Covenantal Forgiveness

PDF: Forgiveness sem pap 2014

Whatever forgiveness is, or whatever forgiving someone implies, it has to be clear, intentional, and uncompromisingly moral. It should also be free, unmerited, and gracious. Yet any account of forgiveness as simply ‘letting go’ cannot be sufficiently moral. There has to be some kind of reconciliation among persons. But in order for reconciliation to be possible, there must be a relatively clear state if affairs according to which forgiveness becomes possible. For in a situation where forgiving is appropriate as opposed to otherwise, there have to be at least two parties, one of whom has offended the other, where that offence is sufficient to bring about some kind of division between the two parties. That division may be expressed in terms of animosity between the parties, especially in terms of resentment, which we may take to be some set of negative reactive attitudes direct by the offended party toward the offender as its proper object and which causes the offended party to feel resentful toward the offender and perhaps to treat the offender in negative manner. Forgiving someone does seem to involve the offended party ‘letting go’ of his justified resentment somehow, but in such a way that is relational and reconciliatory.

But it seems difficult to think about how this may come about. If we propose, as Charles Griswold has, that there surely are some actions which the offender may do, where he may perform some activity or performative utterance (“I am sorry”) in order to mitigate or seek to appease him whom he has offended such that he may morally obligate the offended party to forgive him, then forgiveness is seen as the repentant offender’s due. But then it does rather seem, as Lucy Allais has pointed out, that the offender has earned his forgiveness. But the concept of earning forgiveness seems to defy the free, gracious nature of forgiveness. If I am obligated to forgive someone for something they have done, given those actions or speech-utterances they have performed, I feel I have been morally obligated to forgive them, and I do not feel to have forgiven them freely and graciously. This is especially pertinent in cases where ‘letting go’ of that resentment which grips me emotionally is difficult. For oftentimes, when we try to reconcile with somebody, it seems like we have entered into a process of attempting to forgive them, and when we have reconciled with them we believe that we have forgiven them ipso facto. Say that we do reconcile with them presently. Very well. However, sometimes in memory we inadvertently remember that person’s original wrongdoing, and we find that is conjures up old resentment we thought we had been rid of. This seems like a problem, for then it seems we may not have forgiven them after all.

This paper therefore seeks to explore the issue of forgiveness from a conditional angle versus an elective angle, especially with reference to Allais’ work on elective forgiveness in direct response to Griswold.[1], [2] I will argue that Allais’ account captures something essential about forgiveness—that is, its free and unmerited nature—but that there are certain highly intuitive situations which we can conceive that seem to illustrate situations in which it would just be crazy for an offended party not to forgive their offender. In such cases it would seem that offended party is obliged to forgive the offender. But then it turns out that at least something like Griswold’s conditional account would be true. But we don’t want to speak of forgiveness as one’s giving the offender “his due”. So it seems like a dilemma. We want to maintain that (A) forgiveness is gracious, unmerited and free, (B) forgiveness can be conditional in such a way that, under certain conditions, the offended party is somehow obligated to forgive the offender, but (C) the forgiveness is such that it is absolutely and completely gracious, and is not somehow earned, merited, demandable, and so on. Ultimately, it will be concluded that perhaps something like a hybrid of the conditional/elective accounts is possible so as to maintain (C) even given (A) and (B). I have chosen to call this a covenantal account of forgiveness, according to which individuals or groups of people can impose an agreement (a covenant) whereby they can freely choose to forgive under certain morally sufficient conditions that obligate forgiveness in the future in such a way that, were one not to forgive under those conditions, one would be doing something wrong, even though one is the rightly offended party.[3] It may even be the case that such agreements—or covenants—are imposed implicitly between morally relevant persons and communities. What should motivate such a self-imposition to forgive under the right conditions? I do not believe it is in giving the offender his due, as such. Rather, I will argue that our forgiveness of others should be couched in a morally virtuous character, namely, a character of grace motivated by love. I will argue this primarily with reference to Glen Pettrigrove’s virtue account of forgiveness in two chapters of his recent book on the subject.[4]

Let us begin by characterising a scenario in which forgiveness becomes possible. Say some person, P, has $50 that gets stolen by V. So V has offended P. Now, say P knows V stole the money. P would rightly be offended at V, and resent V for stealing her money. That is, P’s resentment would be morally justified. P’s resentment of V would be further compounded if P demanded the money back from V, and it turns out that V had spent the money and had no money left with which to repay P. Now the loss is irreparable: P permanently lacks $50 by virtue of the act of V. Thus P resents V, and P becomes divided from V, and so on. Such justified resentment is a necessary condition for forgiveness, and makes possible a situation where P can forgive V. Very well. But under what conditions or for what reasons should P treat V in such a manner other than negatively? According to Pamela Heironymi, forgiving someone involves something to do with mitigating resentment. Yet we do not want to avoid justified resentment, nor do we want to manipulate ourselves out of justified resentment.[5] For when V wrongs P, P has received from V some course of action for which P was not worthy; i.e., P ought to have been treated better by V. So in order for P to be able to forgive V, P must mitigate P’s resentment of V. P must forego her resentment of V, but not just in any way at all.[6] In forgiving someone, we want neither to retract the fact that there has been a wrong perpetrated, nor do we want to adjust our view of the other person so as to make it our as if they have not wronged us after all. For forgiveness to be possible, there really has to be a wrong, and that wrong must be worthy of genuine, justified resentment.

But if resentment against a wrongdoer is justified, then it seems like a necessary condition for forgiveness that this resentment be mitigated, or negated, somehow. But this is difficult as the offended party’s resentment is justified. So to mitigate this resentment, the wrongdoer has to do something which somehow overrides the justification the offended party possess in order to resent him. Griswold has argued somewhat in this regard. Griswold believes that forgiveness is a four step process.[7] First, the wrongdoer acknowledges responsibility for the wrong; he really did steal the $50. Secondly, the wrongdoer repudiates the action; he denies that his theft was a valid course of action for him to take, and that it really was wrong. Thirdly, the wrongdoer commits to a new way of life: he repents of the wrong and seeks to be a better person. Fourth and last, the wrongdoer expresses and acknowledges an understanding of the hurt wrought on the offended party by his wrongdoing and seeks reparation. Any resentment on part of the offended party has is said to be unwarranted in light of this penitent action, and forgiveness is required as a proper moral response to the situation, and it has now become the offender’s “due” to be forgiven by the offended party. So the offended party now ought to forgive his wrongdoer.

But something seems desperately wrong with this account. For Griswold’s forgiveness-conditions are wholly dependent on what the wrongdoer must in order to morally override the warrant or justification of the offended party’s resentment. But say there just are no conditions under which at least one of these four conditions can be fulfilled by the wrongdoer. Say the wrongdoer killed one of your close family members. This would surely bring about justified resentment in you. But then say, by some other completely unrelated freak-accident, this same wrongdoer gets placed into a comatose state out of which he shall never awake. Does it really follow that you are not able to forgive him? Must you be forced to carry around the burden of unforgiveness for the rest of your days? For if a person is in a comatose state, they cannot possibly fulfil any of the conditions Griswold has suggested. But it still feel like you could forgive them for the great wrongdoing—and that not on account of pitying them for being in an irreversible comatose state. It seems like you should be able to forgive them freely, without regards to any conditions being fulfilled on their part. It may even turn out that you are an extremely admirable person for being able to do so, in such a manner that you serve as a moral example for others. So there is intuitively something wrong with an absolutely conditional case such as this. It falters in situations where the wrongdoer simply cannot fulfil the forgiveness-conditions. So it seems Griswold’s account is false.

One philosopher has also responded directly to Griswold conditional account (which she calls ‘over-moralized’) on account of its missing something intuitive about forgiveness, namely, its free and unmerited nature. Allais argues that “there is a central sense in which forgiveness can be undeserved” and this lies in forgiveness’ “elective” nature.[8] Forgiveness, says Allais, is something like a freely given gift; something given to the wrongdoer that is unearned by them: “It is something which the wrongdoer is not in a position to demand, something to which the wrongdoer does not have a right, and there is a central and crucial sense in which the decision to forgive is at the victim’s discretion.”[9] That is to say, the offended party—or victim, in Allais’ usage—may elect to forgive or not to forgive at will. For if something like Griswold’s account is true, then it seems that Allais’ objection to Griswold’s account is accurate that the wrongdoer could demand to be forgiven if he has done the right things by fulfilling the appropriate conditions. As Allais rightly puts it, “But why is what the offender is due not something she is entitled to ask for? Why should the victim have so much discretion with respect to giving the offender what is her due?”[10] Proponents of the view could bite the bullet on that one, but it seems they shouldn’t. Rather, it seems that Allais is right that if forgiveness is supposed to be a generous and praiseworthy thing, then it should be free and unmerited. It should be entirely up the offended party to forgive or not to forgive at will. This way forgiveness could be a gift of grace to the exclusion of all merit, or due: “forgiveness involves giving up resentment to which you are entitled.”[11]

So far so good. But there seems to be a glaring error in Allais’ account. For whereas it does seem that being able to forgive someone in a morally acceptable manner freely and wholly apart from conditions is a gracious and attractive feature of Allais’ account, it seems that Allais’ negation of conditions for forgiveness absolutely implies that unforgiveness is also just as morally acceptable in any situation whatever, again, wholly apart from conditions or actions undertaken by the wrongdoer. But that does not seem right at all. Imagine, again, that V steals $50 from P.  On Allais’ account, there just are no conditions under which P ought to forgive V. If P forgives V, P forgives V wholly of P’s own free will, apart from anything V has done or will do. P can forgive V freely when V decidedly chooses not to repent, or even when V does not think V has done anything wrong at all. But that implies that P may also choose not to forgive V under any circumstances whatever. But that seems crazy. Imagine V steals $50 from P. So P rightly resents V. But now imagine that V falls prostrate before P in dust and ashes, and pleads with V to forgive and restore and mend the broken relationship between P and V. Then imagine that V produces a valid $5000.00 check to replace the $50 which V will put into P’s bank account immediately. Then, imagine V offers to become P’s slave forever and ever! And so on. It would just seem crazy for P not to forgive V in this situation. But, then, is not P somehow intuitively obligated to forgive V in certain circumstances? V is, after all, truly repentant. Apparently not, on Allais’ account. Why? Because forgiveness is free. There are no conditions under which one is obligated to forgive. But that seems absurd. Yet this kind of morally absurd situation is just what Allais’ account implies. If there are no conditions whatsoever under which P is obligated to forgive V, then P can choose to forgive or not forgive V no matter what V does. But that seems insane. It also seems to make morally permissible the vicious characteristic trait of being unforgiving. But that is false; it is not morally permissible to be an unforgiving person. So Allais’ account is false.
It seems, then, that neither an absolutely elective account nor an absolutely conditional account of forgiveness will work. But there is something attractive on both accounts. On the conditional account, it is easy to avoid crazy and morally absurd counterexamples to elective accounts like the one just outlined. Such accounts also give a clear and concise account of how and when it is morally permissible or obligatory to forgive. But the elective account makes easy sense of how might have the freedom to forgive, and to be gracious, and selfless. Consider, again, the comatose state. One ought to be able to forgive a person like that. So perhaps something like a hybrid of these two highly intuitive yet mutually incompatible and independently problematic accounts probably makes the most sense. We need an account of forgiveness, according to which the mitigation of resentment is worked towards but is not absolutely necessary for forgiveness. We want to say (A) forgiveness is gracious, unmerited and free, (B) forgiveness can be conditional in such a way that, under certain conditions, the offended party is somehow obligated to forgive the offender, but (C) the forgiveness is such that it is absolutely and completely gracious, and is not somehow earned, merited, demandable, and so on and so forth. Furthermore, we do not want a fake solution, where the third, new account of forgiveness offers no new insights, and suffers from the weaknesses of both those accounts of which it purports to be a hybrid.

We may propose one such account: the covenantal view. A covenant, roughly, is an agreement that is imposed by one party upon another, either explicitly or implicitly. For example, when P is on the street with V, P implicitly agrees with V not to harm V only if V does not harm P, and vice versa. I will give you my pen, thinks the student to his peer, only if you do not use it to stab me in the eye, or whatever. These implicit agreements could be multiplied. In the moral case, it seems we can think about forgiveness through a covenantal point of view. For if a covenant is imposed by one party on another, then it may be possible for P to impose a conditional state of affairs on V in such a way that P was entirely free in imposing them on V, and those conditions now orientate the way P relates to V on whom P has posed this covenant or agreement. This may take place in various ways. But, fundamentally, it does turn out that the one making the covenant or agreement has full free will concerning the conditions he or she may impose onto his or her covenant partner. Think of the $50 theft scenario. Say V wrongs P in this regard. P therefore justly resents V, and if V wants to be forgiven it may simply just be that V has to do certain things towards P such that P may forgive V, such as are a P’s own discretion to impose on V the wrongdoer as the offended party. What might these things be? Well, on a covenantal view, they may be just anything at all such as are elected by P to serve as sufficient conditions on the basis of which V can now be forgiven, and V counts as being worthy of forgiveness insofar as V has fulfilled the conditions imposed by P.

But an objection immediately arises that this seems to make the conditions for forgiveness radically relative. For if P may impose the forgiveness-conditions on V, such that it is entirely free and up to P what those conditions are, then they would imply become the morally sufficient conditions for the forgiveness of V, such that V must act in accordance with them so as to be forgiven. But surely if the conditions chosen by P are freely chosen, then they conditions could be as strict or as lax as P wants. If V steals $50 from P, P may decide to forgive V free of charge, or force him to pay back $5000.00. But that seems wrong.

Perhaps, then, the imposition of forgiveness-conditions that may be imposed after the fact of the wrongdoing will surely have to be balanced by some other intrinsically good factor, such as fairness, where the conditions for forgiveness that are elected by P must be of such-and-such a degree of severity such that count as being the right kinds of conditions for the morally responsible forgiveness of V. Whatever conditions P freely chooses to impose on V in order to mitigate the resentment of P in such a way that unforgiveness on P’s part should become unwarranted on the completion of said conditions by V, these will have to be of such a severity such as is in keeping with the severity of the wrongdoing of V against P. In other words, P cannot just elect anything as conditions. The conditions for forgiveness will have to be kept within the bounds of fairness and justice. This ‘fairness principle’ implies that P cannot just set any conditions whatsoever in response to a wrong. It is not morally justifiable when V steals $50 from P that P simply declare that V pay back $5000.00 and be P’s slave forever and ever.  Something like the ancient Hebrew principle is probably true: “But if there is harm, then you shall pay life for life, eye for eye, tooth for tooth, hand for hand, foot for foot, burn for burn, wound for wound, stripe for stripe.”[12]

But then have we not negated entirely the freeness of Allais’ elective account of forgiveness? It does not seem so. All we have negated is absolute freedom—the freedom to impose just anything at all. Other moral virtues like fairness and justice seem to demand for this restriction. But it also seems right to argue that there are two morally admirable virtues, on the basis of which the imposition of forgiveness-conditions might be balanced as is in accordance with what is fair and just. The imposition of forgiveness-conditions by the offended party who has been wronged may be imposed freely after the fact of the wrong, but they have to be imposed as is in accordance with what is also morally acceptable.

Here, Pettigrove suggests two factors in terms of morally admirable and virtuous attributes as sources which ought to direct the conditions under which one chooses to forgive others. First and foremost is the attribute of love. Being a loving person, says Pettigrove, is a virtuous attribute to have. More precisely, biblical agape (pron. /‘agәpi/) implies a form of love that is generous, self-sacrificial and beneficial to its object. For one to love someone in an agape sense implies that one is concerned for the fundamental wellbeing of the other person. This kind of agape love is essential to a maximally morally upright character. It may even be directed toward a person who is completely unworthy of the honour. Love in this sense is closely related to Pettrigrove’s second attribute, which is the attribute of grace. Recall Griswold’s account where under specific conditions the offended party must give the wrongdoer his “due” of forgiveness. But grace, in contrast, implies giving unmerited favour such that the beneficial thing given to the wrongdoer is neither merited nor deserved by him.[13] Grace is related to other morally admirable virtues like kindness and mercy, and a graciousness of character is marked by one’s eagerness to see to the other’s good. Grace and love are related: a loving person will be a gracious person, and grace will motive some form of unmerited kindness toward the wrongdoer that will seek his good and restoration in right relationship with you, which is your mutual reconciliation. Insofar as justified resentment goes, Pettrigove is right: “love and the hostile attitudes forgiveness overcomes are unhappy bedfellows…”[14] But from this it does not seem to follow that “…love must in the end yield forgiveness.”[15] This modal claim ‘must’ does not follow. For in many circumstances, it may be more harmful to the wrongdoer if a loving person were simply to forgive without respect to any conditions at all. Part of the desired outcome in forgiving someone is that they learn not to act in the way that they did. It may be possible for a loving and gracious person to forgive freely in such a way that is morally admirable or permissible. But in most practical cases, simply forgiving a particular person freely and without conditions might actually be unbeneficial for them in the long run, and hence unwise. Wrongdoers will never learn how to be better people if we never hold them accountable for their actions, and require them to confess and commit to turning from their wrong course of action to a right course of action. Loving forgiveness may permit free forgiveness without conditions, but it is more likely than not that a loving and gracious person who is also wise will impose certain, fair conditions onto someone as prerequisite to their being forgiven as a way of teaching that person to do better and to live well. That would most certainly be the most loving thing to do.

