aremental states whosecontents correspond to normative reasons.It is a mistake to assume that meeting the constraint requires theidentity of normative and motivating reasons (Mantel 2014).
To determine the success of this response we need to disambiguate it.On the one hand, the response can be taken to argue that the reasonsthat motivate us are thecontents of our mental states ofbelieving. This meets the normative constraint (on this view, onecan act for a good reason) but this does not vindicatepsychologism because the contents of mental states are not themselvesmental states: they may be propositions, or facts, etc. If, bycontrast, the response is construed as simply asserting that a mentalstate with the right content can be a good reason for acting (inDancy’s sense), then it is not clear how this engages withDancy’s objection.
The above and other arguments (see, e.g., Stout 1996; Stoutland 1998;Hyman 1999, 2015; Raz 1999; Dancy 2000, 2008; Williamson 2000; Bittner2001; Schueler 2003; Hornsby 2007, 2008; Alvarez 2008, 2009, 2010;McDowell 2013) suggest that being motivated by a reason is not beingmotivated by, or acting in light of, or guided by, a mental state.Accordingly, psychologism is now a minority view.
There are currently two main alternatives to psychologism:“factualism” (motivating reasons are facts) and“propositionalism” (motivating reasons are true or falsepropositions).
Each face difficulties. For factualism, a central problem concerns“error cases”: cases like Othello’s, where the agentis motivated to act by a false consideration. In the example whatmotivated Othello, what he would give as his reason—thatDesdemona has been unfaithful—is false. And so, Othello cannotact in light of the fact that Desdemona has been unfaithful sincethere is no such fact. A possible response is that, in error cases,the agent is motivated by something that he treats as a reason butwhich is in fact merely an “apparent reason” (Alvarez2010).
Propositionalism, it seems,can accommodate this problem: inerror cases, the positions holds, agents act for a reason that is afalse proposition that the agent believes. So, in the example above,Othello’s reason is the (false) proposition about Desdemona,which he believes. Note his motivating reason is nothisbelieving that she’s unfaithful, which would bring us backto psychologism, butwhat he believes. According to thisproposal, then, Othello did act for a reason: a putative fact that theagent takes to obtain. The view is defended or endorsed by many,including Dancy (2000, 2008, 2014), Hornsby (2007, 2008), Setiya(2007), Schroeder (2008), McDowell (2013), Comesaña &McGrath (2014), and Singh (2019).
Against this, it has been argued that the proposal leads to paradox orinfelicitous claims. As Unger puts the point:
it is inconsistent to say “His reason was that the store wasgoing to close, but it wasn’t going to close”. (1975:208)
If so, the response that in error cases the agent’s reason is afalse proposition is problematic.
But is this anything other than a terminological dispute? After all,if some philosophers choose to call false propositions that motivate“motivating reasons”, while others choose to call them“apparent reasons”, that seems perfectly unobjectionable.The substantial issue behind this disagreement seems, then, whetherthe notion of a reason we apply in different contexts is a unifiednotion; and, if it is, what aspect or features of reasons, if any, areessential or central to that notion—for instance, that they candetermine deontic status, that they can explain (all sorts of things),or that they can be used in reasoning. If the former two are taken tobe central, then it follows that a false proposition cannot be areason. In contrast, the view that the central feature of reasons isthat they can be premises of reasoning allows for more latitude so asto include false propositions as reasons.
Motivating reasons are considerations in light of which we act.However we conceive of them ontologically, there are issues about whatit takes for an agent to act for a motivating reason.
Most accounts of acting for a motivating reason require as a conditionthat the agent be in some kind of epistemic relation to the reasonthat motivates her. One view is that this epistemic relation isbelief: for an agent to act for the reason thatp, the agentmust at least believe thatp. It is this thought that ledsome to psychologism.
