The sentence "I did p because of q" is ambiguous in an obvious way. It may mean that event q determined me causally to do p, but it can also mean that I believe, or believed, that q entitles me to do p. A more interesting (philosophically) example is "I did p because of fear (or anger)". It may mean that a mental state (fear/anger) caused me to do p; I was in such a state that I could not help p-ing. But it can also mean something like: "In a situation such as the one I was in, fear/anger occurred, and for someone who was afraid/angry, doing p is appropriate".(When asked "Why did you do p?", someone can answer "I did p because i was afraid" both to justifiy himself - given my mental state, i was entitled to do p - and to exculpate themselves, i.e. to place themselves outside the realm of morality - fear coerced me into doing p, so please don't blame me for doing p).
Ambiguity is even more obviously present in "I believe that p because of q", where it can be accounted in terms of scope. It may mean that "I believe that (p because of q)", i.e. i believe there is a causal relation from q to p, or it may mean that "(I believe that p) because of q", i.e. q is (in my opinion) a ground for believing p.
A cause seems to be a ground in that, if q is a cause of p, then, if I believe that q, I must believe myself entitled to believe that p. (Note the "must": it marks the transition from "mere cause" to "ground"). This may be criticized: not all causes may become grounds in this way. For example, token causes, which are not causally effective in virtue of a law, cannot be so grounded. We don't know what they cause. so to speak, until the caused effect occurs.
But what is more obvious is that not all grounds are causes. The effect is in many cases a sign of the cause: from knowing the sign I am entitled to know the signified. The relation of grounding is here the symetrical of the relation of causing: belief in an effect's occurence is a good ground for belief in its cause's occurrence,
I will leave open the question if, by entering into belief contexts, all natural causes can become good grounds for belief in their effects. I just want to draw attention to the intuitions underlying the distinction between causes, or natural causes, and arguments (or grounds, or reasons).
The discussion of arguments versus causes has some resemblances with the discussion of mental causation, versus �natural� causation. Yet, the two discussions are not equivalent. Even if we accept that there is any mental causation over and above natural causation, it still doesn�t follow that mental events are arguments. It is not apparent why mental states such as beliefs, desires and emotions (�passions�) cannot cause their effects in the same way as natural causes do. The issue of mentality has, in itself, nothing to do with normativity and free will. As I will write in more detail, arguments are intrinsically related to normativity and free will.
I.
Things are produced by causes but people believe, desire and act because of arguments � or at least they sometimes so claim. When people ask for arguments, they don't ask for causes. The autonomy of the realm of arguments is the key of any attempt to vindicate the possibility of rational debate and of free will. On one extreme, we have Kant's position, who thought that people's reasons to act freely belong to a completely different realm from that of (natural) causes. On the other extreme, there are the varities of relativism and irrationalism who claim that argumentation is nothing more than persuasion and that any process by which we get someone to change his opinion or behave in a certain way cannot be essentially different from using rhethoric, psychiatric medication or, why not, bribing him in order to change his mind.
If arguments are irreducible to causes, this poses a methodological problem for attempts to formalize the relation of causation. I.e., the examples of relations of causation should not be selected from the "realm" of arguments. The problem is all the more acute since formal analyses of causation typically consider the relation of causatin as holding between events. But it is plausible that arguments are not events.
For the sake of uniformity, I will use the words "argument" and �reason� interchangeably. If the reader wants, he may read "reason" whenever i used "argument", I am unsensitive to the distinction (if there is any) between reasons and arguments. I don't know if the words "cause", "reason" "arguments" have exact equivalents in other languages, and this issue does not interest me.
I assume that people both act and believe because of arguments. In what follows, I want a unified account of arguments - i.e., I assume that arguments may be effective both for beliefs and for acts. Thus, arguments have effects.