What does this all say about foregoing resentment as a necessary condition for morally permissible/responsible forgiveness? Well, Griswold’s conditional account of forgiveness makes it so that the offended party’s resentment is unwarranted and thus forgiveness is the repentant wrongdoer’s due. Allais’ elective account of forgiveness makes it so that you can resent a person wholly and completely but still choose to forgive them. The so-called covenantal account of forgiveness, however, seems to allow for something like Allais’ view, but which is restrained and directed on the basis of biblical agape love and grace. On the covenantal account, it seems implicit that a person who forgives another may, in love, commit themselves to trying not resent that person, and so forgive them freely, and incrementally mitigate that resentment to a stage where there no longer is any resentment.[16] For forgiving somebody in love and with grace implies that we do not treat them as their sins deserve. The wrongdoer may deserve to be paid back and punished. But the forgiver, freely and of his own will, decides not to punish them in that manner, and instead actively seeks restoration and reconciliation with them. The loving and gracious person will forgive a person under just and fair conditions, and in kindness seek the other’s good. Then he will commit himself, as the offended person, not to resent the person who wronged him.

In conclusion, then, we considered separate accounts of forgiveness. On the one hand, Griswold’s conditional view—according to which P is obligated to forgive V on the grounds that V has fulfilled some set of conditions which then undermines P’s resentment of V, which thus obligates P to forgive V—leads to the strange implication that, if P is wronged by V, and V is somehow placed in a situation where V cannot possibly fulfil the conditions, such as in a comatose state, then P cannot possibly choose to forgive V, and must carry around a burden of unforgiveness. On the other hand, Allais’ elective view—according to which no matter how much P resents V, and no matter what V does, P can still choose to forgive V—suffers from the strange implication that it would also be perfectly within P’s moral rights not to forgive V even if V did every conceivable thing to confess and make reparation towards P; i.e., it would become morally permissible for P to be an unforgiving person. Both seem wrong, but both have attractive features. Thus, we have proposed a hybrid account that attempts to combine the strengths and avoid the weaknesses of each view. The covenantal view states that, if V wrongs P, and P justly resents V, P may freely elect the conditions under which P forgives V. But not just any conditions will do. P will set conditions which are in accordance with both grace and agape love. With respect to resentment, it turns out that P need not immediately cease resenting V. Indeed, on reflection, this may be psychologically impossible. But P can choose to treat V in such a way that is gracious and undeserved, with the aim of ultimate reconciliation between both P and V.


[1] Lucy Allais, ‘Elective Forgiveness’ in International Journal of Philosophical Studies 21:5 (2013), pp.637-653.

[2] Charles Griswold, Forgiveness: A Philosophical Exploration (Cambridge University Press: 2007).

[3] I do not know if the term “covenantal” is used in the forgiveness literature. But for the sake of this paper, I am going to use the term idiosyncratically to refer to that hybrid view that forgiveness is both free and conditioned, where this third-way position is contrasted with an absolutely elective account, quasi-conditional, but is also opposed to certain essential key elements in Griswold’s conditional account.

[4] Glen Pettigrove, Forgiveness and Love (Oxford University Press: 2013).

[5] Pamela Heironymi, ‘Articulating an Uncompromising Forgiveness’ in Philosophy and Phenomenological Research Vol. 62 No. 3 (2001), pp.529-530.

[6] Ibid. pp.530-531.

[7] Griswold (2007), pp.49-51.

[8] Allais (2013), pp.638, 641.

[9] Ibid. p.642.

[10] Ibid. p.643.

[11] Ibid. Emphasis in the original.

[12] Exodus 21:23-24.

[13] Pettigrove (2013), p.126.

[14] Ibid. p.103.

[15] Ibid. Emphasis added.

[16] This is similar to Bishop Butler’s view, but with the added element of actually aiming to rid oneself entirely of resentment at the end of the process.

Allais, Lucy. ‘Elective Forgiveness’ in International Journal of Philosophical Studies 21:5 (2013), pp.637-653.
Griswold, Charles. Forgiveness: A Philosophical Exploration (Cambridge University Press: 2007).
Heironymi, Pamela. ‘Articulating an Uncompromising Forgiveness’ in Philosophy and Phenomenological Research Vol. 62 No. 3 (2001), pp.529-530.
Pettigrove, Glen. Forgiveness and Love (Oxford University Press: 2013).

The Problem of Evil and Alternative Concepts of God

PDF: Thought and Hyperint. seminar paper 2014

On a classical theory of concepts, we stipulate the meaning of a term with respect to a set of jointly necessary and sufficient conditions that must obtain in order for the object that our term denotes to count as being the thing that it is. But not just any conditions will do. One of the fundamental problems with the classical theory is that it often commits us to postulating highly intuitive accounts of what it takes to be something, but then allowing for either too little or too many things to count as that thing. For example, one might take the concept of a ‘chair’ to be a physical thing that one can sit on. That is, a ‘chair’ is any given x such that x has the following two properties: {is physical, can be sat on}. But then too many things will count as a ‘chair.’ Tables, alarm clocks, rooftops, human bodies, mirrors and carpets all share these two properties, but none of these are chairs. At most, {is physical, can be sat on} would count as necessary but not sufficient conditions for something to count as a chair. So we may make our definition more specific. Perhaps x is a chair if and only if x has the following two properties plus at least one new one: {is physical, can be sat on, has a certain shape/size/context}, where that shape/size/context is of a particular form so as to preclude from being chairs those objects that are obviously not kinds of chairs, and so on. For example, perhaps you think a chair is something only if it has an L-shape of some such description, such as having a back and having four legs. But then that would preclude things like bar-stools, or three-legged chairs, which are obviously types of chairs. It also has the odd implication that if we got a thing precisely specified in the shape that you desired, say an L-shaped object with four legs, but then cut a leg off, then the object it would cease to be a chair. But that does not seem right either. Examples like this could be multiplied. So a classical theory of concepts seems to make extraordinarily ambiguous just what anything in the world is at all. This may lead someone to reject the classical theory of concepts.

But perhaps this kind of response is wrong-headed. Perhaps such a process of trial and error with elimination for clarification is just what we want in making sense of complicated and philosophically interesting concepts. For we can all identify chairs and tables and other paraphernalia in ordinary situations according to how they are signified by others in a social context. But when it comes to philosophically interesting subjects such as time, the mind-body problem or, the topic of this paper, God, it may simply be that the concept under consideration requires just that kind of complicated process of reflection that the classical theory of concepts commits us to given the kind of world that we live in.

It seems we can apply this intuition when talking about God. Give the world that we live in, it may be that certain a priori conceptual stipulations of God simply will not work. For example, on Classical Theism (CT), God is stipulated as an eternal, personal being who is also maximally great.[1] For whatever good and desirable property there can be, God has that property to the fullest and maximal extent. Take just three examples of desirable properties: power, knowledge and goodness. For God to be maximally great in these regards, God would have to be all-powerful, all-knowing and all-good; i.e., omnipotent, omniscient and omnibenevolent. CT postulates the existence of one such eternal, personal being.
To clarify these three technical terms, which we shall call the ‘omni-properties’:

Omnipotence =def. for all x, x is omnipotent iff x can do all things;
Omniscience =def. for all x, x is omniscient iff x knows only and all true propositions;
Omnibenevolence =def. for all x, x is omnibenevolent iff x is morally perfect.

Omnipotence (O1) implies that the omnipotent being can bring about any state of affairs anywhere in the whole world that can be brought about, provided it is logically or metaphysically possible. Omniscience (O2) implies that the omniscient being knows all true propositions and believes no false propositions. Omnibenevolence (O3) implies that the omnibenevolent being has perfectly morally good affections towards all the creatures which are of such a manner that it seeks the good of the creatures in such a way that is beneficial to them, and in this sense omnibenevolence is often connected with love, such that the omnibenevolent being would act in such a way that is in keeping with what we should expect how any good and loving being should act. We may call this kind of being the ‘omnigod.’ If CT is true, then the omnigod exists.

But there is also quite arguably another fact in the world, namely, that evil exists. That is to say, there are states of affairs in the world where the creatures suffer loss or harm in accordance with either the privation of sufficient good in the lack of benefit bestowed on the creatures, or the presence of injuries inflicted on the creatures either by their fellow creatures or by the natural world in terms of natural disasters, diseases, or desperately cruel and immoral activity enacted by one creature on another, and so on and so forth. Prima facie, then, the creatures of the world experience evil in the world in such a way that it would be rather unexpected for them to experience it if the omnigod existed. For the omnigod, (so goes the argument), would have the ability, the knowledge and the moral will to end all evil forever and ever; nay, the omnigod would plausibly never permit evil to come to pass in the first place, much less the very possibility of evil. In short, if the omnigod exists, then evil does not exist. But evil does exist. Therefore, the omnigod does not exist. So CT is false. Letting (G) stand for the proposition “Classical Theism is true” and (E) stand for the proposition “Evil exists” we may formulate an argument as follows:

  1. If G, then not-E.
    2. E.
    3. Therefore, not-G.Modus tollens. This is a standard form of the Problem of Evil (POE) argument against CT.[2] Given (1.) and (2.), one is committed to (3.). Of each premise, (2.) seems to be most obvious. So the debate should rest on premise (1.).

    If it follows on CT that the God specified by the proposition (G) must be maximally great, and if maximal greatness implies that the omnigod has the omni-properties specified above, then that would in turn motivate one to believe that premise (1.) in the preceding argument is true. But then it seems to turn out that CT is false. Yet it does not seem from the falsity of CT that something quasi-godlike could not remain, given evil. For on the preceding argument, it follows that the first premise may be explicated in this way:

    1*. If there is an x, such that (x is O1 and O2 and O3), then not-E.
    2. E.
    3*. Therefore, there is not an x, such that (x is O1 and O2 and O3).

    But (3*), being a negated conjunction, logically implies some disjunction with at least one false disjunct, namely, premise (4) that:

    4. Therefore, there is an x, such that (x is either not-O1 or not-O2 or not-O3).

    Which is to say:

    5. If not-(x is O1 and O2 and O3) then (x is either not-O1 or not-O2 or not-O3).

    And the conclusion (5) will count a true under tge negation of (G).

    On CT, this would prove to be a disaster For God, on the CT view, just is the (O1 and O2 and O3) being. But for us who may want to maintain some form of Theism, this need not strike us as overly problematic, even granted the argument. For there are different ways to think about maximal greatness, some conceptions of which do not necessarily imply that the omni-properties as specific are required. That is, it sounds plausible that we may accommodate some modifications of our concept of ‘God’ given evil, such as in accordance with what maximal greatness entails, but which also takes into an account an argument like the POE. Perhaps we may be able to use the POE and use it, not as a way of negating Theism altogether, but as a way of clarifying exactly what maximal greatness entails. Perhaps we can formulate a concept of God that is sufficient recognisably to count as Theism but which is non-classical; a Non-classical Theism (NCT). That is to say, it seems we can launch a defence of a form of Theism—NCT—if we explicate what it takes to be maximally great given evil. If it turns out that the God of (G) is not possible given evil, it may still be that some kind of God exists, and that some other proposition, “Non-classical Theism is true” (G*) is true. If God is in fact stipulated to be that eternal, personal being who is also maximally great, then it may follow, given evil, that the conjunction of the three properties of omnipotence, omniscience and omnibenevolence is not required for maximal greatness. Perhaps a maximally great being can be a maximally powerful being, but not an omnipotent being, where ‘maximal power’ is somehow less than strictly omnipotence but very, very powerful in such a way that may be said to be quasi-omnipotent. The same could be said for knowledge and moral goodness. Perhaps what it takes to have these attributes maximally is not the same as having them so as is specified precisely by the relevant omni-property, but rather to the maximal extent given the actual world. I am suggesting that we may make sense of maximal greatness, given evil, in such a way that does not require the conjunction of these three omni-properties.

    Let us represent CT by the proposition (G), NTC by the proposition (G*), and the POE by the proposition (E). Both (G) and (G*) postulate slightly different concepts of God. We may represent this on a 2D semantic graph to show how evil does not necessarily commit one to the negation of Theism altogether:

G   G* E
  G* F T T

In the world where CT is true (G) it is not true that evil exists (E), and the same vice versa, as per the POE argument above. Similarly, on (G) NCT (G*) would not exist, despite the fact that, whatever NCT implies, the omnigod of CT encapsulates whatever the God of NCT has, since on NCT there just would not be an omnigod like CT specifies, and so we judge these two accounts (G) and (G*) to be mutually incompatible. According to our table, on NCT it is possible for both (G*) and (E) to be true. On these worlds, it is true that evil exists and NCT is true, provided Theism in general can accommodate NCT, as it seems that it can. Here, then, is the significance of the graph in a discussion about God and the POE: evil does not necessarily imply that there no God at all, and so the original argument attacks only a particular form of theism—CT—and not Theism simpliciter. At most it would follow that we would have to modify our stipulation of what maximal greatness implies, and so clarify our concept of God. For a God is much more than an omnipotent, omniscient and omnibenevolent being. God is, perhaps, an eternal, personal First Cause who has creative purposes which may or may not involve evil, and we may want to retain our belief in such a being given other fact about the way the world is. We can then use 2D semantic graphs like this to clarify issues in Philosophy of Religion where conceptual clarification is involved in the defence of a view like Theism, and provide a legitimate alteration of our concept of God given the world that we live in, provided the former stipulation that was suggested is not metaphysically necessary for the being in question.

Laura Schroeter’s article in the Stanford Encyclopedia of Philosophy provides some interesting discussion surrounding why 2D graphs such as the one presented can guide us in issues of conceptual clarification and resolutions of problems such as is the case in a debate about God and the POE.[3] In addition to clarifying the structure of our semantic understanding, the 2D framework can help justify specific conceptual analyses of Theism. Schroeter explains how the criteria that implicitly guide our everyday use of a term like ‘God’ are often embodied in recognitional or inferential dispositions rather than in consciously accessible rules. We can compare our ability to explicate reference-fixing criteria for particular expressions to our ability to recognise the grammatical accuracy of sentences in our own language.[4] Linguists are able to construct grammars for linguistic expression grounded on one’s judgments about the acceptability of particular sentences. Similarly, one can also construct an analysis of the meaning of a term such as ‘God’ on the basis of the application of the term in hypothetical cases, such as when evil exists in a world where God, also, should exist (for the committed Theist). The correct analysis must then capture the full range of one’s confident judgments involving the target expression, while abstracting away from performance errors and other interfering factors.[5]

The 2D framework, Schroeter explains, is particularly helpful for providing an analysis of the meaning of proper names or natural kind terms, because it prompts one to consider possible cases in two different ways: as actual or as counterfactual. One’s point of view on whether or not the God of NCT counts as a maximally great being, for instance, depends on whether one thinks of NCT as a hypothesis about what is the case in the actual world or as a purely counterfactual possibility. When the A- and C-intensions for the term ‘God’ diverge 2D empiricists suggest that the correct analysis of the term will take the form of rigidified definite descriptions using the indexical term ‘actual’: God =def. the actual thing that has such-and-such a set of properties. Moreover, careful attention to one’s reactions to the hypothetical scenarios where God has alternative properties will allow one to make explicit which properties implicitly motivate one’s application of the term ‘God’. For instance, you may discover that your implicit criterion for applying ‘God’ is that God =def. the actual supernatural kind that is personal, maximally powerful, maximally knowledgeable and maximally good. Alternatively, your use of the term ‘God’ may be guided by the dominant cause of your community’s past use of the term ‘God’. It is possible that standard causal theories of reference are based on this method of conceptual analysis. The conceptual analyses produced by this method count as a priori, according to the 2D empiricist, because you can know them to be correct “independently of knowing what the actual world is like.”[6] The evidence that supports such analyses consists in purely hypothetical judgments: judgments about how to classify cases on the supposition that your environment is thus or so. Given the fact that such hypothetical judgments don’t require you to determine what your real environment is actually like, your justification for accepting an analysis is not based on empirical knowledge. And to change your judgment about a purely hypothetical case would be to change the meaning of your term. According to empiricist 2D semantics, the truth of any metaphysical reduction depends on the reference-fixing conditions conventionally associated with the terms involved—and these reference-fixing conditions are always knowable a priori. So a priori conceptual analysis plays a crucial role in vindicating metaphysical reductions.[7]

Metaphysical reductions, says Schroeter, provide a constitutive account of some target domain—for example: beliefs, free will, water, moral rightness—in terms of more basic features of the world such as the properties postulated by an idealised physics, the perfect and complete conceptualisations of divine ideas, or the complex mosaic web of sense data. A Physicalist about mental states, for instance, is committed to there being specific facts about the microphysical structure of the world that suffice for the existence of beliefs, desires and sensory experiences. The Physicalist is thus committed to metaphysically necessary ‘entailments’ connecting claims about the two domains. That is, it is metaphysically necessary that if such-and-such physical facts obtain, then such-and-such mental facts obtain. This metaphysical entailment relation can arguably be cashed out in terms of global supervenience.[8] On the empiricist view of meaning, the relevance of conceptual analysis to metaphysics is then straightforward: conceptual analysis establishes that a putative reduction respects the original meaning of the target expression.[9] Successful reductions must be in accordance with our original shared understanding of the target expression, and they must elucidate this original understanding. This seems to be so in the case of the God of CT versus the God of NCT in light of the POE. If conceptual analysis of what it takes to be a God are knowable a priori, it follows that metaphysical reductions must always be backed by a priori entailments between the base-level claim (such as ‘evil exists’) and the target claim (such as ‘God exists’). Conceptual analysis thus plays a modest role on the empiricist account. A priori conceptual analysis merely tells you about the relations among your ideas; it cannot tell you whether there are any objects, kinds, or properties that satisfy your current reference-fixing assumptions.[10] Moreover, the meaning you currently associate with a term may be pragmatically deficient: e.g., it may not be determinate enough to settle certain hard cases or it may not allow you to draw useful distinctions in your actual environment. In such cases, you would have good pragmatic reasons to change the meaning of your term.[11] What the empiricist denies is that changing your current criteria for applying a term can ever get you closer to the truth about what you meant all along.