More recently, some have argued that, at least when someone acts for areason that is also a fact, the agent needs toknow and notmerely believe the relevant fact, if they are to act in light of it.The view is defended by Unger (1975), Hyman (1999, 2011, 2015),Williamson (2000, 2017), Hornsby (2007, 2008), and McDowell (2013).The basic thought is that, if the agent does not know the fact, shecannot be guided by it (Hyman), or respond rationally to it(McDowell). The relationship between the agent’s acting as shedoes and the relevant fact will be fortuitous, a matter of luck orcoincidence. To illustrate, suppose someone believes on a hunch thatPegasus will win the Grand National, and so places a large bet. Evenif Pegasus does win, the person was not guided by the fact thatPegasus will win; they just got lucky (Hyman 1999: 447). The same maybe true when an agent acts on a belief that is both true andjustified. Suppose someone believes that it is 5pm because they glanceat what—unbeknownst to them—is a stopped clock, and soleave for their appointment. Even though it actually is 5pm, they arenot guided by this fact—they just got lucky. (This point extendsGettier’s (1963) famous arguments that justified true belief isnot sufficient for knowledge; see entry onthe analysis of knowledge).
These arguments remain contentious. For example, Dustin Locke (2015)uses so-called “fake-barn” cases to argue that we can actin light of a fact without knowing it. Suppose a man is driving in thecountryside and sees a (real) barn ahead. Unbeknown to him, he’sdriving in “fake-barn country”, which is littered withfake barns: barn façades designed to look like real barns. Itis widely held that, although the man is looking at a real barn, hedoes not know that it’s a barn (Alvin Goldman 1967).Nonetheless, Locke argues, he might, for instance, drive towards thebarn guided by the fact that it’s a barn. Thus, being guided bya fact does not require knowing it. For further views and discussion,see Hawthorne 2004; Brown 2008; Neta 2009; Dancy 2011, 2014; Lord2018; Hawthorne & Magidor 2018, Whiting 2021: ch.5.
As we have seen, motivating reasons need not be normative reasons: wedon’t always act for good reasons. However, an influential andattractive view is that they must betaken to be normativereasons. As Davidson (1963) puts it, when an agent acts for amotivating reason then “from the agent’s point of viewthere was, when he acted, something to be said for the action”(1963: 691; see also Darwall 1983: 32; Scanlon 1998: 23). The view isa relative of the ancient idea that all action takes place“under the guise of the good”—that whenever we actintentionally, we take what we are doing to be in some way good. Thisview is defended by Anscombe (1957), Stampe (1987), Quinn (1993), Raz(1999), Schueler (2003), and Oddie (2005), amongst others. Indeed, ifwe assume that taking there to be a normative reason to act involvestaking acting in that way to be good, the view embraces the guise ofthe good thesis. Nonetheless, this assumption may be questioned andsome claim that a “guise of reasons” view that doeswithout it has advantages over the guise of the good (Gregory2021).
There are important questions about how exactly these views are to beunderstood. For instance, what is it to “take” aconsideration to be a normative reason? On one natural view, to take aconsideration to be a normative reason is just tobelievethat it is a normative reason (Gregory 2021). Another view is thattaking a consideration to be a reason is a quasi-perceptual state ofthat consideration’sseeming to be a reason (compareStampe 1987; Oddie 2005; Tenenbaum 2007; Moss 2012), and yet othercontenders have emerged in the recent literature (Tenenbaum 2008;Schafer 2013; Singh 2019).
Much of the debate concerning the guise of reasons and the guise ofthe good has focused on putative counterexamples—cases in which,it is claimed, someone acts intentionally or for a reason withouttaking any consideration to count in favour of what they are doing. Totake a few representative examples, some claim that negative emotionssuch an anger or frustration might lead us to do what we recognise tobe in no way good, as when someone shouts at or destroys an“uncooperative” household appliance (Stocker 1979;Hursthouse 1991). Others claim that we can be motivated to act in wayswe take to be entirely bad, and indeed be so motivatedprecisely because we take the action to be bad—as whenMilton’s Satan declares “evil be thou my good”(Velleman 1992). And others suggest that animals and young childrenmight act for reasons, while lacking the conceptual sophistication torepresent their actions as good. Various responses have been offeredto these examples. For instance, some take cases of negative emotionsto undermine the view that we act only in ways webelieve tobe good, but not the view that we act only in ways that, at the time,appear good (Tenenbaum 2007). Others emphasise the varietiesof ways in which something can be good, or a reason—thus forinstance, while Milton’s Satan might be motivated by what hetakes to bemorally bad, he may still do what he takes to begood in other ways (Anscombe 1957; Tenenbaum 2007). And there has beenmuch debate on “expressive actions”, for example, onwhether actions done out of strong emotions are intentional or donefor reasons (see Bennett 2021 for an overview).
