Originally published in The Philosophical Quarterly, Vol. 44, No.176(July 1994)
(Hist-Analytic.org)SPEAKING OF SELVES By Bruce Aune In an essay called "The Transcendental `I'," the late John Mackie asked what theme unites the images of Zeno Vendler mounted on a horse and watching through Hannibal's eyes the battle of Cannae, Peter Strawson blazing a trail through a jungle of phenomena, and Elizabeth Anscombe floating in a tank of tepid water.1 The answer he gave was "Of course the transcendental `I'." To his question about philosophical worthies doing strange things, I want to insert a reference to Roderick Chisholm enjoying the company of other incorporeal monads and to John Mackie himself, whittling down what he considered the partly Cartesian concept of I and you and she to the bare concept of "a subject, whatever it may be," of experiences and actions. Also, in place of Mackie's answer, I want to defend what I think is a more satisfactory one--namely, that what unites most of these worthies supposedly doing such strange things is not a transcendental “I” but a vivid imagination and some bad arguments. I shall say just enough about their arguments to expose an instructive error that seems to motivate exotic theories of "the self."I begin with Roderick Chisholm, whose remarks on my topic are the most recent of the group I mentioned. In an essay called "On the Simplicity of the Soul" Chisholm expressed the belief that he is a monad, a "simple substance that is a soul." He did not argue that this belief is true or probable; he merely tried to show that it is, as he put it using the words of William James, "very much of a live option." He tried to show this by defending the thesis that persons are "metaphysically unique" in having "a nature wholly unlike anything that is known to be true of things that are known to be compound physical things."2 To defend this thesis he adopted what he called a Cartesian approach, presupposing that he could conceive entia rationis such as properties or attributes and could "tell them apart...and see what they logically require to be exemplified." He also presupposed that "our mental properties provide us with the most assured information that we have about any individual thing or substance" and that, on the basis of what we know about our own thinking, we may derive conclusions about our metaphysically unique nature (pp. 167-69).
I am not enthusiastic about the Cartesian assumptions that Chisholm inserted into his discussion, but the criticism I want to make does not depend on my attitude to those assumptions. What it does depend on is the plausibility of a premise he used and the reasoning that probably made it acceptable to him. His explicit inferences, the ones he laid out for the reader, are very simple and unproblematic; to give you a reasonable sense of the peculiar view he defended, I want to begin by describing them. As you might expect, they are related to a fairly complicated definition that is built on other definitions.3
At the beginning of his paper, Chisholm explained some of his special terminology. He said he uses the word "soul" in the way that St. Augustine, Descartes, and many others have used it: to mean the same thing as "person." He also said that by the term "simple substance," he means an individual thing (a particular) that is not compound, or made up of further things--as (to use his example) a chair is made up of a back, a bottom, and some legs. His belief that he is a monad thus amounts to the belief that he is a person, a thinking substance, that is not made up of further things: unlike a chair, he lacks physical parts. This reference to him, the person Roderick Chisholm, is, as he might say, "strict and philosophical," for in a "loose and popular" sense he can no doubt be said to have physical parts since he can be said to "have" a body.4 Yet his body, which includes a brain that he concedes is perhaps physically necessary for his ability to think, is not, strictly speaking, a part of him: it is something that he makes use of in his earthly life. He sees with his eyes, he said (p. 171), in the same sense in which he sees with his glasses; evidently, he moves by means of his legs and thinks by means of his brain.
As I said, Chisholm did not attempt to prove that he is a monad lacking physical parts; his aim was merely to show that this view of himself is a "live option." To accomplish this aim, he offered an argument supporting a very limited thesis about his "metaphysically unique" nature. Although this thesis, as he initially stated it, is focused on his nature rather than the nature of persons generally, he intended it to have this greater import, and he therefore used "we" rather than "I" in formulating his argument. The premises he used are these:
His explicit conclusion, which he reached by an obvious intermediate step, is, as I said:(1) We have qualitative properties (our mental properties are examples of them).
(2) The qualitative properties we are acquainted with "are known to be possibly such that [they] are exemplified by simple substances."
(3) No qualitative property is known to be such that it may be exemplified by compound substances.
(4) We have a nature which is wholly unlike the nature that anything known to be a compound thing is known to have.
Chisholm's second and third premises are supposed to be Cartesian; "they tell us," he said, "what rational beings can know about the nature of the psychological properties they have" (p. 177). The first premise is not explicitly Cartesian, for Descartes did not speak of qualitative properties; but its acceptability can be determined, if Chisholm is right, by Cartesian criteria. The reason for this is that qualitative properties, as Chisholm defines them, are supposed to be invariably psychological. Our knowledge that we have such properties is therefore assured, if Chisholm is right, by the Cartesian assumption that our mental properties provide us with the most assured information that we have about any individual thing or substance. If our mental properties provide us with this kind of information, we certainly know, if we have perused his definition, that we have "qualitative properties.”