So where does this leave us concerning a legitimate alteration of our concept of God given the POE? It shows us that it is possible that there can be legitimate ways of making sense of an eternal, personal being who has powers, knowledge and moral goodness all to the maximal degree even given evil. Perhaps omnipotence, for example, is incompatible with there’s being evil. But perhaps maximal power is not. Perhaps God condescends and self-limits himself in order to enter into loving relationships with human beings. This is merely a possibility. But if there is indeed such a possibility, such that there are logically necessary reasons or divinely willed contingent purposes for which God might alter or restrict his own power, knowledge or goodness, then it seems we can accommodate some coherent concept of a maximally great being even given the evil in the world, and thus not be committed to a negation of Theism altogether, even if it would negate CT. It may even be that maximal greatness requires us to formulate alternative accounts of divine power, divine knowledge and/or divine goodness given the evil in the world, yet still maintain a non-classical concept of Theism.

Somebody might object that if we reduce the concept of God to a certain degree, that reduction will either negate any meaningful of God, or it will be arbitrary precisely as to what degree one ought to alter any given omni-property. But it seems to me that these objections to the general point in this essay—which is to explore legitimate ways we may theoretically alter our concept of God and maintain a kind of sufficiently robust Theism in the event that we think the POE is a good argument against a certain conception of God—do not hit the mark sufficiently well. With respect to the first, if a new concept of God is created to deal with the issues inherent in the POE, and that alteration is legitimate, then legitimacy will surely imply that our concept is meaningful. As for God, the idea seems to be that there is some kind of eternal, good, moral being who is also the cause or sufficient reason for the existence of everything else that exists, who is also very powerful and highly knowledgeable. That is a coherent concept. Notice we did not specify the extent to which these properties go, but placed them in a high range. Moral goodness implies the being does the right thing. Being very powerful seems to speak to the being’s being able to do a lot, though perhaps not absolutely everything. And having a high amount of knowledge, perhaps the most knowledge of anything else that does exist, implies being very intelligent and able to consider things in ways we have not before, in such a manner that the being is (so far as we can see) virtually omniscient. There is nothing incoherent about this concept, and it does seem meaningfully godlike.

As to the second objection about the arbitrariness of specification of the properties on our new concept of God, this does seem more difficult. But even here it seems to be right to say that the specification of the properties will depend on which property you think can be sacrificed and in what degree necessary in such a way that is sufficient so as to maintain a sufficiently godlike being given the evil in the world. For instance, one might suggest like Stephen Law has that perhaps there is an Evil God who takes pleasure in evil and suffering.[12] In that case, omnibenevolence would not be required, and evil is easily accommodated. On the other hand, John bishop has suggested that perhaps God is freely self-limiting and thus freely not omnipotent in order to sustain divine omnibenevolence and enter into loving relationships with the free creatures.[13] Or perhaps we may agree with certain doctrinally revisionist Christian philosophers and theologians that God’s omniscience is not quite like we believed it to be, and that God in creating (libertarian) free beings could not, by logical necessity, know what they would freely do in the world at future times, and thus cannot arguably be put to blame for his good intentions in forming a creation which turned sour due to human free will.[14] Whatever we choose to deny, it will have to be in keeping with something which is sufficiently godlike to count as a maximally great being as an omnigod, and which omni-property can be altered and to what degree this can be so will depend on each person’s relative intuitions.

In conclusion, then, it does seem as if we can think of a means through 2D semantics by which we can alter our concept of God legitimately so as to accommodate a belief in NCT in the case where we think an argument against CT from the POE goes through after all. Whereas the original concept of the omnigod (G) is incompatible with both an alternative concept of God (G*) and the existence of evil (E), nevertheless, both (G*) and (E) are mutually compatible, and it is not obvious that (G*) is inherently arbitrary or meaningless, or insufficiently theistic. Therefore, we may theorise an alternative stipulation of our concept of God that is sufficiently robust so as to count as a form of Theism using 2D graph-work, and that stipulation will include variations of one or more of the original omni-properties, in such manner that it follows that it is not necessarily true that the existence of evil in the world implies that God absolutely does not exist.


[1] William Wainwright, ‘Concepts of God’ in The Stanford Encyclopedia of Philosophy (Spring 2013 Edition), Edward N. Zalta (ed.), URL = .

[2] Michael Tooley, ‘The Problem of Evil’ in The Stanford Encyclopedia of Philosophy (Summer 2013 Edition), Edward N. Zalta (ed.), URL = .

[3] Laura Schroeter, ‘Two-Dimensional Semantics’ in The Stanford Encyclopedia of Philosophy (Winter 2012 Edition), Edward N. Zalta (ed.), URL = .

[4] Frank Jackson, ‘Reply to Yablo: What do we communicate when we use ethical terms?’ in Philosophical Books (2000), 48: 24–29.

[5] Frank Jackson, From Metaphysics to Ethics: A Defence of Conceptual Analysis, Oxford: Oxford University Press. 1998, pp.31-41.

[6] Ibid. p.51.

[7] Frank Jackson, ‘Armchair Metaphysics’ (1994) in Meaning in Mind, M. Michael and J. O’Leary-Hawthorne (eds.), (Kluwer Academic Publishers: 1994), pp. 23-42.

[8] Frank Jackson, From Metaphysics to Ethics (1998), pp.6-14.

[9] Ibid. p.28.

[10] Ibid. pp.42-44.

[11] Ibid. pp.44-46, 53-54.

[12] Stephen Law, ‘The evil-god challenge’ in Religious Studies (Cambridge University Press: 2009), pp.1-21.

Available on CJO doi:10.1017/S0034412509990369. Granted, this article is essential a challenge to the moral argument in favour of Theism and is thus not inherently a case for the actual existence of such a deity. I would also argue that an Evil God is ultimately an incoherent concept when applied with our ordinary moral intuitions. But I do take it to be an example of a concept of a quasi-godlike being with two out of three of the omni-properties in such a way as would make sense given evil: there is an all-powerful, all-knowing, wholly wicked Evil God who takes joy in suffering. That would certainly make the Problem of Evil less forceful if true!

[13] John Bishop, ‘The Argument from Evil and the God of “Frightening” Love’ in Sophia (2013) 52:45-49, DOI 10.1007/s11841-012-0344-y.

[14] Here, I am referring to Open Theism, the view that divine foreknowledge and human (libertarian) free will are logically incompatible. But since human beings do have such free will, God cannot foreknow what they will freely do. Thus, God “risks” when creating the world of free beings whom he is able to lovingly enter into a saving relationship with, and cannot be held responsible for the freely chosen decisions of the creatures which bring about evil and suffering. This view does not compromise omniscience, as such. It simply shows that divine knowledge does not include truths about future free acts of creatures. Cf. e.g. John Sanders, The God Who Risks: A Theology of Divine Providence (IVP Academic: 2007).

Bishop, John. ‘The Argument from Evil and the God of “Frightening” Love’ in Sophia (2013) 52:45-49, DOI 10.1007/s11841-012-0344-y.
Sanders, John. The God Who Risks: A Theology of Divine Providence (IVP Academic: 2007).
Schroeter, Laura. ‘Two-Dimensional Semantics’ in The Stanford Encyclopedia of Philosophy (Winter 2012 Edition), Edward N. Zalta (ed.), URL = .
Tooley, Michael, ‘The Problem of Evil’ in The Stanford Encyclopedia of Philosophy (Summer 2013 Edition), Edward N. Zalta (ed.), URL = .
Jackson, Frank. ‘Armchair Metaphysics’ (1994) in Meaning in Mind, M. Michael and J. O’Leary-Hawthorne (eds.), (Kluwer Academic Publishers: 1994).
____________. From Metaphysics to Ethics: A Defence of Conceptual Analysis, (Oxford University Press: 1998).
____________. ‘Reply to Yablo: What do we communicate when we use ethical terms?’ in Philosophical Books (2000), 48: 24–29.
Law, Stephen. ‘The evil-god challenge’ in Religious Studies (Cambridge University Press: 2009), pp.1-21
Wainwright, William. ‘Concepts of God’ in The Stanford Encyclopedia of Philosophy (Spring 2013 Edition), Edward N. Zalta (ed.), URL = .

Defining Descartes’s Doctrine of Consciousness

PDF: Seminar Paper [EModP&C]

* * *

According to Alison Simmons (2012, p.3), René Descartes never defines the term ‘consciousness.’ Yet, throughout his writings, Descartes ascribes both thought and mental cognition to some things, such as humans, God and angels. Yet he denies the same attribution to other things such as automatons and animals. Things may have perceptions, sensations and physical, mechanistic functions, but Descartes does not believe one can derive consciousness from these kinds of non-thinking physical aspects alone. Whatever consciousness is, it is not reducible to physical, mechanistic functions and perceptions. Consciousness therefore must either depend upon or be related to thinking in some way. This paper seeks to provide an interpretation and analysis of Descartes’s doctrine of consciousness from a selection of key texts found in Descartes’s writings, in conjunction with certain key issues raised primarily by three separate recent works.

To do this, we shall work in three steps. First, special attention will be paid to Gordon Baker and Katherine J. Morris’ (2002) discussion surrounding the exegesis of the word cogitatio, where it has been suggested that this key phrase of Descartes’s may extend beyond mere thought to include all mental activity in its seventeenth century context. Secondly, we will assess Janice Thomas’ (2006) claim that Descartes’s may allow different kinds of consciousness in animals to try to discern the differences in the cognitive lives of animals and human beings, especially given the wide array of neuro-physiological properties shared in common between human and non-human animals. Thirdly, we will then extend the account of human consciousness and cognition in Descartes to include distinctive features such as second-order reflexivity, as Vili Lähteenmäki (2007) has suggested. Ultimately, it will be concluded that, for Descartes, the conscious thing simply just is the thinking thing: thought characterised by second-order reflexivity and awareness, which should be thought about in terms of what characterises human thinking in second-order reflexive terms.

First, Baker and Morris (2002) maintain the thesis that Descartes’s term cogitatio (French: pensée; trans. ‘thought’) may include all mental cognitive activity. Consciousness, in Descartes, is closely related to the conception of the mind and the mental. Thus in his Meditations, Descartes, by method of doubt, reduces a priori the human person to an Archimedean point of reference, whose essence is thinking (cogito) which, by axiom, requires its being or existence (sum).[1] For Descartes, the agent is, indubitably (CSM, II, p.19): “A thing that thinks… A thing that doubts, understands, affirms, denies, is willing, is unwilling, and also imagines and has sensory perceptions.” I, the agent, am a thinking thing, and my thought is characterised by all kinds of phenomenal and cognitive phenomena. Notice how Descartes mentions only those kinds of activities which could only be performed by rational, intelligent agents: doubting, understanding, willing, denying, imagining and sensory perceptions.[2]

This intentional, reflective intelligence characterises Descartes’s conception of thought in general. In the Meditations (CSM, II, pp.50-51), ‘imagination’ for instance is the application of a cognitive faculty to an object of perception which is immediately present to a person by perception. Imagining is a reflective, intentional and intellectual activity of a person to an object of sense-perception, and is a reflective, intelligent activity by which the mind comprehends objects, which may be both internal objects like thoughts, and external objects like trees and bodies. If a person perceives a triangle through a sensory perception, one’s physical body captures the shape, which, being apprehended by the mind, is transferred by way of the imagination into the understanding so as to create a conscious, sense-perception of the triangle. This faculty of the intellect may require more personal effort given the complexity of the entity being perceived.

Mental complexity is also demonstrated in Descartes’s May 1638 letter to Reneri (CSMK, III, pp.96-97), where Descartes defines ‘thought’ in terms of “all the operations of the soul.” In the Passions (CSM, I, pp.345-358), this soul is distinguished into two parts: ‘higher’ or ‘rational’ soul, which is the seat of the will, volition and intelligence; and ‘lower’ or ‘sensitive’ soul, which is the seat of the appetites and base desires, which are the receptible sensations and physical perceptions, these being apprehended by the higher or rational part of the soul to form an understanding of the world.  The mind features in the relationships between these distinctions in the soul in that the mind dictates between them to form rational behaviour in accordance with the will of the creature, either willingly restraining or being propelled by his appetites or base proclivities to act in certain ways. The mind, it seems, is in some kind of an active, higher state governing over these two aspects of the soul, and it has awareness over each one, both actively and intelligently moving the agent in relation to them.

In light of Descartes’s doctrine of the complexity of mental cognition, consider the claim made in Baker and Morris (2002, pp.13-14) that the Latin noun ‘cogitatio’ (French: pensée), often translated ‘thought,’ covers more than our ordinary modern English understanding and use of the word ‘thought’ naturally covers. Cogitatio, in its standard, seventeenth century usage, covers not only ratiocination, but also “any form of conscious state or process or activity whatsoever.” They continue:

            “The term ‘thought’—pensée, cogitatio—had in Descartes’s time a much wider meaning than it has now. It embraced not only ‘thought’ as it is now understood, but all mental acts and data. Indeed, on this basis, ‘cogitatio’ is sometimes translates as ‘experience’ or ‘consciousness.’ […] The idea that Descartes expanded the term ‘cogitatio’ to cover all (and only) ‘states of consciousness’ we will call the ‘Expansion Thesis.’”[3]

If Baker and Morris are right about the Expansion Thesis, it seems like Descartes’s doctrine of consciousness can be derived from his doctrine of thought, such that whatever characterises thought also characterises consciousness, and vice versa. For interpreting ‘thought’ as ‘consciousness’ allows consciousness to be characterised by whatever thought is characterised by. By implication consciousness would then have to be interpreted so as to exclude any form of physical basis for its existence, since, for Descartes, the essence of mind is thought and the essence of the physical body is extension. If consciousness simply just is thought, then its essence is also thinking, too. The body, then, could not be conscious, regardless of its complex physical arrangements; consciousness could neither emerge from nor supervene on any particular arrangement of material particles. Anything that is physically extended could be interpreted with reference to mechanical principles alone without thought and without consciousness. Baker and Morris (2002, p.17) call this the consciousness/clockwork distinction—Descartes distinction between the life of the mental and the mechanical functions of the body.

Baker and Morris cite two primary texts in support of the Expansion Thesis. Firstly, they cite the Principles (2002, p.31; cf. CSM, I, p.195) thus (emphasis original; square bracketing removed):

            “By the term ‘thought’, I understand everything which we are aware of as happening within us, in so far as we have an awareness of it. Hence, thinking is to be identified not merely with understanding, willing imagining, but also with sensory awareness. For if I say ‘I am seeing, or am walking, therefore I exist,’ and take this as applying to vision or walking as bodily activities, then the conclusion is not absolutely certain. This is because, as often happens during sleep, it is possible for me to think that I am seeing or walking, though my eyes are closed and I am not moving about; such thoughts might even be possible if I had no body at all. But if I take ‘seeing’ or ‘walking’ to apply to the actual sense or awareness of seeing or walking, then the conclusion is quite certain, since it relates to the mind, which alone has the sensation or thought that it is seeking or walking.”