II. Are arguments causes?
If arguments are just a special kind of causes, and if their �effectiveness� can be formalized in a counterfactual or probailistic way, they are events. A further question is: what kind of events are arguments? When someone says �I ran away because I the enemy was near�, he may imply that the reason of his running away was (a) the physical fact that the enemy was near; (b) his belief (a mental state) that the enemy was near; (c) the words �the enemy was near�, pronounced silently or otherwise as a justification for his decision to run away. I have in mind three possibilities: (a) they are natural events, perhaps with the property of �potentially good�; (b) they are mental events; (c) they are linguistic events. A fourth possibility is that arguments are platonic entities, existing in a realm of their own. Popper�s three worlds ontology and Kant�s belief in the realm of reasons are examples of the platonic account of arguments.
The easiest answer is that arguments are a special sort of causes. They are natural events plus some mental events or properties.. The least risky way would be to characterize arguments are believed causes. We would say that any event can causally determine me; but if I believe that an event caused me, that belief is an argument.
This is obviously wrong: I may believe, even know, that I do A because of B yet B is not an argument of A, For example, I may be drugged and behave in bizzare ways, and at the same time realize that being drugged is not an argument. Perhaps I may not only do, but even believe some things, because I am drugged, and simultaneously realize that my beliefs (be they about moral or about matters of fact; be they corerct or not) are caused by my being drugged � yet my being drugged is not a reason why I have these beliefs. Such a situation should not necessarily prevent me from entertaining those beliefs.
A more refined attempt would be to define an argument as a cause that is both believed and desired. If I do or believe A because of B, then it is not only the case that I believe that B; I also desire that B should causally lead to A. I, so to speak, agree with the fact that B should lead to A. This is again wrong, and for obvious reasons: I may desire that ingesting a certain drug, or listening to a certain music, should make me believe something or behave in a certain way; I may know (and believe) that I have ingested that drug or listened to that music. Yet I would not call ingesting a drug or listening to music arguments.
Note that I spoke about "arguments as believed and desired causes" because I want to avoid, for the moment, the appeal to normativity: Therefore I will not say that I argumentatively do/believe A because of B iff there is a logical or somehow inferential relation leading from B to A. The reason why I do so is that an argument can be effective even if it is not correct: even if it doesn't deserve, so to speak, to be effective. Unlike causation, the relation of "argumentatively leading to" has obviously a flavor of normativity: there are good and bad arguments for a conclusion, there are good and bad inferences. But bad inferences are still inferences; when people use invalid inferences, they still use arguments; they are not caused simply for being wrong).
Pushing to the extreme this issue, skepticism about good arguments is plausible if we pay attention to deviant causal chains. When someone utters �I believe that P and that P --> Q. Therefore I believe that Q� the conclusion, we can never know if the cause (or a cause) why he believes that Q is his belief in the premises that (correctly) lead to Q. Perhaps just a second before utterring that Q he forgot the premises and everything he knew about logic; perhaps he merely rationalizes (he would have believed that Q even without justification, even without believing the premises. And if we can always be skeptical about other people�s arguments being causally effective, why should we be more confident of our own inferences? It seems hard to be skeptical about good arguments being good arguments if we are skeptical about good arguments being causally effective; yet, for any particular case, it seems nevertheless easier to doubt about an argument�s validity than of its being an argument. There is a connection between validity and effectiveness, although the two notions need not be confused. Invalid arguments may well be effective (or, better to say: arguments can be effective in invalid ways)..
Another reason why arguments cannot simply be defined as believed causes: we need to know _when_ an argument is believed. Someone may do something without mentally (or otherwise) uttering any reason why he did this: then he may be asked for his reasons and offer a plausible one. "I did B because of A, but when I did B i didn't think of A" is an intelligible and correct way of speaking. (This may pose a problem for Kantian ethics, insofar as Kant says that moral acts are those who are done by duty, and not simply according to the duty. This may suggest that a person is moral in his acts only if he mentally utters, before and while doing the act, the sentence "I do this because this is my duty".). Of course, the defender of the "arguments as believed causes" may reply that people have not-yet-conscious beliefs. (I avoid the word "unconscious" because it seems to refer to something necessarily unknown to the person who acts unconsciously).