Since arguments by counterexample are often inconclusive, recentdiscussions have focused more on what role the guise of reasons, orthe guise of the good, might play in a satisfying account of what itis to act for reasons. Thus Setiya (2007) argues against these theseson the grounds that they play no role: he claims that the central taskfor an account of what it is to act for reasons is to explain ourdistinctive knowledge of our own actions, and that this can beprovided without any appeal to the guise of reasons, or the good.However, many other theorists have thought that there are indeedprincipled reasons for action theory to accept one of these theses.Perhaps the most important argument appeals to the idea that, instating an agent’s motivating reasons, one shows theagent’s action to be intelligible in a distinctive way—away characteristic of rational creatures. Proponents of this argumentsuggest that explanations of agent’s actions which don’treveal the way the agent took so acting to be good, or supported byreasons, fail to do this. To adapt a famous example of WarrenQuinn’s, someone who turns on a radio because of a brutedisposition to do so, without seeing anything to be said for doing so,seems not to be acting for a reason. Articulations of this idea in theservice of the guise of reasons, or the good, can be found in Anscombe1957; Davidson 1963; Quinn 1993; Scanlon 1998; for recent developmentsand discussions, see Copp and Sobel 2002; Gert 2003: ch.9; Schapiro2009; Setiya 2010; Schafer 2013; Sinhababu 2017: ch.7; Gregory 2021:ch.4.
This section discusses some of the many interesting issues about hownormative and motivating reasons relate to each other. One cluster ofquestions (section 4.1) centres on the guiding role of normative reasons. A second (section 4.2) focuses on what it is to act for normative reasons, and thesignificance of doing so.
An attractive and popular idea is that normative reasons are supposedto guide us, in deliberation and in action. That is what normativereasons seem to befor. But how should we spell out thisthought? One way is normative: a reason is a fact that we ought to bemoved by, or that it is rational, or appropriate, or good reasoning,to be moved by. This might be called the “deliberativecondition” on normative reasons. Another way is descriptive:normative reasons must be facts by which you can be moved: they mustbe potential motivating reasons. Call this the “abilitycondition” on normative reasons. Often these ideas are combined:normative reasons must be facts that you are able to be moved bythrough sound deliberation, or in so far as you are rational (Nagel1970; Williams 1979, 1989; Korsgaard 1986; Smith 1994). Call this the“combined” condition.
We might ask which of these conditions, suitably refined, bestcaptures the initial thought, or is most plausible on reflection.Rather than engage with this directly, this section will illustratehow the differences between them matter by looking at ways in whichthe three versions have been put to work. Since the deliberativecondition has been discussed insection 2, this section will mostly focus on the ability and on the combinedconditions.
The most famous appeal to the combined condition is found inWilliams’ (1979, 1989) case for a broadly Humean view ofnormative reasons—at least on one common reading of that muchcontested argument (see entry onreasons for action: internal vs. external). As noted insection 2, Williams claims that we have reason to do only what we could bemotivated to do via a sound deliberative route. Since Williams alsofinds it plausible that what we can be so motivated to do will varywith our desires (broadly construed), he concludes that our normativereasons will also vary accordingly. One problem with this argument isthat whether sound deliberation exhibits this relativity depends onwhat exactly sound deliberation involves (Korsgaard 1986; Hooker1987). Some Kantians hold that sound deliberation will lead anyone,whatever their motivations, to the same conclusions—inparticular, to do what is morally right (Smith 1994; Korsgaard 1996;Markovits 2014). Williams (1989) concedes this point, while notingthat the burden is on Kantians to make their case. In so far as one isunpersuaded by Kantian efforts, Williams’ argument may thusretain some force.