Secondly, similarly from the Meditations (2002, p.31; cf. CSM, II, p.19):

            “But what then am I? A thing that thinks. What is that? A thing that doubts, understands, affirms, denies, is willing, is unwilling, and also imagines and has sensory perceptions.”

Taken together, these two citations seems conclusive for the truth of the Expansion Thesis.[4] The first citation from the Principles shows that Descartes understands the word ‘thought’ (cogitatio, pensée) to include just those kinds of intentional, intelligent, reflective activities we have already mentioned. In particular, take note of sensory awareness. I may have a brute sensations from the sensitive part of the soul, my body has a certain reaction which causes my mind or rational soul actively and intelligently to apply the imagination to that mere sensation or physical perception, to form an understanding or comprehension of the object of my physical sensations, and thus I can be said to have a sensory awareness and understanding of the object of my sensations. ‘Thought,’ then, includes a kind of third tier where I, as an agent, possess an awareness of my sensing things, which provides me an ability as a thinking creature to reflects on my sensations, and to perceive that I perceive, to understand that I understand, and to bring about an awareness that I am aware (Lähteenmäki, 2007, pp.181-182, 183-186).[5] For it would just follow, given that I am a conscious, thinking thing, that to truly perceive in the higher, rational sense, I would be conscious that I do, when my intellect is applied in the right kind of way.

But what about nonhuman animal creatures? Let us briefly consider their cognition by way of comparison with human beings, to clarify Descartes’s understanding of human consciousness. For although animals, to all appearances, exhibit the same kind of neuro-physiological properties such as perceptual and sensory awareness that we humans do, yet Descartes did not believe that animals are consciously aware in the same way that we humans are, if at all. This does not mean that animals do not have neuro-physiological properties like physical perceptions and bodily sensations, or that their bodies cannot move in complex ways as ours do. But clarifying just what it is about animals that makes them able to move as machines and have perceptions but not be in possession of conscious thought will help us be able to refine just what it is about us mind-possessing human creatures that makes us conscious in a way that animals are not, and therefore to clarify Descartes’s doctrine of consciousness in general.

Janice Thomas’ (2006) research serves as a challenge to my view that animals are not conscious in Descartes. Thomas believes that Descartes essentially delivers a doctrine of different kinds of consciousness. To animals belong certain kinds of consciousness; to the humans, another and more special kind. She distinguishes between five different kinds of consciousness (2006, pp.246-356): (1) ‘Organism Consciousness,’ which is different degrees of functioning, like being awake or asleep; (2) ‘Perceptual Consciousness,’ a form of sensory awareness   which may allow an animal to navigate but without reflective, rational thought or reason; (3) ‘Introspective Consciousness,’ which is reflexive, self-consciousness; (4) ‘Access Consciousness,’ which is access to another agent’s intention to act in a certain way and a preparedness to respond to that potential activity; and finally (5) ‘Phenomenal Consciousness,’ which is mental qualia, i.e. the felt reality of conscious experience. “Clearly,” says Thomas, “Descartes’s opinion is that animals, including humans, have inner states of feelings which can be expressed in outward signs…” (ibid, p.343). And again (ibid. p.363), “My overall answer to my title question [‘Does Descartes Deny Consciousness to Animals?’] [is] that Descartes never denies organism consciousness, perceptual consciousness, access consciousness or phenomenal consciousness to non-human animals. The consciousness Descartes does deny to them is merely (not that this is a small thing) the capacity for self-conscious reflective receipt or awareness of one’s inner states which he [Descartes] believes belongs uniquely to the fortunate (and immortal) animals endowed with rational souls.”

The problem with this analysis (as Thomas herself seems tacitly to admit) is that this consequence of her research is not mere at all. What Thomas has just told us is that animals possess, inter alia, certain sensory, perceptual and phenomenal qualities which do not amount to the kind of consciousness Descartes also wants to ascribe to human beings who are endowed with rational souls (thinking things), which is a kind of self-awareness of the intelligent, deliberate, introspective sort. But Thomas never justifies her choice of the word ‘consciousness’ to describe these kinds of neuro-physiological properties ascribed to animals. She kind of just assumes it. But even if she could justify it, then it seems the properties she is attributing to animals to allow them to share in some sort of consciousness is just not the interesting sense of intelligent, self-reflective consciousness ascribable to human creatures with rational souls. Simmons (2012, pp.13-14) allows for the possibility for varieties of consciousness in Descartes, but I am not convinced that this supports Thomas on the basis of other, stronger, primary texts which seem to deny thought to animals altogether, and thus consciousness. Descartes surely takes very careful precautions to describe the functions of material bodies in relation to immaterial minds, so as to be able to explain how animals move and act in complex ways but without thought. In the final analysis, one is simply not convinced that Thomas’ attribution of the word ‘conscious’ to the kinds of bodily states ascribable to animals in the way that she does is quite accurate on a strict reading of Descartes’s primary text source material. However the properties she ascribes to the bodily operations of the animals are correct and helpful. My biggest concern with Thomas’ account here is that she never argues from Descartes himself that the particular kinds of properties exhibited in each case are describable as therefore being ‘conscious.’ Worse, Descartes himself seems to contradict Thomas in her attribution of consciousness to animals on the basis of these neuro-physiological properties, since Descartes does not even hold that animals have minds, strictly taken as non-extended, thinking things.

Indeed, many primary texts from Descartes contradict the attribution of any kind of consciousness to animals, strictly taken, in short since animals are not thinking things, though they may be wondrously complex mechanical physical things. The main problem is that animals are not thinking things endowed with a rational soul, according to Descartes—they are not even intelligent creatures. But thought is a necessary prerequisite condition for consciousness in Descartes. In the first place, in his Discourse Descartes does not seem to think that any internal arrangements of our bodily organs could make any difference to the nature of the soul that we possess, whether it be rational or merely sensitive, vegetative, or animal. This implies that any functional arrangement of the material body cannot produce a thinking mind that is intelligent, rational and introspective; conscious, in a robust way. In physical bodies, things simply function like mechanical gears moving a clock, without need to refer to a thinking thing (CSM, I, p.134): “These [physical] functions are just the ones in which animals without reason may be said to resemble us.” This implies they do not actually resemble us internally, though they look like they do. Similarly, in the Passions (CSM, I, p.348), animals possess all the physical faculties from which arise the same kinds of sensations we have, but in lacking a rational soul they simply lack the intelligent ability as a thinking thing to determine between which passions to emphasise and which to supress and so are casually determined by “unreasoning instinct” (my emphasis) to do what they do, and so do not on free choice of a rational, acting agent choose to do what they do (Lähteenmäki, 2007, p.178). Again, in his May 1638 letter to Reneri (CSMK, III, p.99), animals are said to resemble humans in many ways such that we often think they have a rational soul as we do, implying they really do not. Yet most certainly the twin death-blow to attributing consciousness to animals on Descartes’s understanding is found in a positive argument against animal thought in his 5th of February 1649 letter to More (CSMK, III, pp.365-367) and in his denial of the same in his 11th of June 1640 letter to Mersenne and especially in his 30th of July 1640 letter to the same (CSMK, III, pp.148-150; particularly p.149; see below).

In the letter to More, Descartes rejects the argument from analogy which says since animals move and behave like we do, therefore they must have some kind of soul just as we do. Descartes instead proposes that we distinguish between what he calls the ‘corporeal soul,’ which is the body and its mechanisms, and the ‘incorporeal soul,’ which refers to the mind which thinks and exercises volitions. There, Descartes thinks everything animals do can be reduced down to the former corporeal type of soul. Indeed, Descartes can only think of one reason possibly to attribute to animals incorporeal soul or a mind, and three reasons against attributing it. In favour, Descartes noticed animals have in them all kinds of sense organs just like us, and can have all kinds of similar sensations. However, against this, Descartes says three things: first, that bodily parts can often be arranged to form movement without thought, such as uncontrollable epileptic convulsions; second, that we can imagine that men might make automatons that are capable of movement without thought; and third, animals are incapable of forming complex languages and intentional speech—they merely mutter grunts and roars on impulse, and we have no evidence that they are thinking before communicating in this way. Indeed, in his Discourse Descartes is even more explicit about his opinions concerning the implications of animals’ incapacity for complex language acquisition (CSM, I, p.140; my emphasis): “This shows not merely that the beasts have less reason than man, but that they have no reason at all.” Again, animals have, according to Descartes (CSM, I, p.141; my emphasis) “no intelligence at all”. Animals, whatever they are, are not thinking things, and therefore cannot be conscious things, according to Descartes.

Similarly, in the former letter to Mersenne, cited above, Descartes makes clear that he does not explain what appears to be the conscious feeling of pain without reference to the soul, since pain properly understood exists only “in the understanding” which requires the kind of intentional, reflective intelligence found only in minds, that is, human creatures. Hence Descartes denies animals feel pain “in the strict sense” since they lack this intelligent understanding. And this is followed up most explicitly in Descartes’s latter letter to Mersenne, also cited above, where Descartes employs more forcefully his automaton thought experiment so as to think that the inference from complex animal behaviours to animal mindedness or awareness simply does not follow (CSMK, III, p.149):

            “As for brute animals, we are so used to believing that they have feelings like us that it is hard to rid ourselves of this opinion. Yet suppose that we were equally used to seeing automatons which perfectly imitated every one of our actions that it is possible for automatons to imitate; suppose, further, that in spite of this we never took them for anything more than automatons; in this case we should be in no doubt that all the animals which lack reason were automatons too. […] [T]he bodies of animals contain all the organs which an automaton needs if it is to imitate those of our actions which are common to us and the beasts…”

The plain reading of this passage suggests that Descartes desires that we rid ourselves of the opinions that animals feel in the same way that we do solely on the basis of observing their behaviour. He likens them to lifeless yet complex automatons which are mindless and unconscious, even though they have all kinds of complex physical and mechanical functional properties. Descartes’s thought experiment from automatons is simply to reject the argument from analogy, and to refute the concept that animals must have thought simply because they behave in complex ways just like we may. What seems to follow from this is that, for Descartes, in order to think and to be conscious, it takes more than merely having complex physical make-up as animals do. The conscious, thinking thing directs the mechanisms and movements of the body, such as directing the small animal spirits, which are those smallest, physical particles which move the limbs through the pores in the body (CSM, I, p.139; CSM, I, pp.331-332, 335). But the conscious thing neither emerges from nor can be inferred by the mechanisms and movements of the physical body, no matter how complex. One therefore feels inclined to agree with Lähteenmäki (2007, p.178) as he characterises it, that animals are conscious only if they have rational souls. But animals do not have rational souls. So animals are not conscious. Descartes most certainly rejected animal thought, and therefore any meaningful concept of animal consciousness in his writings.[6]

Citations such as these therefore seem sufficient to argue that Descartes did not believe that animals, strictly taken, are conscious. But they do have sensations and perceptions, and can move (or, rather, be moved) in complex ways. They simply lack an additional necessary property which is required for intelligent, self-awareness: a rational soul or mind. The impartation of an immaterial mind or rational soul by God is a necessary condition for what it takes for something to be conscious, and this seems to support my thesis that that Descartes’s doctrine of consciousness requires us to think that the conscious thing simply just is the thinking thing, that consciousness is therefore a property of human beings, and that consciousness in Descartes may therefore be thought about in terms of what characterises human thought.

So far we have argued that Descartes’s Latin term cogitatio (French; pensée) may be extended in its seventeenth century context so as to be able to include all the elements of human thought. We also considered elements of animal behaviour and concluded that they do not strictly imply animal consciousness in any interesting and robust kind of way. Yet my thesis not only requires that the conscious thing simply just is the thinking thing, but that this thinking thing also be characterised by a form of second-order reflexivity, so as to characterise a person’s cognition with an awareness of whatever occurs in himself—an awareness than one is aware, perceiving that one perceives, and thinking that one thinks.

Here, Vili Lähteenmäki (2007) has done some helpful work in this regard. As we have already demonstrated, consciousness requires as a necessary prerequisite condition that the thing that is conscious be a thinking thing. This thought cannot be reduced to the mechanistic bodily function and sensations in which animals also share. Indeed, for humans, in Descartes, thought has the property of having second-order reflexivity: thinking about thinking, and being aware of that second-order thought. Lähteenmäki (2007, pp.187-188) takes there to be a distinction in Descartes between ‘direct’ and ‘reflective’ thoughts, which seems somewhat related to Descartes’s disambiguation of the soul into two ‘higher’/‘rational’ and  ‘lower’/‘sensitive’ parts (CSM, I, pp.345-358). A direct thought would be a first, and simple thought related to a direct perception of an entity. A reflective thought would be a higher, intellectual recognition of that entity in relation to and distinguished from other entities, but would be simultaneous with the direct thought. This finds support in Descartes’s 29th of July 1648 Reply letter for Arnauld (CSMK, III, pp.356-359; cf. esp. p.157), where Descartes distinguishes between direct and reflective thoughts in the same way one may make a distinction in vision between light’s hitting the retina of the eye of causing one to see objects, and one’s actual recognition of the fact that they do perceive: “[W]e make a distinction between direct and reflective thoughts corresponding to the distinction we make between direct and reflective vision, one depending on the first impact of the rays and the other [an agent’s actually seeing] depending on the second.” This is like unto an infant’s having a simple sensation (e.g. pain) and then goes no further, and an adult’s having that same sensation, yet simultaneously having a higher perception that one is in pain in contrast to not being in pain. This distinction seems directly related to the intellectual capacities of the agents in question, and their ability to reflect on their thoughts and feelings, as opposed simply to having thoughts and feelings.

Lähteenmäki wants to make an analogy between these cases. Just as there is a difference between simple seeing with regards to light’s reacting on the retina in the right way and seeing in regards to an agent’s seeing one thing as opposed to another given that sense of sight, so there is a difference between, on the one hand, merely thinking or being conscious, and, on the other hand, thinking that one is thinking, or being consciousness that one is consciousness. My thinking that I am thinking, and being aware of that second order relationship, is what characterises second-order reflexivity. Consciousness, in Descartes, must therefore be characterised by this kind of second-order reflexivity in order to be consciousness in the fullest, most robust sense Descartes wants to give to humans.

Despite the wealth of textual evidence in this direction, two prima facie counterintuitive consequences provide legitimate caution for heading in this direction. Firstly, with respect to Descartes own example corning the consciousness of human infants CSMK, III, p.157), they should count as not wholly-conscious on this view. This may seem unacceptable, since it would seem that human infants should be able to count as just as conscious as adult are, since human infants are thinking things just like adults are. But it seems to me Descartes should simply say that the thinking thing can learn to engage in higher forms of more complex thought, given the kind of experiences it undergoes, and the higher forms of intelligence it obtains. Maybe another way to look at it would be that the human infants really are as conscious after all, but it simply takes more experiences and intelligence for human infants to able to fully realise the capacities of their human consciousness that they already possess. As Lähteenmäki (2007, p.188) says, “[W]e can be conscious by having direct thoughts and having reflective thoughts.” So human infants are still conscious, even if on a lower level or capacity than an adult.[7]

Secondly, by virtue of a conscious thing’s being a thinking thing and requiring this second-order reflexivity as a property of that consciousness, the human person would have to be in perpetual thought about something. Lähteenmäki notes this in his paper (2007, p.188): “By maintaining that the mind always thinks in one manner or another, Descartes subscribes to the view that there are varieties or degrees of consciousness between which we constantly vacillate […] [.]” So by virtue of being a thinking thing, the person must always be thinking, and this sounds counterintuitive if we think of things such as sleep, or the perception of objects in our peripheral vision. But we could imagine a sliding scale on which to place all human thoughts: {0, 0.1, 0.2 … 0.8, 0.9, 1.0}, where 0 designates something completely out of one’s thought and 1.0 being something’s being in the fullest, most robust attention it could possibly have from someone. All Descartes’s view commits us to is that there be at least some single object which, at all times, lies somewhere from 0.1 to 1.0 on the scale. And that’s not at all difficult to imagine.

In conclusion, then, we have argued the following: First, Descartes’s phrase cogitatio extends to all mental acts and data, and therefore also to consciousness, so as to establish that consciousness is at least related to if not dependent upon thought; second, we have seen that Descartes denies consciousness to animals, regardless of their complex neuro-physiological similarities with human beings, implying that consciousness is neither reducible to nor emerges from complex physical arrangements; and finally, we say that human thought is characterised by complex, second-order reflexivity, by which we mean an awareness about one’s thinking about thinking. Therefore, my argument suggests we interpret Descartes’s doctrine of consciousness thus: Consciousness is that second-order reflexive awareness had by thinking things (humans).