Moreover, someone may act "having in mind" a certain reason, i.e. being prepared/disposed to offer a justification for what he did, yet other people may believe (and even convince him) that he did what he did for a different reason. "I believe that B because of A". "No, you believe that B because of C". "Oh, yes, indeed, I believe that B because of C". In such a case, it is plausible to say that C is the argument that "leads" to B. It can be replied that in such contexts C is just a rephrasal, a translation of A, at least insofar as the person who B-d is sincere. But this is another issue.
Arguments are not believed causes, or believed and desired causes, but aren�t they simply beliefs or desires endowed with causal force? Why can�t we just say that arguments are mental events? (I assume that a beliefs and desires are mental events; I suppose that Wittgenstein would say they are linguistic events, and contemporary followers of the language-of-thought-hypothesis would say the same, although pointing to unconscious utterances in mentalese). Why does the distinction between causes and arguments not coincide with the distinction between natural causes and mental causes? I wrote above that even a wrong argument is an argument. Thus, it could be claimed that we don�t need normativity to define arguments; we only need mentality.
There are two reasons why arguments are not simply beliefs with causal powers. One reason is perhaps obscure, the other is, in my opinion, more convincing. The more obscure reason why arguments cannot be identified with beliefs with causal powers arises from the ambiguity of beliefs� individuation. When someone justifies himself by resorting to an argument (e.g. �Why did you do p?� �-In those given circumstances, I had no better alternative but to p�), s/he obviously doesn�t mean that that just before he acted s/he uttered mentally �In these circumstances I have no better alternative but to p�.) Perhaps he wasn�t aware of any words before his minds� eye, so to speak; perhaps was aware of some mental representations, or perhaps he wasn�t aware of anything that could be related to his decision to p. It could be replied that he still believed that p, although he wasn�t aware of p; that we can model conditions of individuations for beliefs on the conditions of individuations for arguments (or vice versa, depending on which of these seem to have clearer individuation conditions). Since I think this reply is a good reply, I find the objection it addresses more obscure. This objection is valid only if it makes sense to say �I am now realizing that I did p because of q, although at the time when I did p I did not believe q�. It makes sense to say this only if we want to exclude beliefs of which the believer is unaware, but want to preserve arguments of which the person who does something because of an argument is unaware. I am not sure this is a promising strategy.
The more convincing reason why arguments cannot simply be identified with causally effective beliefs is that a belief may be effective in different ways. Some of these ways are appropriate than others. When a belief is effective in a way that is not appropriate, we have a deviant causal chain: the belief may cease to be a valid argument, or even an argument at all. Deviant causal chains undermine the relation of �argumentatively leading to�, although they do not undermine the relation of �causing�. I may have the belief that it will rain and, as a result of this belief, I may decide that I should always take with me a red object when I walk through the center of the city. (Perhaps my brain�s abnormal biochemistry, or an unconscious memory from childhood, causes me to connect the belief that it will rain with the need to take red objects). I have a red umbrella and I need to walk through the center of the city, therefore I take my umbrella. It would be misleading to say that in such a case my belief that it will rain is an argument for my taking the umbrella. But it is obvious that this belief is a cause.
It could be replied to my example that my belief that it will rain is an argument for my taking the umbrella, but it is not a good argument. This would show that deviant causal chains undermine the relation of leading argumentatively in a correct, or valid way � which brings us back to the problem of normativity. To meet such an objection we only need to modify slightly the example: I don�t �feel� any connexion whatsoever between my belief that it will rain and my decision to take the umbrella; if asked why I took the umbrella, my answers don�t make any reference to rain. I answer something like �I needed a red object and the umbrella was the first red object I saw�. If I am taken at my word (and it is plausible that I am taken at my word), noone would think of me that the reason why I took my umbrella is my belief that it will rain, although both other people and I would ascribe to me the belief that it will rain. My belief that it will rain and my decision to take the umbrella are unrelated to each other; there is no connexion between them. This is compatible with the possibility that my belief that it will rain is still a cause of my decision to take the umbrella: the causal power of the belief is, so to speak, active at a level behind by consciousness. Thus, because of deviant causal chains, not all beliefs with causal powers are (good or bad) arguments.