The ability condition is often appealed to in support of an epistemicconstraints on normative reasons, or perspectivism about what we oughtto do. The basic idea is that a normative reason must fall within anagent’s epistemic perspective (e.g., they must know it or be ina position to; seesection 2), if they are to be able to be moved by it. This argument turns on afairly strong interpretation of the ability condition. In one sense, Iam able to be moved by a reason if I might be motivated by it, were Ito become aware of it. In another sense, I am not able to be moved bya reason if I am not right now aware of it. Compare: in one sense, Iam able to drive, as I sit here typing in my office; in another sense,I am not able to do so, since my car is several miles away. The firstsense depends only on my having the ability to drive; the second on myalso having the opportunity (what J.L. Austin (1956) called the“all-in sense of ‘can’”). The argument forperspectivism requires that we understand the ability condition in themore demanding sense. It is unclear whether this version of thecondition is well-motivated (for discussion see Way & Whiting2016; Lord 2018: ch. 8).
A version of the ability condition features in Dancy’sinfluential argument against psychologism about motivating reasons. Asdiscussed insection 3.1, Dancy claims that since it must be possible to act for a good reason,motivating reasons must be of the same ontological category asnormative reasons—they must be facts, or putative facts. Thisargument requires only a very weak version of the ability condition,since it involves no commitments about the conditions under whichnormative reasons might actually motivate us.
These examples illustrate some of the different ways in which theinitial attractive idea—that normative reasons are supposed toguide us—can be elaborated, and how those differences mightmatter. Some philosophers, however, reject all forms of this idea.This is largely because of putative examples of “elusive”or “self-effacing” reasons: roughly, normative reasonsthat would not survive if you were to respond to them. The mostfamiliar example is that of Nate, who loves successful surpriseparties but hates unsuccessful surprise parties (Schroeder 2007: 33;165–66). If there is a surprise party waiting for Nate at home,then that seems a reason for Nate to go home. After all, an informedfriend might advise Nate to go home by saying, “although Ican’t tell you what it is, there’s a very good reason foryou to do so!” But Nate cannot go home for the reason thatthere’s a surprise party there; this would require him to beaware that there’s a surprise party, which would spoil thesurprise, and thus undermine the reason. That there’s a surpriseparty waiting for Nate thus seems to be a normative reason that is nota potential motivating reason.A fortiori, it is not a reasonhe could be motivated by through sound deliberation. (For otherexamples and references, see McKeever & Ridge 2012; Markovits2014; Way & Whiting 2016; Rossi 2021).
There are other putative counterexamples to these conditions which arenot based on self-effacing reasons. Julia Markovits (2011: 48)describes the case of Captain Sullenberger, who successfully, andwithout any loss of life, emergency-landed an Airbus 320 that had lostall thrust in its engines. Sullenberger said afterwards that he wasnot thinking of the potential loss of life and suggested that it wouldhave been a distraction to do so. Of course, the threat to life was anormative reason for him to act as he did but, it might seem, not oneit made sense for him to deliberate in terms of.
The most common response to these objections is to deny that suchcounterexamples are genuine. For example, even if the surprise partywaiting for Nate is not a reason for him to go home, it mightnonetheless make itgood for him to go home, be a reason forothers to encourage him to do so, a reason for him later to regret notdoing so, etc. Moreover, related considerations, such as that Natewould enjoy going home, might be reasons for him to go home. For some,these observations suffice to capture our intuitions about the case(Kiesewetter 2016: 769–71; Paakkunainen 2017: 68–70).Others hold that the cases call for particular interpretations of theability and combined conditions. For instance, perhaps Nate can bemotivated by the surprise party in so far as he can go home on accountof his discreet but better-informed friend’s advice (McKeever& Ridge 2012; Sinclair 2016; for alternative interpretations seeSmith 1994; Way & Whiting 2016; for discussion see Rossi2021).
Debate over these questions goes on. And while it can be tempting todismiss such examples as artificial or peripheral, they may havesignificant implications. For example, some take them to underscorethe need to sharply distinguish the deontic and deliberative roles,concluding that there is no unified concept of a normative reason thatcan play both roles (Wedgwood 2015).
Regardless of whether all normative reasons must be potentialmotivating reasons, we do sometimes act for normative reasons. What isinvolved in this? What does it take to act for a normative reason?