[1] There is some dispute in Descartes scholarship whether or not Descartes’s famous Cogito phrase is strictly an invalid syllogism, which has as its first premise “I am thinking” leading to the conclusion “Therefore, I exist.” Scholars who criticise Descartes on this point simply find it to be invalid to make this inference from the premise. But I think stronger scholarship agrees that a more charitable reading of Descartes takes the Cogito to be the statement of an axiom from which we then start to build an infallible epistemology. We do not have time to delve into this subject here. Suffice it to say that I find Descartes’s Cogito both valid and coherent.

[2] Only one—having sensory perceptions—is possibly passive. But not probable since, as we shall see later in greater detail, Descartes distinguishes between mere physical sense-perception and actually perceiving rationally as an agent.

[3] Baker and Morris cite three scholars in support of their point in the citation: Williams 1967:347; Koyré 1954:xxxvii; Rée 1974:94ff. (Their citations.)

[4] Baker and Morris do provide some reasons against being too hasty in saying these two passages support the Expansion Thesis. But it seems to me that they ultimately conclude that the passages do support it, and for that reason, and for sake of space, I will be working as though their opinion is correct.

[5] It’s possible that this may regress infinitely. So if I am aware, consciousness suggests to me that I am aware that I am aware. If I become aware that I am aware, then on reflection I may become aware that I am aware that I am aware… and so on, ad infinitum, potentially. (One need not reflect past one or two levels of that kind of awareness to get the picture.)

[6] Thomas seems to recognise these points at the beginning of her (2006), p.341. She notes that Descartes rejects argument by analogy, and that he employs the automaton argument against physical consciousness.

[7] The idea that things can possess consciousness on different “levels” or “capacities,” to me, raises questions similar to Janice Thomas concerning the coherence and legitimacy of Descartes’s denial of animal consciousness. But we do not have space to go into that.

Baker, Gordon; Morris, Katherine J. Descartes’ Dualism (Routledge: London and New York: 2002).
Descartes, René. The Philosophical Writings of Descartes (Cambridge University Press: Cambridge: 1985), Vol. I, trans. eds. John Cottingham, Robert Stoothoff & Dugald Murdoch.
Descartes, René. The Philosophical Writings of Descartes (Cambridge University Press: Cambridge: 1984), Vol. II, trans. eds. John Cottingham, Robert Stoothoff & Dugald Murdoch.
Descartes, René. The Philosophical Writings of Descartes (Cambridge University Press: Cambridge: 1991), Vol. III, trans. eds. John Cottingham, Robert Stoothoff, Dugald Murdoch & Anthony Kenny.
Lähteenmäki, Vili. ‘Orders of Consciousness and Forms of Reflexivity in Descartes’ in Sara Heinamaa, Vili Lähteenmäki & Pauliina Remes, Consciousness: From Perception to Reflection in the History of Philosophy (Springer: 2007), pp.177-202.
Simmons, Alison. ‘Cartesian Consciousness Reconsidered.’ Philosophers’ Imprint Vol. 12 No. 2 (2012), pp.1-21.
Thomas, Janice. ‘Does Descartes Deny Consciousness to Animals?’ Ratio Vol. 19 No. 3 (2006), pp.336-363.

Historical Explanation vs. Historical Understanding

PDF: Images Sem Ppr

 * * *

            In Critical or Analytic Philosophy of History, explanation forms a key component for justifying historical propositions. That is to say: an historical proposition is more likely true than not if we can provide an explanation of why that proposition is true. By an historical proposition we roughly mean a truth-apt sentence in the past tense, being characteristically a member of a set of like propositions, denoting events or sequences in the world, which are necessarily causally related to each other in some way, either as a preceding cause or subsequent effect. Since this description is consistent with both historical occurrences of both human action and nonhuman action (human action and meteorite strikes, for example), we shall restrict ourselves for the purposes of this paper to parameters of interpersonal and intercultural human affairs. Examples of such propositions of historical human action include: “Caesar crossed the Rubicon,” which is a simple stand-alone statement of human action. In contrast, the proposition, “The Norman Conquest of Anglia in 1066 influenced rapid phonetic change in the English language,” expresses a causal relation between the past tense Norman Conquest and linguistic change in the English language at a past time. It also serves (partially) to explain linguistic change in Anglia from 1066 onwards. If we asked the question, “Why did English language undergo phonetic change from 1066 onwards?” at least a partial explanation afforded by the proposition would be: “The Norman Invasion.” Historical propositions in this context therefore concern sets of past tense truth-apt sentences about human affairs that denote events which are causally related in some way, and which are explanatorily related to their surrounding propositions in the relevant sense. An historical explanation would therefore be a set of historical propositions which provide a true description of how it is true that an historical proposition holds, and therefore how plausible any given historical theory is concerning any event human action whatever.

Yet in attempting to explain historical propositions, there is a dispute in the literature concerning which kinds of methods and what logical form an historical explanation should take. In this paper, I will attempt to sketch two broad alternative theories that have been proposed to explain historical propositions: Covering Law or Regularity Theory, and Narrative Theory or Interpretationism. Morton White will be taken as a paradigm Regularity Theorist,[1] and Georg Henrik von Wright will be taken as a paradigm Interpretationist.[2] I will work with each theory independently in a respective section. In a third and final section, we shall attempt to tie some of the related concepts together, in attempting to integrate the perceived benefits of each theory intuited by me. In so doing, I will attempt to give a broad summary and exegesis of each author. I will then aim to assess the perceived strengths and weaknesses of their view. I will suggest that the Regularity Theory provides a sound logical structure for historical arguments, but suffers from intuitive metaphysical problems about free will for people who share my metaphysical presuppositions. This weakness may be amended by an application of Interpretationist intuitions, which seem to allow for free will contingency. But as Interpretationism arguably tends to be logically loose on its ‘Practical Inferences’ syllogisms, I therefore suggest that we employ something of the insights of the structure of the arguments of the Regularity Theory in order to preserve Interpretationist intuitions but with a more formidable logical form that is characteristic of the explanatory deductive arguments of the Regularity Theory. I will then propose a very rough sketch of what such integration may look like. Overall, my implied suggestion is that some kind of integration is not only possible but preferable in explaining historical propositions.

First, we want to assess the Regularity Theory as represented by Morton White. According to White, the standard version of the Regularity Theory of historical explanation is a way of explaining any given historical explanandum event or state of affairs E at t with reference to some set of other events or states of affairs {E1…En} under subsumption of some set of nomological covering laws {L1…Ln}, which jointly explain E at t by way of a necessary and logically deductive sequence, whereby {E1…En} and {L1…Ln} jointly serve as the explanans of E at t, where E at t simply falls out under the appropriate initial conditions specified by the covering laws. On this view, the explanation of any human activity in the historical past, such as wars, revolutions, migrations, or economic (in)stability is, according to its logical form, like unto how one would come to explain how a particular fire started in some particular wastebasket at some time t.[3] Here, we are able to explain the existence of some given fire from the necessary and sufficient conditions surrounding that event, subsumed under a relevant covering law concerning the presence of such conditions and the initiator of events, such as a spark’s falling into the wastebasket, where there is dry paper, and a sufficient amount of oxygen (say, one litre) to bring it about that a fire takes place. Given certain physio-chemical facts we know about fire, we theorise some nomological covering law that whenever a spark falls into a waste basket under such appropriate conditions, a fire will ensue. Thus, if a fire ensues at time t, then the presence of the necessary and sufficient conditions in conjunction with the nomological covering law at t jointly explain the presence of that fire at t. So the fire is deduced from necessity with reference to formal deduction and covering laws. In a simplified form:

1. E occurs at t.
2. E occurs at t iff {E1…En} and {L1…Ln}.
3. Therefore, {E1…En} and {L1…Ln}.

The proposition expressed on the right-hand of the biconditional on premise (2) serves as the explanans of the proposition expressed in the left-hand side statement in the same premise (2), which is also the explanandum in premise (2). The conclusion then serves to make our explanans explicit: explanandum E at t is explained with reference to explanans {E1…En} and {L1…Ln}. So stated, the standard version of the Regularity Theory provides us with an explanatory logical deductive argument for the occurrence of E at t. This is a logically air-tight form of historical explanation grounded in valid logic and verifiable laws and conditions. As White says, “The fundamental drive of [the Regularity Theory], at least in our time, is […] in the direction of analysing explanation so as to avoid reference to mysterious powers and unanalysed connections.”[4] On the Regularity Theory, an explanation is legitimate qua explanation if and only if it provides the means deductively to derive a description of the explanandum from a statement of a law and a description of some initial conditions which serve as explanans.

This form of stating the Regularity Theory has the benefit of a valid, logical structure in a deductive syllogistic form. This makes it easy to determine the logical coherence of an historical theory, so as to be able objectively to assess the validity or invalidity of any historical explanation. This point may seem rather basic for a philosopher. But this form of rigorous logical structure is surprisingly atypical in professional history.[5] It should not be.Yet histories as written by professional historians ought to be characterised by this form of logically air-tightness in their explanatory formulations so as to provide a valid method of assessing the strengths and weaknesses of their theories. The logical-deductive method of the Regularity Theory allows for explanatory coherence, and objectivity, and thus provides just the right kind of structure we ought to see in an academic context. Therefore, my intuition is that the Regularity Theory shares the benefit of sharing particular logical-formal strengths that are not characteristic of the form of Interpretationism that will be discussed below.

Yet the Regularity Theory plausibly suffers from some potential weaknesses. I shall list three proposed weaknesses and one actual weakness.[6] First, it is argued that, in most instances, the complexity of covering law conditions in explaining some historical occurrence E at t is such that the only real instance of the covering law we use in trying to explain just is E at t at nothing more. The second objection is like the first: Even if the regularity or general law may apply to more than one historical occurrence, there would be a paucity of examples anyway in the actual world. Therefore, covering laws are otiose explanations. They are not sufficiently general, and surely not useful for explaining historically.

Such objections are intuitive. But as Morton White argues, even if it were it the case that a covering law applied solely to one individual occurrence in the world, it may still be true, and may also imply statements about other individuals of the same kind, were they to exist. Matchsticks, for instance, when struck, produce a flame under the relevant covering law alongside the right set of general conditions concerning sulphur and oxygen, and striking a match. The law would still hold even if there were only one matchstick in the whole world. If another matchstick were mad, it would then under the same laws and conditions, such that we could predict that it would also produce a flame under them. The restriction of actual instances of a general law in history is simply not relevant to the truth of the general law itself, especially if we have evidence for the proposed covering law in the world. So much, then, for such objections. As White says, citing J. S. Mill, “Sheer paucity of supporting instances [of a covering law] is not enough to condemn an induction, nor is sheer abundance of them enough to certify it as valid.”[7]

Another more persuasive objection comes from the fact that not all explanations need to make reference to covering laws. “The Norman Conquest of Anglia in 1066 influenced rapid phonetic change in the English language.” No covering law there. Yet the implied statement “Rapid phonetic change occurred in the English language from 1066” is explained by the statement “The Norman Conquest of Anglia in 1066” connected by the causal word “influenced,” without reference to a covering law. There therefore exist singular, explanatory statements which do not make reference to covering laws. Nor do such singular, explanatory statements need to be put into a deductive sequence. Yet they are explanatory. Here I agree with White that the Regularity Theorist may agree that deductive sequences with reference to covering laws may not be necessary to explain some historical occurrence E at t. Yet surely putting explanations into a logical, deductive form is better than these standalone individual statements in a technical context. Here I actually differ with White that we should not at least try to translate such singular, explanatory statements into a deductive form that is valid so as to draw out the covering law and relevant conditions implicit in the statement. For, on reflection, the statement, “The Norman Conquest of Anglia in 1066 influenced rapid phonetic change in the English language,” in assuming itself to be explanatory, actually seems to imply the following universal conditional statement: For any entity x, if x is a people and language group and x is invaded sufficiently rapidly and all-encompassingly by a non-identical people and language group y, then the language of x changes rapidly. From this statement, we are able to formulate premises in an argument of the explanatory deductive form that is in keeping with the original Regularity Theory. To be sure, it is not at all obvious we need to do this. But as long as we ca show (as I think we can) that singular, explanatory statements do not necessarily falsify the explanatory, deductive method of argument characterise of the Regularity Theory, this argument falls flat as an objection.

What objections, then, may be brought against the Regularity Theory? Just this: On particular metaphysical presuppositions about Libertarian Free Will, it seems hard for the Regularity Theorist to justify the employment of a general law and a set of necessary and sufficient conditions, according to which certain outputs just follow by necessity given certain law/condition inputs, such as E at t simply falling out under {E1…En} and {L1…Ln}. Especially in the context of an historical explanation of any individual or communal human action, if we grant the metaphysical presupposition that all human agents have Libertarian Free Will (by which I roughly mean the rational ability of a conscious agent to do otherwise in any given set of circumstance of choices, A or B), then it seems just impossible to guarantee that that any historical set of circumstances E at t will just fall out in the case some set of necessary and sufficient conditions and covering laws {E1…En} and {L1…Ln}. What would it even mean for there for be a ‘necessary’ condition for some free human action E at t? Would {E1…En} and {L1…Ln} necessitate E at t? It seems to me that they would. But ths is simply unacceptable for proponents of Libertarian Free Will. In short, despite its attractive logical structure, White’s Regularity Theory seems, at the base of it, thoroughly deterministic, and this is simply unacceptable on my metaphysical presuppositions. One may admittedly not share in my presuppositions—perhaps one is an Incompatibilist Determinist, or some form of Compatibilist with Determinism. In that case, this objections falls flat. But for people such a me who do feel inclined to be Indeterminists and yet believe that we do have Free Will, that counts against the Regularity Theory.[8] We therefore want some theory which retains the logical structure of the Regularity Theory, but which can account for genuine human libertarian free agency, on my view.

Georg Henrik von Wright’s Interpretationism may help us here.[9] In his account of historical explanation, von Wright proposes we distinguish between two distinct yet interrelated concepts: ‘explanation’ (Erklären) and ‘understanding’ (Verstehen). Explanation, say von Wright, is more relevant to scientific analyses concerning necessary nomological connections between events in nature, and as such takes place under necessary nomological deductions.[10], Understanding, on the other hand, is a way of making sense of human actions, inter alia, in terms of reasons, intentions, and motivations, in much the same way as takes place in the social and cultural sciences as opposed to the natural sciences. Human action in history cannot be understood from a strictly explanation vantage point comparable to the natural sciences, but only with reference to the understanding as charitably taken as a interpersonal and intercultural study of complex, often alien, affairs.

At the heart of the Interpretationist model is the concept of a Practical Inference.[11] Roughly stated, a Practical Inference just seems to be a way of understanding some human activities with reference to the reasons, intentions and motivations of conscious creatures to act in the way that they do. Briefly summarised, a Practical Inference admits of the following form:

1a. Entity a intends to bring it about that E at t.
2a. Entity a considers a cannot bring it about that E at t unless a does A at time t*.
3a. Therefore, a sets a to do A at t*.

Here, (3a) is said to be a practical inference from premises (1a) and (2a) in that it provides an understanding of how it was true that a sets a to do A at some time t* (which may or may not be prior to E at t) leading to the outcome that E at t. We therefore come to understand why E at t with reference to the reasons, intentions and motivations of a in doing A at t*. The inference is therefore practically valid for all intents in purposes of understanding E at t.[12]

In order to bring out this intuition more clearly, consider a concrete case in the world. Historians generally agree that: “the cause of the outbreak of the First World War was the assassination of the Austrian archduke at Sarajevo in July 1914.”[13] This assassination purportedly led to a declaration of war by Austria, and hence to all the events we place under the signification of World War One. Here, the explanandum is the outbreak of the war, and the proposed explanans is the declaration of war motivated by the assassination of the archduke at Sarajevo. But there is more to it. Suppose Austria intended to restrict Russian control in the Balkans, but needed an excuse to move in such a way so as to bring about this restriction. Suddenly, understanding the historical human relations becomes more complex. Now we must not only specify the explanandum and explanans, but the explananda and explanantia also, which often serve as contextual features and contributory causes in an explanans for the explanandum. All are logically independent events in the world which nevertheless bear some kind of non-nomological, non-Humean causal relation to each other. They do not stand in a necessary casual relation. They are not directly causally related. Yet they are causally related in a causal chain, even if the causal relevance of any number of causes is contributory only and not unilaterally so. How, then, do we make sense of this causal connection? How do we understand it? The Interpretationist says: by way of Practical Inference. What must be done is clearly to delineate the likely motivations of agents as discerned in the relevant source documents leading to the declaration of war by Austria at the relevant past time. We can summarise this course reasoning as follows:

1b. Austria intends to bring it about that Russian control in the Balkans is restricted.
2b. Austria considers that it cannot bring it about that Russian control in the Balkans is restricted unless Austria finds some reason to declare war at time t*.
3b. Therefore, Austria sets itself to do find some reason to declare war at t*.