But it is not yet obvious that another way of characterizing arguments as a sort of causes is not plausible.I rejected some tempting attempt to define arguments as a species of causes. But perhaps other attempts could be more successful. In particular, I did not prove that arguments are not beliefs; perhaps they are some kind mental or linguistic entities but we need to refine the definition. That�s why in the next section I will try to present some general reasons why arguments are different from causes, no matter how we define them.
III. Differences between causes and arguments
(1) Noone believes, desires or does anything for a single reason; noone offers a single argument when asked to justify himself. Perhaps this phenomenon says nothing important about the �ontology� of arguments; perhaps this phenomenon is just an effect of the richness of our language. Our language is such that people can use many different arguments in support of their beliefs or deeds � at least of those beliefs and deeds upon which they reflected the most. Thus, thesis 1. seems to depend on a particular conception of arguments: arguments are identical to their linguistic formulation. If arguments are not identical to their linguistic formulation, thesis (1) would show nothing more than that an argument can be described in more ways than a cause can be.
It can be replied to thesis (1) that it is trivial, since no natural event occurs because of a single cause. It is almost a clich� to say that any cause is effective only given an adequate background, perhaps given some laws of nature. But arguments are different insofar as a conviction is the better founded the more arguments supporting it exist; a deed is the more understandable the more arguments justifying it it has. (by �understandable� I don�t mean �moral�). There is no parallel in the realm of natural causes: we cannot say that an event is more existent or �real� if it has more �causes�. Identifying one or several causes of an event is a measure of the attention we are ready to pay to the event, it depends on the particular interests of the person who identifies the cause.
(2). It is a gradual matter if someone does or does not do something because of an argument. In any context, some arguments are more effective than others � but the less effective arguments are effective as well. �I had many reasons for believing X is dishonest; but I was not convinced of this until I discovered that X stole from Y�: in such a case we should say that there is a decisive reason for believing X is dishonest (my belief that he stole from Y) but there were many other � less effective - reasons for believing X is dishonest. A metric of effectiveness seems to be required for arguments, but not for causes. (In the realm of natural causes, the notion that some causes are more effective than others seems to make little sense).
(3)
Related to (1). and (2).: In the case of arguments, the phenomenon of preemption collapses into redundant causation. Indeed, it is counter-intuitive to say that �John did C based on argument A,; but, hadn�t A been effective, he would have done C based on argument B�. If someone had two arguments, at least if he thought of two (or more) arguments for doing, believing or desiring something, then these arguments combine each other. For example, if someone had both purely egoistic and moral reasons for doing something, then doing that thing is a fortiori a free deed of his; it belongs a fortiori to his essense, so to speak: there is no doubt that he chose to do that thing. This thesis is closely related to previous two theses: an argument�s effectiveness is a gradual matter (arguments are more or less effective) and noone does anything for a single reason.
(4) Unlike most causes, arguments are the domain of norms. There are good and bad arguments; there are good arguments for false beliefs, or for immoral acts, and there are bad arguments for true beliefs, or for moral acts.
(5) Arguments are not coercive. The connection between arguments and normativity implies its connection to free will. (Norms cannot be effective on unfree agents). No matter what is the relation of an agent (or a believer) to any argument, s/he still would have been able to refuse to reach the consequences of that argument. I feel there is a strong intuition in favor of this position, although it can be challenged both by extreme intellectualism, like Socrates, who thought that one cannot know the Good without doing it, and by pragmatists, who denied the separation between beliefs and acts).
If arguments are not coercive, then they are never sufficient conditions for their consequences. Someone may believe and understand the premises of a syllogism yet reject the conclusion. More rigurously speaking, the uncoerciveness of arguments means that whenever an argument A leads to consequence B (where B is a belief, a desire or an act), for any moment of time before B it would have been possible for ~B to occur. Given A, any non-B course of world is consistent with the world-history until B. (If in the above sentence we remove the �Given A�, the result is not obviously correct: perhaps after believing myself entitled to do B because of A, I started a course of action that, from some moment on, made it impossible for me to refuse to B. What I want to stress is that an argument can never be the sufficient cause of its effect. For example, people�s free decisions may lead them to do unfree acts: for example, I may decide to take drugs based on my moral convictions that it is good to take drugs, then become addicted and unable to refuse to take drugs). This is a case when an argument (the conviction that it is good to take drugs) s responsible for an event (feeling unable to stop taking drugs) that was not freely chosen, i.e. couldn�t have helped to occur given the world history until it; yet, it is not the argument that is responsible for the fact that its effect was �bound to happen� in this way. The argument itself was not coercive.