The answer might seem straightforward. Dancy writes,
[i]n the best case, there is some good reason for doing the action,and the reason that motivates the agent coincides with that reason.(2000: 3)
This suggests that when the fact thatp is a normative reasonfor you to act in a certain way, you act for that reason when you actin that way and your motivating reason for doing so is the fact thatp. However, even if we allow ourselves the notion of anaction motivated by the fact thatp—seesection 3.2—there are further complications.
Consider: the house is on fire and a child is trapped inside. This isa normative reason to run inside and help the child escape: thechild’s life matters and you can save it. Suppose that you dorun inside and that your motivating reason is that you can help thechild escape the fire. But suppose you are so motivated not becauseyou are concerned for the child, but because you expect praise andreward for your action. Do you then actfor the reason thatthe child is trapped inside the house? In a sense you do, since yourun in to help the child escape. But, in an important sense, youdon’t, because you do not respond to this fact as themoralreason it is—one connected to the value of thechild’s life. (This example is due to Markovits 2010: 227; forother examples, see Lord 2017; Way 2017; Singh 2020; Howard 2021).
What is missing? In this example, the problem seems to be that youdon’t act from the right kind of concern: you don’t desireto save the child for their own sake. Perhaps, then, acting for anormative reason, in this special sense, requires acting from theright kind of desire. This idea is developed in different ways byMarkovits (2010: 227ff), Howard (2021), and especially by Arpaly andSchroeder (2014). Alternatively, perhaps acting for a normative reasonrequires acting from a certain kind of normative knowledge—e.g.,knowledge that your motivating reason is a normative reason, orknowing how to treat a normative reason appropriately. Versions ofthis view are developed by Lord (2017; 2018: ch.5), Isserow (2019),and Singh (2020). Further views have also been developed (Mantel 2014:ch.3; Way 2017; Fogal & Worsnip 2021; Whiting 2021: ch.5).
One reason this issue matters is because of its bearing on questionsabout responsibility. Whether we act for normative reasons often seemsto matter for whether we merit credit or praise for doing what weought to do. The example above illustrates this. The person who runsinto the house, since the child is inside, but from the desire for areward does the right thing but not in a way that is a credit to them,or for which they deserve praise. Kant famously put this by sayingthat in such a case the action lacks “moralworth”—though the issues are not specific to morality.
How exactly do the reasons for which we act bear on moral worth? Ontheright reasons view, moral worth depends on acting for theright reasons, that is, the normative reasons that make so actingmorally right. On this view, if the person had responded to thechild’s need as the moral reason that it is, their action wouldhave had moral worth. This view, which in some ways developsAristotelian themes, though not necessarily in terms of virtue, isinfluentially defended by Nomy Arpaly and Julia Markovits (Arpaly2002; Markovits 2010; Arpaly & Schroeder 2014). The view has manyattractions but as this case illustrates, it turns crucially on thenotion of acting for a normative reason and thus requires asatisfactory account of what this amounts to. This shows howlongstanding questions about responsibility are bound up withunresolved questions about the relationship between normative andmotivating reasons.
The above is an overview of a range of problems about practicalreasons and their widespread significance. It should be sufficient toshow how the problems and their many ramifications reach into manyaspects of our lives and have important consequences for ourunderstanding of ourselves as rational agents.
How to cite this entry. Preview the PDF version of this entry at theFriends of the SEP Society. Look up topics and thinkers related to this entry at the Internet Philosophy Ontology Project (InPhO). Enhanced bibliography for this entryatPhilPapers, with links to its database.
action |agency |bias, implicit |cognitivism vs. non-cognitivism, moral |Davidson, Donald |facts |metaethics, normativity in |moral particularism: and moral generalism |moral realism |rationality: instrumental |reasons for action: agent-neutral vs. agent-relative |reasons for action: internal vs. external |value theory
Thanks are due to Connie Rosati and to SigrúnSvavarsdóttir for their very helpful suggestions forimprovement on earlier versions of this entry.
View this site from another server:
The Stanford Encyclopedia of Philosophy iscopyright © 2024 byThe Metaphysics Research Lab, Department of Philosophy, Stanford University
Library of Congress Catalog Data: ISSN 1095-5054