From this we may infer the understanding that the reason war came about was not only the assassination, but Austria’s prior intentions to restrict Russian control in the Balkans. The assassination may be said to be the cause only in relation to the prior contributory cause of the intention of the Austrian government, and so come (at least partially) to understand why World War One began.

Interpretationism seems correct on a number of levels. First, von Wright’s distinction between explanation and understanding seems intuitive. There is indeed an arguable difference between explaining something in a scientific manner, and understanding the activities of human action in the social sciences. Scientific explanations under necessary nomological deductions are qualitatively different from how we ordinarily understand human action in that nomological statements predict regularities in nature which can be objectively measured by the means employed in the physical sciences. But human activity is obviously measurable in this way. Human actions are often complex occurrences that are beset with various interrelated motivations, reasons and intentions, which are contributory causes that do exhibit causal relations, but not so in the same kind of way as events in nature do. I therefore find this distinction an attractive feature of von Wright’s account.

Second, given the complex interrelations that are postulated on Interpretationism, in terms of complex motivations, intentions and reason with various streams of contributory causes in an event, it is very easy to discern the spirit of human contingency which I have already indicated I would like to see preserved in our theory of historical explanation, given my metaphysical presuppositions. Interpretationism makes better sense of Libertarian Free Will agents relating in complex ways for various reasons, and this appears to me as a attractive feature.

What is not an attractive feature, however, is the apparent logical looseness of a Practical Inference on Interpretationism. By ‘apparent logical looseness,’ I simply mean the classical invalidity of the conclusions found in the practical inferences in the two examples above. It is not easy to see how it is that the conclusion is supposed to follow on the premises, nor is the spooky concept of ‘practical validity’ obviously comprehensible. The idea seems to be that, if we just dwell on the reasons, intentions and motivations of human agents like Austria in their contexts, we can come to intuit a kind of ‘sense-makingness’ in the actions of the country. We can just ‘see’ that, if Austria wanted to restrict Russian control in the Balkans, and if the assassination of the Austrian archduke at Sarajevo in July 1914 provided the perfect excuse to declare such as would lead to that restriction, then it would make sense if Austria declared war for just this reason.

But I agree with Bernard Williams that there is a difference between making sense and being reasonable.[14] Some activity A at t*  may make sense in that we can tell a coherent story demonstrating truthfully that A at t* stands in a causal relation to activity E at t. But it does not follow from this that either A at t* or E at t are inherently reasonable, sensible or rational. This seems to be a problem for von Wright’s account.
Might the general idea of Interpretationism be saved by connecting it with the explanatory deductive method of the Regularity Theory? It seems to me that it can. It seems to me that an integration of the strengths of these two views may just avoid each other’s weaknesses. I have already argued why I think the classical, logical and deductive form of the Regularity Theory is a good one. But I mentioned it does not seem to be consistent with human contingency. Interpretationism, on the other hand, makes good sense of human contingency, but employs a kind of loose logic. We therefore ought to try to craft a theory of historical explanation and understanding that takes cognisance of both human contingency, and explains these contingent historical propositions of human actions in a logically robust, explanatory deductive sense.
Here is what I suggest. Let us suppose we are able to craft an explanatory deductive argument whose individual parts make sense, and whose conclusion is also inherently reasonable, given its being logically deduced from true premises and valid logic. What might such an argument look like? I suggest the following:

1c. E occurs at t.
2c. E at t makes sense iff some other set of events {E1…En} takes place, and there is a designated set of intentions/desires/motivations {S1…Sn} on part of human agents at the times of those other events which jointly serve as contributory causes for E at t and provide understanding for E at t.
3c. Some explanation, P, explains E at t iff P makes sense of why E occurs at t.
4c. Therefore, P explains E at t iff some other set of events {E1…En} takes place, and there is a designated set of intentions/desires/motivations {S1…Sn} on part of human agents at the times of those other events which jointly serve as contributory causes for E at t and provide understanding for E at t.

What follows from this? It follows that we have specified the necessary and sufficient conditions for what counts as an historical explanation of E at t, namely, some explanation P that is able to specify the surrounding events, intentions, desires and motivations of the agents involved in them, which are shown to serve as contributory causes of E at t.  The main debate should then be centred on the identification of just what those other events {E1…En} and intentions, desires and motivations {S1…Sn} precisely are, as represented in source evidence. That, however, is purely an empirical question best left to be solved by the historian. What I have proposed here forms a rough condition for what counts for explaining any historical event E at a given time t in deductive form. Precisely what the method of explanation ‘P’ is may be left unspecified at this point. Perhaps it is a probability analysis. Perhaps it is a self-contained narrative description. All that any given explanation P needs to do is make sense of E at t in the appropriate way that fulfils the relevant conditions for what it takes to make sense in premise (2c). The proposed method therefore preserves both positive elements we wanted it to: it has a valid logical structure and does not obviously conflict with free libertarian human contingency, and the complex non-necessary contributory causal relations implied by it. That forms just the kind of explanation theory that ought to be desired. It integrates in an intuitive way the strengths of both Regularity Theory and Interpretationism, and avoids their respective weaknesses.[15]


[1] Morton White. Foundations of Historical Knowledge (Harper & Row Publishers: New York and London: 1965).

[2] Georg Henrik von Wright. Explanation and Understanding (Cornell University Press: Ithaca, New York: 1971).

[3] White (1965), p.15.

[4] Ibid. p.53.

[5] David Hackett Fischer. Historians’ Fallacies: Toward a Logic of Historical Thought (Routledge & Kegan Paul: London: 1971).

[6] The first three are assessed by White (1965), pp.23-25, 29, 47; the fourth hails somewhat from my interpretation of some issues raised in Richard Eldridge (forthcoming).

[7] White (1965), p.37.

[8] The reader should be reminded that only form of the Regularity Theory or Covering Law model I am criticising is Morton White’s (1965) model. Indeed, as Georg Henrik von Wright briefly mentions in passing in his (1971) work, p.13, some forms of Covering Law models do not proscribe necessary, deductive, nomological explanations but advocate rather for inductive probabilistic explanations, where some set of conditions C make probable that E at t, and the best explanation for E at t is just that set of conditions C or C* which makes most probable that E at t. This does indeed allow for libertarian contingency, and thus avoids the objection. There simply is not enough space to go into the issues, especially given my restricted purposes here.

[9] Wright (1971).

[10] By ‘necessary’ here, I simply mean to refer to what is supposed to follow in nature given the set of laws we find ourselves having in our world. I do not mean to imply that the laws themselves are necessary. Perhaps another universe with different physics could have existed. But I hope you can see that whole issue is not exactly relevant in this context, and is simply assumed.

[11] Ibid. p.96.

[12] Following von Wright, we can replace the phrase “intends” with phrases like “aims at,” “pursues,” and “wants;” the phrase “considers” with phrases like “thinks,” “believes,” and “knows;’ and the phrase “sets [himself] to do” with phrases like “embarks on doing,” “proceeds on doing,” or simply “does,” etc. Each phrase denotes some complex activity of the human entity, be it individual or communal.

[13] Wright (1971), p.139.

[14] Bernard Williams. Truth and Truthfulness: An Essay in Genealogy (Princeton University Press; Princeton and Oxford: 2002), pp.233-235.

[15] I would like to thank Prof. Richard Eldridge of Swarthmore College, PA, for permission to use his paper, “Introduction: Historical Understanding and Human Action” (forthcoming), for thought, in preparation for this seminar paper.

Fischer, David Hackett. Historians’ Fallacies: Toward a Logic of Historical Thought (Routledge & Kegan Paul: London: 1971).
White, Morton. Foundations of Historical Knowledge (Harper & Row Publishers: New York and London: 1965).
Williams, Bernard. Truth and Truthfulness: An Essay in Genealogy (Princeton University Press; Princeton and Oxford: 2002)
von Wright, Georg Henrik. Explanation and Understanding (Cornell University Press: Ithaca, New York: 1971).

Classical Pyrrhonian Scepticism as Epistemic Caution

Download WORD2013 document: PHIL2649 Maj. Ess.

Philosophical problems in classical Epistemology arose in the context of ancient metaphysical disputes concerning the nature of what is, and how the human mind is able to gain epistemic access to what is.[1] Pyrrhonean Scepticism (henceforth Pyrrhoneanism) was one form of classical scepticism which advocated the suspension of dogmatic judgement on matters of fact in metaphysics, especially with respect to the true nature of substances, as opposed to one’s personal understanding or immediate perception of those substances. This essay seeks to offer an interpretation of Pyrrhoneanism, defend its coherence, and to argue that this form of scepticism could be applied to modern epistemological debate as a form of epistemic virtue.

Most of our textual information about Pyrrho of Elis (c.360-270BC) and Pyrrhoneanism is found in the writings of Diogenes Laertius about Pyrrho and the reflections of Sextus Empiricus on the Pyrrhonean tradition in general.[2] According to Sextus Empiricus, three principle kinds of philosophy: Dogmatic, Academic, and Sceptical (Pyrrhoneanism).[3] Dogmatic philosophers tend to claim that they have discovered the real truth about logical, physical, metaphysical and normative matters and therefore into the true nature of things that exist. Dogmatists therefore feel justified in affirming knowledge about many things about the world. Academic Sceptics, on the other hand, hold that nothing can be known.[4]

Pyrrhoneanism may be distinguished from Dogmatic and Academic philosophy, in that it purports to make no claims about the true nature of our world without denying the possibility. Pyrrhoneanism is characterised by skepsis (inquiry) on part of the skeptikos (inquirer).[5] In general, the Pyrrhonean tradition cast doubts on our cognitive ability to grasp the nature of the world as it really is, as opposed to my mere perception of the world. There are important similarities between Academic philosophy and Pyrrhoneanism.[6]

Pyrrhoneanism, Richard Bett tentatively suggests, may have shared influences from several philosophical predecessors.[7] These include: the Indeterminacy Thesis,[8] Aristotle’s Law of Noncontradiction,[9] the self-critical Instability Thesis of Plato’s Theaetetus,[10] the Eleatic doctrine of the deception of appearances,[11] Xenophanes’ assent to mere belief (dokos) as opposed to rationally justifiable belief about the external world according to the relativity of individual sense-perception,[12] Protagoras’ scepticism about theology because of the disagreement of the polytheistic faiths,[13] Democritus’ atomic theory of perceptual ambiguity and doctrine of the end-goal of philosophy as possessing a ‘good spirit’ (euthumia) or ‘well-being’ (euestō),[14] the Cynics’ doctrines about perceptual experience as possible illusion reaming and general attitude of tentativeness towards theoretical enquiry in general,[15] the Megrarian opposition to predication,[16] and more interestingly the far-eastern Buddhist doctrines of impermanence the insubstantiality of ordinary objects.[17] Bett seems tentative about positing each of these in the development of Pyrrhoneanism directly, bbut aintains it is nevertheless possible that they had some relevance to the Pyrrhonean movement. For in common to each these beliefs is the garnering of doubt concerning the legitimacy and accuracy of our perceptual and conceptual powers about the real world past mere belief. And that is Pyrrhoneanism.

The limitations and ambiguities of sense perception and consequent conceptual formulations led Pyrrhoneanism to postulate that philosophers ought to suspend judgement or belief (epochē) about all and every ambiguous or debatable subject of interest as the telos or goal of philosophical enquiry.[18] For the Pyrrhonean both Dogmatist and Academic are being too bold in their assertions about the true nature of things. This suspension of judgement or belief, says the Pyrrhonean, will lead to mental tranquillity (ataraxia), and this tranquillity or mental calm should lead to an ‘ordinary life’ (bios) that is ‘without opinions’ (adoxastos).[19] Alan Bailey asks us to consider this claim: ‘There is an even number of grains of sand in the Libyan Desert.’ When one encounters this claim, suspension of belief arises naturally. Even though one knows that there is a potential answer one could give, it is just be extraordinarily difficult and impractical to determine the outcome. Therefore the proposition is neither believed nor disbelieved; judgement concerning is merely withheld. The goal of Pyrrhoneanism is ‘suspension of judgement, upon which freedom from anxiety follows like a shadow.’[20]

In order to show some more precise conditions for suspension of judgement, the  primary sources indicate that Pyrrhoneans employed certain ‘modes’ or arguments indicating ambiguity that implies that judgement about the true nature of things needs to be suspended by the enquirer. Fifteen different modes are discussed by Diogenes Laertius in his writing about Pyrrho—fifteen different basic arguments or observations in favour of Pyrrhoneanism by way of which Laertius calls “the Pyrrhonean strategy”.[21]

Firstly, Pyrrhonean Sceptics noted that there is ambiguity between different species of animal about pain-reception, and therefore about the objectivity of the reception of pain due to common causes; secondly, the idiosyncratic nature of man and each man’s individual tastes and affections spreads doubt on human qualities of taste and perception; thirdly, differences in sensory experience such as in taste distinctions (the same apple may be sweet to one and bitter to another); fourthly, the fluctuation of human dispositions of love/hate, sadness/joy, etc., should suspend one’s judgement about the nature and cogency of human personalities; fifth, the diversity of human cultures, moral beliefs and religious practices spreads doubt on the true nature of theological matters and ethical concerns; sixth, the emergence of properties that only seem to emerge and change in relation to other properties (e.g. skin complexion in the sunlight); seventh, perceptual ambiguity according to distance (e.g. a large building seems small far away); eighth, quantities and qualities can seem contradictory (e.g the wine that fortifies oneself in small amounts but in large quantities intoxicates and makes weak); ninth, the occurrence of odd happenings or phenomena in the world (e.g. miracles, natural disasters) casts doubt on the uniformity of nature; and tenth, there are comparisons and relations, i.e., something 100kg of weight is heavy, but it is light compared to a 10,000kg thing. Then, the ‘Agrippa School’ of Pyrrhoneanism also added five more general arguments which should cause one to suspend judgement in ordinary philosophical conversation.[22] Firstly, the argument from disagreement shows that many brilliant minds have disagreed about many matters and this ought to deter one from making a judgement concerning them; secondly, the argument from infinite regress, whereby when one thing must always be based on some other thing, shows that we cannot really ground anything, since explanation would continue ad infinitum and nothing would be explained; thirdly, the argument from relativity suggests that things are never understood in relation to themselves but only in reference to other things, the problem of which is a variation on the infinite regress argument; fourthly, the argument from hypothesis posits that for ever hypothesis there may exist an equally plausible antithesis; and fifthly, the argument from insularity or circular argumentation shows that whenever in a case where some proposition P is based on Q, but where Q needs P, both P and Q are unjustified, and we must suspend judgement about them both. For each mode of argument leading to the suspension of judgement about whatever particular philosophical puzzle, it is easy to see from Laertius’ characterisation of the modes how modern scientific tools might help with solving them easily. Others are not so easy to solve on raw empirical data, however.[23] For the purposes of this essay it is sufficient to note that the Pyrrhonean took there to be the relevant phenomena in question, but suspended his judgement about the true nature of the phenomena in question due to the confusion and ambiguity that arises for him.[24]

One primary challenge to Pyrrhoneanism is the claim that it is a self-defeating and incoherent view. If one is supposed to suspend judgement about all things, should not one then suspend judgement about suspending judgement? But this kind of argument is just a mistake. Whereas the Pyrrhonean Sceptic suspends judgement about the essential nature of ideas such as ethics or substances such as gods, that does not mean they cannot act on what appears immediately evident to their senses and inclinations personally. The Pyrrhonean does not assent to nothing at all, but he only suspends judgement about non-evident things. Phyrroneanism is a way of responding to controversial and ambiguous propositional claims about the nature of substances.[25]

Machuca’s recent research on ancient Scepticism notes how ancient issues remain relevant for contemporary epistemologists.[26] So far as Pyrrhoneanism goes, it seems we could cast it as a form of intellectual humility in epistemic enquiry—something often agreed upon as being a virtue by contemporary epistemologists, whereby one’s epistemic claims are situated in the consciousness of one’s own fallibility and limited evidence. Pyrrhoneans were tentative about postulating anything about the true nature of objects and therefore they suspend judgement concerning the true nature of substances in the world. Contemporary epistemologists Robert C. Roberts and W. Jay Wood note that caution in judgement is a virtue in the life of the mind. Caution is ‘a disposition of proper fear […] a disposition to fear what is worthy to be feared, to fear it enough (but not too much) and in the right way, and then to be able to avoid what is feared’, namely, error or unjustified statements. They continue, ‘Intellectual caution is wise fearing in cognitive matters. Wise cognitive fears are based on the love of knowledge, and the love of knowledge is based on a concern for the epistemic goods that situate them among the other goods of human life—a concern for information, understanding, and acquaintance that is significant, relevant, and worthy in the context of a human life.’[27] The Pyrrhonean’s tentative attitude towards making claims about the true nature of substances plausibly exhibits this kind of epistemic caution, which is birthed from a form of intellectual humility on part of the intellectual philosopher.