(6). Arguments are most of the time intensionally effective, while the relation of causation is always - we are told � extensional: arguments are not extensionally effective, at least not in a trivial sense. If A causes C, then the relation of causation between the two holds irrespective of the descriptions used to designate A and C..If someone acts or believes something because of argument A and argument A has the same content with argument B, it does not follow that he acts/believes that thing because of argument B. (Classical examples of intenstionality appear exactly here: Oedip shot an arrow with his bow at the stranger whom he met because he had some beliefs and desires about the stranger; that stranger was his father, yet Oedip did nothing � at least not consciously � because of his beliefs and desires about his father).
IV. Do we have first person authority about the arguments we follow?
It could be claimed that there is another difference between causes and arguments: we are infallible about our arguments, but fallible about causes. In my opinion, this is not a true difference between causes and arguments. We do not have first person authority about our arguments.
�Arguments are privately known. Whether an argument was effective or not, this is only known to the person who did or believed something because of that argument�.
V. Arguments as idealized causes
The best definition of arguments I can imagine is that arguments are idealized causes of human acts/beliefs or desires. They are ideally intelligible causes of events caused by humans. By this, I mean the following: when someone justifies himself by resorting to an argument, he describes a world history in which an ideal relation of causation holds. When we ask �why did he steal the carpet�, we may receive causal answers (he could not help stealing the carpet given the situation he was in; his brain chemistry was such and such and trigerred the physical gesture of stealing the carpet). But we may receive answers of the type �because he believed that the carpet did not had any importance to its owner� or �because he believed that the carpet�s owner was a bad person and it was appropriate to take revenge on him� or �because he believed that the carpet was necessary to his purposes�. These answers point to arguments for, not to causes of, stealing the carpet. They may be bad arguments � if it is wrong to steal in general; but they are still arguments: unless causes, they may be criticized and debated. They are governed by norms.
Laurian Kertesz
I want to emphasize that the distinction between valid (good) and invalid (bad) arguments does not parallel the distinction between effective causes and �preempted� potential causes (i.e., events that could have been effective causes but failed). Someone may entertain beliefs (generally false, but not necessarily) based on bad arguments. If we want a place for �effectiveness� when dealing with arguments, it is obvious that bad arguments may be effective.
Perhaps this seems to be the most obvious feature of arguments and I should have started with it. However, I think that the relation between arguments and normativity should at least be qualified: causes too can be governed by norms. Artifacts, works of technique, mechanisms can work well or badly. They are parts of the causal word (they are appearances, in Kant�s terminology), yet their causal relations � insofar as they are judged from the point of view of human purposes � are governed by norms. Therefore it is false that arguments are simply causes governed by norms. On the other hand, it is true that all arguments can be evaluated as good or bad (perhaps with intermediary degrees: plausible but not good enough, good but not clear enough etc). Unlike arguments, causes need not have anything to do with normativity: they may be governed by norms, but this is not necessary. Most � but not all � causal relations are outside the province of morality or validity.
It could be replied that the case of artifacts can be eliminated if we define arguments as mental events or states which can be causally effective according to norms.
For more riguour, the �uncoerciveness� of arguments should be qualified. I have in mind three possible situations:
(a) I may do A or B, it doesn't matter; it is a matter of taste if I do this or that;
(b) I should do A: it is the best thing to do in this situation.
(c) I am bound (causally coerced) to do A.