Pyrrhoneanism appears to be coherent doctrine of scepticism which models just the kind of epistemic humility required in critical investigations into metaphysics today. It was supported by argument and reason in the network of discussions about metaphysics and knowledge in the classical Greek world, and it provides an interesting example of classical Greek argumentative style in general. A softer form of the Pyrrhonean method is probably a good one for philosophers of knowledge to look into.


[1] Unlike in modern philosophy, classical philosophy did not necessarily distinguish metaphysical and epistemological issues. See for example Nicholas P. White’s comments in reference to Plato: “For some time philosophers have thought of epistemology and metaphysics as different branches of philosophy, investigating, respectively, what can be known and the basic properties and nature of what there is. It is hard, though, to see any genuine boundary here. The issues irresistibly overlap. Certainly in Plato there is no such divide.” Nicholas P. White, ‘Plato’s Metaphysical epistemology’ in Richard Kraut (ed), The Cambridge Companion to Plato (Cambridge University Press, 1992), p.277.

[2] Diogenes Laertius, ‘Skepticism’ (9.74-108) tr. Brad Inwood and Lloyd in C. D. C. Reeve & Patrick Lee Miller (Trans. Eds.), Introductory Readings in Ancient Greek and Roman Philosophy (2006), pp.375-378; Sextus Empiricus, ‘Outlines of Pyrrhonism’ (Bks. 1, 2, 3; excerpts) tr. Sanford P. Etheridge, Brad Inwood, and Lloyd P. Gerson in C. D. C. Reeve & Patrick Lee Miller (Trans. Eds.), Introductory Readings in Ancient Greek and Roman Philosophy (2006), pp.433-442.

[3] Ibid. p.434.

[4] David E. Machuca, ‘Ancient Skepticism: The Skeptical Academy’ in Philosophy Compass 6/4 (2011), pp.259-263.

[5] David E. Machuca, ‘Ancient Skepticism: Pyrrhonism’ in Philosophy Compass 6/4 (2011), p.251.

[6] David E. Machuca, ‘Ancient Skepticism: Overview’ in Philosophy Compass 6/4 (2011), pp.234-245.

[7] Richard Bett, Pyrrho, his Antecedents, and his Legacy (2000), Ch. 3: “Looking Backwards” pp.112-188.

[8] Ibid. pp.114-123.

[9] Ibid. pp.123-131.

[10] Ibid. pp.132-140.

[11] Ibid. pp.140-143.

[12] Ibid. pp.143-149.

[13] Ibid. pp.149-152.

[14] Ibid. pp.152-160.

[15] Ibid. pp.160-165.

[16] Ibid. pp.165-169.

[17] Ibid. pp.169-178.

[18] David E. Machuca, ‘Ancient Skepticism: Pyrrhonism’ in Philosophy Compass 6/4 (2011), pp. 247, 248, 252; Cf. Alan Bailey, Sextus Empiricus and Pyrrhonian Scepticism (2002), p.119. This theme arises in various philosophers. Cf. Sextus Empiricus, ‘Outlines of Pyrrhonism’ (Bks. 1, 2, 3; excerpts) tr. Sanford P. Etheridge, Brad Inwood, and Lloyd P. Gerson in C. D. C. Reeve & Patrick Lee Miller (Trans. Eds.), Introductory Readings in Ancient Greek and Roman Philosophy (2006), p.437.

[19] Alan Bailey, Sextus Empiricus and Pyrrhonian Scepticism(2002), p.122.

[20] Diogenes Laertius, ‘Skepticism’ (9.74-108) tr. Brad Inwood and Lloyd in C. D. C. Reeve & Patrick Lee Miller (Trans. Eds.), Introductory Readings in Ancient Greek and Roman Philosophy (2006), p.378.

[21] Diogenes Laertius, ‘Skepticism’ (9.74-108) tr. Brad Inwood and Lloyd in C. D. C. Reeve & Patrick Lee Miller (Trans. Eds.), Introductory Readings in Ancient Greek and Roman Philosophy (2006), pp.375-378.

[22] Ibid. pp.377-378.

[23] In any case, that would be unfair to ancient Pyrrhoneanism since ancient Greeks obviously did not have the tools we have today to investigate the natural world.

[24] David E. Machuca, ‘Ancient Skepticism: The Skeptical Academy’ in Philosophy Compass 6/4 (2011), pp.259-263.

[25] Gail Fine, ‘Concepts and inquiry: Sextus and the Epicureans’ in Benjamin Morison and Katerina Ierodiakonou. Episteme, etc. (2011), p.90; Alan Bailey, Sextus Empiricus and Pyrrhonian Scepticism(2002), p.146; Julia Annas, ‘Ancient skepticism and ancient religion’ in Benjamin Morison and Katerina Ierodiakonou (Eds.) Episteme, etc. (2011), p.75.

[27] Robert C. Roberts and W. Jay Wood, Intellectual Virtues: An Essay in Regulative Epistemology (2007) pp.234-235.


ANNAS, Julia. ‘Ancient scepticism and ancient religion’ in Benjamin Morison and Katerina Ierodiakonou. Episteme, etc.: Essays in Honour of Jonathan Barnes (Oxford University Press, 2011), pp.74-89.

BAILEY, Alan. Sextus Empiricus and Pyrrhonean Scepticism (Clarendon Press, 2002).

BETT, Richard. Pyrrho, his Antecedents and his Legacy (Oxford University Press, 2000).

FINE, Gail. ‘Concepts and inquiry: Sextus and the Epicureans’ in Benjamin Morison and Katerina Ierodiakonou. Episteme, etc.: Essays in Honour of Jonathan Barnes (Oxford University Press, 2011), pp.90-114.

FLORIDI, Luciano. Sextus Empiricus: The Transmission and Recovery of Pyrrhonism (Oxford University Press, 2002).

KRAUT, Richard. (Ed.) The Cambridge Companion to Plato (Cambridge University Press, 1992).

MACHUCA, Diego E. ‘Ancient Skepticism: Overview’ in Philosophy Compass 6/4 (2011), pp.234-245.
__________. ‘Ancient Skepticism: Pyrrhonism’ in Philosophy Compass 6/4 (2011), pp.246-258.
__________. ‘Ancient Skepticism: The Skeptical Academy’ in Philosophy Compass 6/4 (2011), pp.259-266.

MORISON, Benjamin; IERODIAKONOU, Katerina. (Eds.) Episteme, etc.: Essays in Honour of Jonathan Barnes (Oxford University Press, 2011).

REEVE, C. D. C.; MILLER, Patrick Lee. (Trans. Eds.) Introductory Readings in Ancient Greek and Roman Philosophy (Hackett Publishing Company, Inc., 2006).

ROBERTS, Robert C.; WOOD, W. Jay. Intellectual Virtues: An Essay in Regulative Epistemology (Oxford University Press, 2007).

WHITE, Nicholas P. ‘Plato’s Metaphysical epistemology’ in Richard Kraut (ed), The Cambridge Companion to Plato (Cambridge University Press, 1992), pp.277-310.

Defence of Semantic Internalism: Essay against Putnam’s Twin Earth

Download PDF: PHIL2621 Brendan BURNETT (Maj. Ess.)


Lewis’ Modal Realism and Putnam’s Twin Earth: Two Essays

Download WORD2013 document: Brendan Burnett [PHIL2600 Takehome with comments]


Question 4: What were David Lewis’ reasons for positing a multiplicity of concrete possible worlds, and do you agree with them?

David Nolan defines a ‘possible world’ as a maximal description of what exists; a sum total of all reality. In his famous (1986) work, David Lewis posited a theory of modality which made use of such a concept of possible worlds. Controversially, however, Lewis also claimed that “for every possible world there is, there is such a world that exists.” In other words, for each possible world we can think of from the infinite set of all possible worlds, there really is such a world that actually exists, and it is just as tangible and concrete as our world. There is a world with cities made only of doughnuts. There is a world where everything is the same except for a subtle difference in the shade of a single colour of a single section of a particular carpet in a particular room somewhere. However, for any one particular world from the infinite set of all possible worlds, that world is both spatiotemporally and causally unrelated to any of the other worlds. This essay provides two reasons to suggest that David Lewis’ Modal Realism suffers from epistemological and metaphysical problems, and that we therefore ought to prefer another theory. The first argument will bear on Lewis’ epistemic justification for his theory as presented in his original work.  The second argument will retain Lewis’ possible worlds intuitions but postulate that a divine conceptualist view that serves Lewis’ purposes better but without ontological inflation—if one is willing to pay the ‘cost’ of theism[E1] .

According to Nolan (2005, Ch. 3), modality is a real phenomenon in the world. We are interested in not only what does happen, but also in what might have happened, or what must happen. Metaphysically, we intuit the idea that we want to be able to explain the modalities of possibility and necessity in such a way as to retain most of our common-sense intuitions about what is real, in a clear and logical manner, and without maximum costs to our theory such as ontological inflation (the postulation of too many entities). David Lewis takes just this line.[E2]  In his (1986, p.3), Lewis talks of concrete modal realism as “serviceable” to our theories. On Lewis’ view, these worlds give us the metaphysics we want, how we want it. For him, the cost of ontological inflation (infinite concrete entities) simply does not outweigh the benefits[E3] . Lewis calls this a “reason” to believe in concrete possible worlds, even though it is not a conclusive reason, he says.

As for the metaphysical theories and notions Lewis believes possible worlds explain most clearly and cleanly, Brian Weatherson (2006, Sec. 6),  presents the proposed benefits as follows. First, concrete possible worlds provide us a clear-cut theory of modality. Possibilities such contingent existence modal comparatives and supervenience theses can be described as quantification across possibilities. So saying that I possibly exist for example would be like saying that I (and my relevant transworld counterparts) actually exist in a certain subset of concrete possible worlds much in the way of indexical phrases. I exist “here” (the actual world), and I exist “there” (some other possible world W), multiplied n times. This possibility also captures the idea of my existing “not-there” (not at some other possible world ). Again, modal comparatives are relations across worlds. If I could have possibly been taller that simply means at possible world W I am a certain height and in another possible world my counterpart is a certain height. Secondly, concrete possible worlds provide a clear definition of closeness. Counterfactuals (what would have been true were a certain set of circumstances the case) on modal realism are comparisons [E4] between worlds. The reasons we can make counterfactual claims in the actual world is because there is a possible world W (take it to be the actual world—our world) that is different from another possible world . Modal realism allows us to compare conditions in both worlds W and and see that in a world sufficiently local or close to us (that is, not too wildly different in any major way) the conditions were a certain way such that if the conditions of the actual world were like the conditions of the second possible world then the actual world would have been like the second world. That’s easy to think about. Thirdly, Lewis believed concrete possible worlds informs content. Mental and verbal content could now be analysed in terms of sets of possibilities. And lastly, Lewis thinks possible worlds retains clear explanation about our intuitions about properties  as quantification of sets of possibilia. If I say ‘There are no blue swans in the whole world’ I am implicitly quantifying over my world alone. It says absolutely nothing about whether or not blue swans exist in some other possible world. Maybe they exist abundantly in another possible world. And, in fact, they do. Just not here, in the actual world. That is a face-value description of properties.

The only severe problem Lewis seems to have found with his own theory is that it massively inflates our ontology. Many—indeed an infinite number of—things exist. But, he says, the vast benefits of the consequent systematic metaphysical coherence and explanation far outweighs this cost of ontological inflation. However, that implies that if there were another conceivable way of thinking about possible worlds which did not require this kind of ontological inflation, [E5] then that theory of possible worlds would be better than Lewis’ theory on Lewis own grounds. I will offer that alternative possibility after a brief note on some epistemological issues I think David Lewis fails to address at least in his (1986) work.

Epistemically, a person knows that P iff that person believes that P, P is true, and P is justified in the right way.[1][E6]  Consider Lewis’ case. According to Lewis, concrete possible worlds are things which are entirely useful in that they explain things well with a whole lot of coherence. But how can Lewis justify the inference from the usefulness of a particular postulation to its epistemic justification, which is to say, the way how we know that the thing is there? Since (to my knowledge) Lewis made no attempt to address this in his (1986) work, it may be charitable to interpret Lewis as being a kind of pragmatist. [E7] On epistemic pragmatism, a thing is justified iff is it useful to believe that it is. However, I simply do not share this intuition at all. Consider the case where it is useful for University of Sydney student 311202969 to pass his PHIL2600 philosophy exam. Does it therefore follow that student 311202969 is epistemically justified in thinking that he really will pass his exam? That is not at all obvious. Rather it seems like if student  311202969 thinks that way, and has no other reason to think he will be justified (such as having studied sufficiently well and written a cogent couple of essays with valid premises and sound arguments respectively…), then student 311202969 will not be justified in the right way[E8] . His justification is just not robust enough. In the same way, modal realism, so justified, has insufficient justification and is therefore not strictly known, under our definition. But if David Lewis does not strictly know that modal realism is true on our definition (assuming it holds), why postulate concrete possible worlds in the first place?

A second, metaphysical, issue confronts Lewis on his own grounds. As previously stated, Lewis thinks that severe ontological inflation is the only cost of his theory, but that the benefits far outweigh the cost. Let us grant the usefulness justification despite our immediately preceding epistemic criticism. Now it becomes possible to craft a new theory of possible worlds whilst retaining all of Lewis’ intuitions about how possible worlds work and not having such inflation. This comes from theism, specifically, divine conceptualism about possible worlds. One of the more amusing criticisms that has been launched against modal realism by its critics is that Lewis’ theory implies that theism is true. Lewis grants this, but only with respect to a subset of all possible worlds— and the actual world is not one of those worlds. We can therefore truly say ‘God does not exist’ in the actual world just because, relative to the actual world, God does not exist. But then it also seems maximally obvious if we grant Lewis’ own theory for sake of argument that we could think of a possible world extremely close to our world. Everything in this new possible world is this same as the actual world, except for one different, namely, God exists. God would be a being with a particular set of properties of personhood, will, power, knowledge, and rationality. He would be timeless, immaterial, and spaceless under the relevant conditions. In short, God would be a single immaterial being with many ideas that one would expect a maximally knowledgeable, intelligent and wise conscious being to have. But now let us make a new assumption. Let’s suppose for sake of argument that the world in which this divine being exists is the actual world. On this view, there is therefore a mind in the world with a multiplicity of concepts—concepts about possibility and necessity, properties, instantiation, and universals, content and closeness, etc. Moreover, this being would also know and perhaps be able to think about different ways the creation could have been. In other words, this kind of being would have ideas about the actual world and the possible worlds. But in that case, the thoughts of this being could serve as the placeholders for all the relevant functions Lewis attributes to his concrete possible worlds. Except now, in the divine conceptualist case, there is no ontological inflation. Instead there would be a single divine mind—a relatively simple concept—with complex ideas, indeed. But the ontology is relatively simpler and neater than modal realism[E9] . If ontological inflation counts as a cost of Lewis’ theory, and we can craft a new theory that does the same thing but without that cost, the new theory ought to be preferred. The only ‘cost’ of the new theory of divine conceptualism is that it commits its holder to theism. I, however, would not see such an implication as a ‘cost’ but as a feature. That, of course, depends on one’s prior metaphysical commitments.[E10]

In conclusion, then, we have provided two reasons to think that modal realism is not the best solution for systematic metaphysics.  It suffers from epistemological issues and metaphysical issues. We firstly saw that modal realism is plausibly not justified in the right way, and secondly that divine conceptualism about possible worlds (possible worlds as complex divine ideas) plausibly provides all the benefits without the costs of Lewis’ theory if we want to take the line of possible worlds functioning the way they do as truth-makers for various metaphysical theories as Lewis explains it.


Question 6: What does Putnam’s Twin Earth thought experiment really teach us about reference? You may wish to discuss variations on the experiment, and the different things they may teach us?

In his Twin Earth thought experiment, Putnam (1973; 1975a) seeks to show that how our terms refer to things is a causal-historical matter about how the external world is, not a matter of our psychological states alone. Contextually, in large part, it is a critique of Gottlob Frege and his famous “intension determines extension” dictum from his paper Sinn und Bedeutung (Sense and Reference) and his/Bertrand Russell’s classical descriptivist theory of reference where the meanings (semantic contents) of names are identical to the descriptions associated with them by speakers, while their referents are determined to be the objects that satisfy these descriptions. This essay first seeks to characterise the thought experiment properly, and then tries to offer two interesting criticisms from the philosopher Ǻsa Wikforss concerning the internal coherence of the thought experiment. Finally, we consider a form of causal-historical theory of reference exemplified in the writings of David Lewis, Frank Jackson, and David Chalmers, among others, which the thought experiment brings out in the history of the discussion about reference. We shall call this new theory ‘neodescriptivism.’