If instead of �do� we replace �believe�, the issue may become more complicated, since it is not obvious that we can believe what we want. My intuition is that we can believe what we want (that�s why talk about �ethics of belief� is meaningful), but exploring this issue is beyond the scope of this paper. I think it is possible to have situation (a). collapse into (b). Indeed, we need arguments even to act on our subjective preferences: people sometimes argue why they should be allowed to follow their tastes. Arguments are presented, so to speak, even when they are meant to prove that further arguments are unnecessary. This issue perhaps needs a more in-depth analysis and perhaps formalization.
i
This is more controversial than it may seem. If we are committed to the thesis that arguments are linguistic entities, it could be claimed that arguments are extensionally effective. In order to �extensionalize� them, we only need to describe them in such a way that there is never identity between two arguments with different linguistic formulation. (Multiply identities in order to vindicate extensionality). To any apparent intension there is an extension corresponding to it. Of course, if we could sanely claim that Hesperus is different from Phosphorus just because they have different names, Frege's puzzle would disappear: there would be no "synthetic identity". And it seems there is no synthetic identity between different linguistic formulations: synthetic identity, according to Frege, occurs when it is discovered that two different linguistic formulations have the same reference. But it seems that there is no point in speaking about synthetic identity when the reference is itself a linguistic entity. We cannot discover that two linguistic formulations are identical otherwise than by means of language itself.
Yet, I maintain that arguments are intensionally effective in a way that causes are not. If we really followed the extentionalization strategy outlined above, it would be impossible to ascribe to a person an argument formulated in different words than he formulated it (or he would have formulated it). But there is a strong intuition that we often ascribe to a person beliefs using different words that that person uses. (Leave aside the problematic case of "unconscious beliefs"; I merely speak about rephrasing someone's arguments. Rephrasing in order to make more clear: a bit of translation. �He says he did B for reason A; but A actually means that C). The phenomenon of rephrasing pleads against the thesis that arguments are linguistic entities.
I have in mind two different ways of extensionalizing arguments. The first is trivial: if someone has a belief about Hesperus, he has the same belief about Phosphorus. This trivial extensionalization is correct in de re readings; but de re readings of beliefs provide little hope for understanding the effectiveness of arguments. It is obvious that someone may act or believe something based on his beliefs about Hesperus, but not on his beliefs about Phosphorus. (This reminds one of the arguments in favor of �narrow content�: mental events broadly individuated cannot have causal powers).
Another way of extensionalizing arguments consists simply in postulating as many extensions as there are intensions. On such an approach, there is not synthetic identity when we refer to arguments. Beliefs and arguments are isolated one from another, so to speak, without doors and windows. We cannot discover that two arguments are identical if we didn�t know this from the beginning. The consequence of this approach is that rephrasing the words uttered by someone else as a justification for his beliefs, desires or acts is never appropriate. Rephrasing assumes intensionality about arguments. I.e., rephrasing assumes that we can discover that two linguistic formulations have the same reference but different sense. But rephrasing is strongly supported by our common intuitions. Therefore, arguments are intensionally effective. (This seems to me a powerful reason why arguments are not identical to their linguistic formulations � although it might be claimed that a relation of translating could be defined, such as an argument and its rephrasal are not coextensive but synonymous).
The first (trivial) way of extensionalizing arguments offers an easy (though restrictive) way for rephrasing someone�s arguments. The way is easy because whenever someone does something based on Fa and a=b, we should rephrase his arguments by saying he acted because of Fb. This way of extensionalization is incorrect, because (1) it does not account for the effectiveness of arguments (i.e., it is not the case than whenever Fa is an effective reason for something, Fb is an effective reason) and (2) it is simply not true that all our ways of rephrasing other people�s arguments are based on this trivial extensionalization.
On the other hand, the second (non-trivial) way of extensionalizing precludes any attempt to rephrase other people�s arguments. Rephrasing would simply be illegitimate: we could do no better than taking someone at his words when he claims that he acts/believes/desires because of A. This would be equivalent to first person authority concerning arguments.
Reprhasing assumes a certain extensionalization of arguments, but this extensionalization is not the trivial one. It is beyond the scope of this paper to state the conditions under which it is acceptable or �correct� to state someone�s reasons in different words than he stated (or would have stated) them. I will mention briefly that we need to distinguish, in my opinion, between two cases: (a) I rephrase someone�s arguments without rejecting the description he offered of his arguments. I rephrase just in order to make it clear for myself or for my hearers why someone acted/believed/desired the way he chose; (b) rephrasing-with-rejection of the words he used. �He says he believes p because of q; but I know that he believes p because of r, not becaue of q�. (This does not imply that the person we are referring to is insincere, it only implies only that s/he doesn�t find the right words for describing his/her reasons).
The phenomenon of rephrasing attests that arguments are not linguistic entities unless we postulate synonymy of linguistic formulations. But if arguments were linguistic entities, then they would be extensionally effective, according to the second (non-trivial) strategy of extensionalization. Therefore, this strategy is wrong. On the other hand, it is true that rephrasing assumes a certain extensionalization: I refer to the cases (a) described above (i.e. rephrasing someone�s description of his arguments without rejecting his words). In such cases, an argument is effective even under different descriptions. But there are limits of rephrasing: it is not the case that whenever someone says he believes/desirres/acts because of argument A and A is coextensive, with B, then he believes/desires/acts because of argument B. Sometimes coextensiveness does the job; sometimes it doesn�t. That�s why arguments are, most of the time, intensionally effective: their description matters, it is either unrephrasable or rephrasable along certain non-trivial lines.
I think it is already clear why it should be rejected. First, if we were infallible about which arguments we follow, we should be all the more infallible about whether we follow and argument or are merely caused. But � see above � to the extent that we follow an argument we are free. To the extent that we are caused, we are unfree. Therefore we would be infallible about whether we are free or unfree. Then the feeling of freedom would be the same as freedom. Freedom would be a private matter. Kant rejected this, and I think for good reasons. Someone may believe he is caused, yet act freely.
Second, if we had first person authority about which arguments we follow, rephrasal-with-rejecting-someone�s-words would be again illegitimate. But this kind of rephrasal is legitimate, therefore we are not infallible about which arguments we follow.
Third, if we were infallible about which arguments we follow, then we would be infallible about whether we follow them correctly or not. (I call �weak first-person authority� the thesis that we are infallible about which arguments we follow; and �strong first-person authority� the thesis that we are infallible about whether we follow correctly or not an argument). It seems plausible that weak first-person authority leads to strong first person authority. But, if strong first-person authority is true, validity, morality, conformity to norms in general would be a private matter. I think it is obvious that validity and conformity to norms cannot be a private matter. However, it could be replied that we can be infallible about which arguments we follow, but not about whether we follow them correctly or not. I am not sure this reply is sound, however the issue needs to be debated. (It is not a coincidence that irrationalism � which denies the distinction between argumentats and causes �is closely related to relativism � the thesis that correctly following an argument is a private matter). If correctly following an argument were a private matter, i.e. if we have strong first person authority about arguments, then the issue of normativity would collapse. Arguments could not be governed by norms. And � if being governed by norms is definitory for arguments � then arguments would be causes. It remains an open question whether even the weak first person authority has the same consequence, i.e., it leads to the collapse of arguments into causes.
This is related to the above discussion about intensionality. If an argument were infallibly known by the person who acts/believes/desires because of them, then arguments would be extensionally effective � because of the impossibility of rephrasing someone else�s argument. It could not be the case that an argument is effective under a description but ineffective under another description � indeed, different descriptions would be different arguments. Leave aside the further problems which would arise from the fact that arguments would become either non-entities, or entities accessible by introspection
Therefore, we do not have first person auithority about which arguments we follow. Someone else can know better than we do what are the arguments which motivated us to do, desire or believe something: we are not infallible about our arguments.
The lack of first person authority about our arguments is well accomodated by the intuition that arguments are objective. By the �objecivity� of arguments I do not mean Platonism, i.e. I do not mean arguments exist in an autonomous, immaterial realm. I simply mean they can be communicated; that more people can justify themselves with the same argument; that the identity of the person who follows an argument is immaterial for the identity of the argument.