First, we must characterise the thought experiment in an appropriate way.   In his Twin Earth thought experiment, Putnam (1973; 1975a) asks us to think about ordinary, Earth-water. Water is a natural kind substance on Earth with a certain set of macroscopic emergent properties S – e.g. colourlessness, odourlessness, tastelessness, and thirst-quenching abilities. If an Earthling, called OscarE, sees the relevant substance and observes properties S, his psychological state would correspond to the S properties such that he would refer to the natural kind by use of the natural kind term ‘water.’[2] Call this substance ‘waterE’. So OscarE’s psychological state captures the meaning of ‘water’ in the right way. Call this psychological grasp of the meaning capturing ‘S-wise.’ We are then invited to imagine that there exists a planet somewhere in our very galaxy that is exactly similar to Earth: a Twin Earth. However, Twin Earth is different in a single way. Twin Earth possesses a substance with an identical set of macroscopic emergent properties S to waterE. Call this substance ‘waterTE’. This substance is also referred to using the natural kind term ‘water’ by speakers of the English language on Twin Earth—people such as OscarE’s real Twin Earth counterpart OscarTE. OscarTE also captures meaning of ‘water’ S-wise as he refers to waterTE. So the psychological states of OscarE and his real-world counterpart OscarTE are exactly similar, since they both capture the meaning of the term ‘water’ S-wise. However, there is a difference between waterE and waterTE. WaterTE has an entirely different microscopic molecular arrangement to waterE. Whereas waterE is microstructurally H2O at the molecular level, waterTE is not H2O but is rather another complex chemical arrangement, abbreviated XYZ. The externalist intuition then follows that, whereas OscarE and OscarTE are thought to be in exactly similar psychological states (capturing the meaning of ‘water’ S-wise), their respective usage of the term ‘water’ nevertheless means different things, because their respective usage of the term ‘water’ picks out different objects in the world. OscarE’s term ‘water’ picks out waterE which is H2O but OscarTE’s term ‘water’ picks out waterTE which is XYZ. On Putnam’s view, natural kind terms serve as rigid designators of objects in the world. So if ‘water’ refers to H2O, it refers to H2O in every possible world. Therefore, if OscarE was on Twin Earth, pointed to waterTE and said ‘water’, he would be saying something false, and so too for OscarTE in the relevant alternative scenario, because their respective usage of the term ‘water’ did not correspond to the object required to be referred to by the term[E11] . Therefore, externalism about semantics is true, because there was something about the external world that determined the meaning of the referent ‘water’ in both contexts for both speakers. As Dale Jacquette explains (2013, p.74), Putnam’s Twin Earth thought experiment tries to show that thought-content or psychological states (what is ‘in the head’) is insufficient to determine the referent of our thoughts, and that falsifies semantic internalism, the thesis that all meanings and referential phrases [E12] just are psychological states.

Wikforss (2008, pp.168-171) raises some interesting concerns about Twin Earth scenarios. According to her, Twin Earth style thought experiments depend on examples of the ambiguity of common speech descriptions to provide conditions for something’s falling under a term. It is ambiguous if the phrase ‘Water is H2O’ is predicative or numerical identity. Putnam’s thought experiment critically depends on the point that we intuitively agree with the Kripkean claim that the statement ‘Water is H2O’ expresses a numerical identity claim, such that having the H2O molecular structure is a necessary and sufficient condition for being water in every possible world. This is because, for Putnam (1973, p.710), natural kind terms serve as rigid designators. [E13] But this claim is invites very intuitive Twin Earth style counterexamples to the claim which render it more plausible than not that the Kripkean thesis (regardless of its truth or falsity) is jointly unsatisfiable with Putnam’s thought experiment, even though the experiment depends on it. For instance, we could easily imagine another possible world where a substance on Earth both looks and feels like what we might call ‘tar’ with the relevant set of macrostructural emergent properties   but which also has the microstructural molecular structure H2O. After all, Putnam’s own thought experiment about allows for situations such as this. But, according to Kripke (and, a fortoriori, Putnam), this kind of a substance ought to meet the necessary and sufficient conditions for being the natural kind water and being able to be referred to by the natural kind term ‘water’. But, apparently, waterTE as XYZ does not meet similar conditions, despite having all the relevant emergent properties S exactly similarly to what we ordinarily call ‘water’ on Earth, waterE. That seems extraordinarily counterintuitive. [E14] Perhaps the (FE) semantic externalist could say that being H2O is a necessary but not a sufficient condition for counting as the natural kind water. Water must also exhibit the relevant macrostructural emergent properties S and not . Granted. But then it seems impossible for the thought-experiment to launch the assertion that waterTE is even able to emerge into properties S to be captured S-wise by both OscarE and OscarTE, since the claim now sounds like saying that whenever H2O is present, properties S are present[E15] . But H2O is not identical to XYZ. Therefore, whenever XYZ is present, properties S are not present. But then the thought-experiment does not work on its own merits, because there is no possible state of affairs where OscarE and OscarTE can be in exactly similar psychological states S-wise about H2O and XYZ respectively. So the thought experiment fails if the externalist advocate of the thought experiment takes this alternative route in response to criticism. Otherwise he should just bite the bullet and accept the incredibly implausible claim that the tar-like H2O substance really is water after all, and that our term ‘water’ really refers to that tar-like thing ins some possible world. But that just seems crazy and should be rejected.[E16]

However, even if the externalist advocate of the Twin Earth thought experiment and its conclusions should disregard my arguments against the thought experiment above as misconceived and false so presented, it does not seem as if semantic externalism would go through in every instance of kind terms even if the thought experiment works. At most, Putnam’s semantic externalism about meaning and reference would only be true in specific local instances. As Wikforss (2008, p.177) explains, “The Twin-Earth argument, if accepted, supports a thesis within foundational semantics… This, again, suffices to accomplish what Putnam set out to accomplish, i.e., [to] reject the idea that the meaning of kind terms is always determined ‘internally’. However, it does not support his grander claims, such as the famous dictum ‘meanings just ain’t in the head’… What follows is that in the case of some terms (and we may not know which, prior to scientific investigations) external factors sometimes (when the speaker is not very knowledgeable) play a role in the determination of meaning and content[E17] .” Whereas it is true that facts about the external environment play a role in fixing the reference of some of our terms, it may still be true that there do exist some terms whose descriptive referents are completely internally determined, and are therefore psychological states[E18] . All the semantic internalist would need to do is to adjust his formal thesis to accommodate local instances of externally-derived referential content, even though perhaps a great number of meanings are entirely internally derived.

Since the late 20th and early 21st centuries, there have been a growing number of philosophers such as Frank Jackson (2004) and David Lewis (2007) who have tried to revive something akin to (but not identical to) the old classical descriptivism of Frege and Russell. These so-called neodescriptivists have tried to hold onto the externalist intuitions about what fixes the referents of our terms, but in a different way to Putnam. When OscarE goes onto Twin-Earth, points to water TE and says ‘water’, neodescriptivists still think OscarE has said something wrong, but not for the kinds of reasons given by Putnam in his own thought experiment. Rather what neodescriptivists think is that, for OscarE and Oscar TE respectively, there has been a causal-historical chain of acquaintance for each respective person has had in the world, such that acquaintance with a particular substance has fixed the reference of each one’s respective usage of the term ‘water.’ Whatever a term refers to, it refers to it in virtue of the context in which it is spoken, and the causal-historical chain of understanding obtained by the individual in each respective context[E19] . For OscarE, he is acquainted with H2O, such that a story about his use of the term ‘water’ and therefore its referent can be traced back to interaction with this substance. Therefore whenever OscarEsays ‘water’ he refers to H2O. This is why when he says ‘water’ on Twin Earth, he makes a mistake. However, one seeming implication of this kind of a view is that, if OscarE began to study and to see how Twin Earthlings spoke their language and then learned about Twin Earth XYZ molecular arrangements emerging in properties S, OscarE  might justifiably and understandably begin to think of this XYZ as a different or new kind of water, and baptise XYZ ‘water.’ Then, in the alternative Twin Earth context, OscarE  would begin to use the term ‘water’ to refer to this new thing, XYZ, but if he went back to Earth he would use the same term to refer to H2O. The same term now has two different referents because of context and acquaintance.  (Of course, one could always craft an exactly similar scenario in reverse for OscarTE on Earth.)  So the causal-historical account of reference has an interesting and seemingly plausible explanation of how terms obtain their referential content, and the Twin Earth externalist intuitions probably push us in this direction.

What we plausibly really learn about reference then, if Putnam’s Twin Earth scenario circumvents Wikforss’ and my criticisms, is probably something akin to what the neodescriptivists were getting at: that terms refer to different things in different contexts, and obtain their referential value from my causal-historical acquaintance with particular substances in the world. This is an attractive view since it allows room for linguistic development and change, whilst providing a robustly cogent account of its occurrence[EE20] .

[1] I am aware of the famous Gettier (Knowledge = JTB) counterexamples, which is why I don’t take the flat “JTB” line about knowledge. I see justification simpliciter and justification ‘in the right way’ are two distinct kinds of justification. Call the previous J and the latter J*. I take J* to be just the kind of justification, something which is precisely not like the kind of justification J which is (on the Gettier case counterexamples) from mere luck or simple falsity. I cannot get into the issue here because of space. But it is not a cheat. I simply take the kind of J* we’re looking for to be something from more robust, reliable sources than the J kind of justification in the Gettier counterexamples. This justification J* will now feature in my critique of David Lewis.

[2] Just to be pedantic, according to the philosopher Åsa Wikforss (2005, pp.67-69), Putnam did not believe that there could be provided a set of necessary and sufficient conditions for naming any term T a natural kind term. Nevertheless I am going to assume a brute, pre-analytic definition of a natural kind term applies, namely, as a term that picks out a natural kind. I will also take a standard modern definition of natural kinds to be true. According to one standard modern definition (Bird and Tobin, 2012), ‘To say a kind is natural is to say that it corresponds to a grouping or ordering that does not depend on humans.’ This is the sense in which I will refer to water as a natural kind.

 [E1]Very nice introduction.




 [E5]“… which also did just as well on the positive grounds that Lewis notes….”

I presume that you meant to say this as well.

 [E6]An alternative to this would be to just argue that S knows that P only if (but not if) P and S has a justified belief that P. That justification is necessary for knowledge is still very plausible, even in light of Gettier cases. And you only need the “only if” part of the JTB theory to make your case.

 [E7]It may be charitable to do so. There is a strong sense in which Lewis argued (as he quite often did) that if we believe P, we can say lots of nice things; therefore we should believe P.

Lewis would perhaps respond to your challenge in the following way. Look at what scientists do. We might presume that we have good epistemic reasons to believe in the existence of electrons. Why do we think this? Why do we accept the theory of electrons?

Here is a very plausible account, and one which I think many philosophers of science would accept in broad outline at least. There were a bunch of phenomena that scientists had noted which were in need of explanation. Things like electrical charges and photoelectric effects and so on. We knew we needed something to explain what was going on behind all these different phenomena.

We could not, and we still cannot, directly observe electrons. But we believe that there are such things because by positing their existence, we can explain a bunch of stuff which needs explanation. And we can explain it in a really cool way; i.e., the explanation is simple, straightforward, coherent, and powerful. We could have come up with lots of different explanations for all these different phenomena that we were seeking to explain. Instead, however, we came up with a single solution to a whole bunch of explanatory problems—i.e., electrons—and this solution was simple, elegant, and it just worked really, really well. In fact it works better than anything else we’ve been able to come up with.

Now sure, we had to posit the existence of a certain kind of thing that nobody had ever seen before, and in fact it was part of the theory that we can’t hope to see it directly even if we wanted to. But by just assuming that it’s true we explain a whole bunch of stuff in a really neat way, so we’re going to believe it even without any direct observational confirmation.

Lewis thinks the situation is basically analogous when it comes to possible worlds qua concreta. Science gives us epistemic (not merely pragmatic) reasons for believing in things we can’t see. It does this when these things we can’t see do a really, really good job at explaining a bunch of stuff. What he’s doing is no different than scientists do all the time. And nobody is going to deny that science gives us good reasons to believe in the existence of a whole range of different unobservable entities. So why can’t he do the same thing with possible worlds? (On a related note, as a student of Quine, and in light of Two Dogmas, how much of a distinction do you think Lewis drew between pragmatic and epistemic considerations?

 [E8]An important question is whether this case is analogous to Lewis’s case. My discussion in the previous comment should give some reason to pause here. The case you come up with is obviously not one where pragmatic considerations provide epistemic justification. But there are many respects in which Lewis’ case is importantly different.

 [E9]There are different kinds of ontological inflation. One kind of inflation is multiplying the number of entities. Lewis’s account does this. Another kind of inflation is multiplying kinds of entities. One of Lewis’s central claims is that his theory doesn’t commit this kind of inflation, and that this is a positive point in its favour. He posits just more stuff which is of the same kind of stuff as we are already well aware exists, such as the kind of stuff of chairs and tables. Is positing a God, like this, positing a new kind of thing? The God being posited is very much unlike anything I’ve ever been familiar with.

More issues: how simple is this thesis? It is ontologically parsimonious with respect to number of entities. But where did this God come from? Does it raise more questions than it solves? Another issue when doing inference to the best explanation is fit with background assumptions and overall a priori plausibility.

Another issue to consider is how much do you really need to assume this God to be like in order for it to do the work you’re supposing? You have said the God is immaterial and timeless—but why is this necessary? Why must it be a single being? Why must it be omniscient even, or maximally rational?

One final question: if talk about modality becomes talk about different ideas and the relationships between ideas in the mind of some strange entity, then why do we care about modality? And why do we suppose ourselves to have access to truths of modality? Lewis might have some struggle with these questions as well, but they appear answerable. I am not sure why anyone would care about modality if this theory were true. A theory of modality needs to be consistent also with the basic claim that understanding modality is important. (Lewis’s answer to this would relate to his thesis of Humean Supervenience.)

 [E10]I think this is a good essay. I have given long comments because I think you raise some interesting issues. The length of the comments is not supposed to indicate any severe issues with the paper.

Your points are well argued, original, and the whole thing is well written.

 [E11]This point isn’t important for the thought experiment you have described, i.e., one where Twin Earth is in the same galaxy as actual Earth. It becomes important when we suppose that Twin Earth doesn’t actually exist, but could possibly exist (which is a variation on the thought experiment).

 [E12]Be precise: semantic internalism doesn’t say that referential phrases are psychological states; this would be an absurd thing to assert.

 [E13]Since you’re taking this route, it would have been better earlier on to discuss the thought experiment where T-Earth is in a different possible world.

 [E14]Intuitions diverge on this point. Many would be happy to accept that at the counterfactual scenario you have described, water is very tar-like.

 [E15]I cannot see why this follows. The claim is that for something to count as water, it must be H2O and also satisfy the properties S. This has no implications for whether being H2O metaphysically necessitates that S is satisfied.

Perhaps you mean this: if the former claim, ‘water is H2O’, is taken as a numerical identity claim, then it would follow from the above plus the identity claim that whenever we have H2O, we have the properties S. And this would seem problematic.

The correct thing to do, it seems is for the semantic externalist to deny your intuition that the tarry H2O-substance isn’t water. A distinction between considering a world as actual, and considering it counterfactually, might help them here. Don’t you find it plausible to say something like “Water could have been opaque and viscous”? That seems plausible to me. And if so, then it seems that water, i.e., H2O, could have been tar-like.

 [E16]How bad would this really be?

 [E17]Not even this follows.

 [E18]Again, be precise. Nobody thinks terms are psychological states. Some people still do think that the referents of our terms are determined entirely by some associated, internal psychological states.

 [E19]Isn’t this exactly what Putnam (and Kripke) says? I don’t see how you’ve characterised neodescriptivism here, which is essentially a tweak of the older descriptivist position so as to avoid Putnam style problems. This certainly doesn’t serve to characterise the positions of Jackson, Lewis and Chalmers as distinct from those of Putnam and Kripke.

 [EE20]This essay begins excellently, with a powerful critique on the coherence of Putnam’s thought experiment.

Unfortunately it is let down in the second half. The account of neodescriptivism is either very incomplete or mistaken, and it’s not fully explained why this position was mentioned in the first place. Slightly let down by not considering responses to Putnam from advocates of narrow content, or neodescriptivist discussions, which would have helped draw out more lessons from Putnam’s experiment.

Overall, however, it is not a bad paper. Again, it is very well written and clear, and you provide an interesting argument in the first half.


Get every new post delivered to your Inbox.

Join 42 other followers