A different problem is whether the lack of first-person authority about the arguments we follow is linked to the lack of first person authority about the beliefs we are entertaining. Since it is intuitive that we often (legitimately) ascribe to persons beliefs which they don�t ascribe to themselves, or rephrase the words people use to report their own beliefs, it seems that there is no first person authority about beliefs either. If arguments are beliefs of some kind the lack of first person authority about arguments follows trivially. The problem is too vast to be discussed here with the hope of exhaustiveness. I will only mention that
(i) it is not obvious what the criteria for individuation of beliefs are;
(ii) it is not obvious that arguments are beliefs (see section II.); ;
(iii) it seems plausible that at least in some cases there is first person authority about beliefs;
(iv) it is plausible that the beliefs� criteria of individuation depend on the arguments� criteria or individuation. When we ascribe beliefs to other people (and it is obvious that we can ascribe beliefs � propositional beliefs � even if they did not ascribed these beliefs to themselves) we respect some basic criteria of rationality. When someone believes that p and that pq, we are generallly entitled to ascribe to that person the belief that q � even if we haven�t heard him utterring q. Thus, we need arguments to identify beliefs.
If arguments need to be believed in order to be effective, they are not necessarily beliefs.. A belief may be a mental event; but arguments are not mental (or otherwise) events. It makes sense to say �I am justified by A for having done B, although when I did B I was not aware, nor did I believe that B�. For example, I behaved in a way that was not known to me to be moral. Afterwards I came to the conclusion that it was moral to behave the way I did. Thus, I may say �I behaved so-and-so because it was moral to behave this way�. The (alleged). morality of what I did is an argument why I did it; but no belief was here causally effective, since when I did what I did I did not believe that it was moral.
The advantage of the definition of arguments as idealized causes is that it does not make arguments compete for �causal effectiveness� with natural causes. When we say that an event (someone�s belief, act or desire) had both causes and reasons, we do not mean that there was here a case of overdetermination. We only mean that the explanation by arguments is the most intelligible explanation; the explanation which makes the relation of causation the most transparent. Explanations of human-caused events by causes are less transparent.
Arguments are thus idealized caused of human-caused efffects not in the sense that they are the �best� causes � since bad arguments too are arguments. The ideal which arguments approximate is the ideal of intelligibility: it is not the ideal of morality (in the case of arguments for acts and desires) or of validity (in the case of arguments for beliefs). Arguments are those causes which are causally effective in the most intelligible world. On the other hand, good arguments are the causes that are causally effective in the best possible world � but the best possible world is an intelligible possible world. As physics idealizez natural events in order to accommodate more conveniently its ecuations, so common sense idealizes causes of human beliefs,acts and desires in order to debate them more clearly.
By defining arguments as idealized causes, I also have in mind one of the formulations Kant gives of the categorical imperative: �Act as if the maxim of thy
action were to become by thy will a Universal Law of Nature� (Fundamental Principles of the Metaphyscics of Morals, Introductory Note, Part ). Kant�s point with this formulation was that a moral deed is one whose maxim can be universalized as a law. But universality is characteristic of laws of nature. Therefore, the morality of an act consists in its being ideally the instance of a law of nature.
Defining arguments as idealized causes does not amount to platonism. Platonism about arguments says that arguments are ideal events, existing apart from natural causes. Platonism has obvious difficulties since it separates arguments from events: How can something situated beyond the domain of causes have effects? By what psychological faculty can we grasp an argument? How can we be sure that we grasped an argument correctly? What does the relation of grasping consist in? In order to avoid these difficulties, I prefer to define arguments as idealized causes, but not as idealized events.
Of course, this definition faces serious problems. First, it should be proved that arguments-as-idealized-causes provides an explanation of the differences between causes and arguments, as I spelled them in section III. Most of all, why are ideally intelligible causes not coercive, if causes are coercive? And what is the ideal of intelligibility? Second, it should explain why arguments, so understood, are not events. Answeting these questions is beyond the scope of this paper; the definition which I offer of arguments as ideally intelligible causes is only a proposal.
First Year M.A.