You cannot be unfair to a writer if you accept his assumptions, and for the sake of argument, at least, I shall accept Chisholm’s definition of a psychological property and his Cartesian assumptions about psychological properties. Having accepted these things, I must then concede that the first and second premises of his argument are acceptable. But what about his third premise? Do we not know that this premise is false? Do we not know that psychological properties are exemplified by rational animals, which are certainly compound substances in Chisholm’s sense? And do not I, a normal human being, actually have, in a perfectly strict sense, a back, a seat, and some legs? Chisholm obviously thinks these questions deserve negative or skeptically noncommittal answers, but he cannot expect his readers to agree with him without further argumentation. How, in fact, can he confidently suppose that the creatures of flesh and blood you see and speak to every day are not known to exemplify a purely qualitative property or that no one knows of a qualitative property that could be satisfied by a compound entity? Does Chisholm have any rational basis for his attitude to the third premise of his argument?
If he has such a basis for this crucial premise, it does not appear in his paper. In fact what he does say in the paper should have moved him to reject it. When he defended his second premise, he appealed to the Cartesian assumption that he can conceive various properties and "see what they logically require in order to be exemplified" (p. 169). His claim was that, when he contemplates the mental properties he is acquainted with, he can see that they do not require compound substances in order to be exemplified: it is "logically possible" that the things having them are unextended beings (p. 171). Does he also see that these properties require simple substances for their exemplification? Does he see that things having them must not have proper parts--that it is "logically necessary" for them to be unextended? The answer, I should think, is "Surely not." The supposition that thinkers are compound substances is every bit as consistent, logically, as the supposition that they are simple: neither supposition is disfavored on purely logical grounds. The sort of considerations that Chisholm advances on behalf of his second premise can be advanced, therefore, in opposition to his third: the two premises are at odds with each other. Chisholm should have seen this incompatibility and, on the basis of it, looked about for better premises.
Although Chisholm's argument does not, all things considered, really support the conclusion he wished to draw, a Cartesian might still be encouraged by the idea that the simplicity of the soul is at least a "logical" possibility. This encouragment would be unfounded. Chisholm's claim that the property of thinking does not logically require that the things having it have proper parts may be acceptable, but it merely tells us (if it is true) that the proposition "S has the property of thinking" is logically consistent with the proposition "S does not have proper parts"--that neither proposition logically implies that the other is false and that their conjunction does not logically imply a contradiction. These facts (assuming them to be such) are not metaphysically exciting because they are shared by "Tom is married" and "Tom is a bachelor." The conjunction of these latter propositions may represent a conceptual impossibility, but it is not formally contradictory. Chisholm may, of course, have used the term "logical possibility" very loosely, so that it encompasses what I would call conceptual possibility. But even the conceptual possibility of a thinker lacking proper parts does not insure that "S lacks proper parts" is consistent with the conjunction of "S has the property of thinking" and other propositions that we know to be true of thinking things or have good grounds for accepting in regard to them. A proposition of this last kind might be "A thinking thing is a highly evolved animal, a creature consisting almost entirely (if not wholly) of physical parts."
Towards the end of this paper I shall have more to say about the strategy of Chisholm's attempt to add credibility to his views about his self, but I now want to turn to Elizabeth Anscombe, the next person on my list. The image of her in a tank of tepid water is represented in her essay, "The First Person."5 Unlike Chisholm, Anscombe has no suspicion that she is a monad (she is wholly confident that she is a living human animal), but she is capable, she said, of being temporarily lulled into thinking of herself as a "Cartesian Ego" when she reflects on a standard assumption about the use of the first-person singular pronoun. To guard herself against these insidious erroneous thoughts, she found it necessary to take some very strong philosophical medicine. Taking the medicine amounts to rejecting the assumption that "I" is a referring expression and insisting that the statement "I am Elizabeth Anscombe" (uttered by her) is not an identity proposition of the form "I = E.A." In my view this medicine should not be prescribed. To avoid the erroneous thoughts she mentioned, less drastic remedies are available.
Anscombe began her paper by observing that if Descartes' argument for the nonidentity of himself with his body were valid, a conclusion that he could have reached with equal validity is "I am not Descartes"--Descartes being the man born of such-and-such a stock, christened "René," educated at La Flèche, and so on. The words "I am not Descartes" (which he might have used) admittedly sound strange, but they are not objectionable if Descartes' presuppositions are acceptable. Anscombe's conditional endorsement of these alternative words is focused on the pronoun occurring in them, which is not, she said, the ordinary reflexive pronoun but a peculiar reflexive, which grammarians call the "indirect reflexive" (p. 22). She found it difficult to characterize the use, or specific function, of this reflexive except by saying that "it can be explained only in terms of the first person." What she meant by "explaining" this pronoun can be illustrated by one of her examples. Suppose Smith fails to realize that he is Smith; suppose he thinks "I am not Smith." If I, Aune, am asked to explain just what Smith fails to realize, I have no means of identifying how Smith conceives of or (to use Anscombe's expression) mentally latches onto the entity to which his pronoun "I" presumably refers. Lacking a special, privileged access to Smith's mind, the best I can do is to proceed as I did and say: "He doesn't realize that he is Smith; he thinks `I am not Smith'."
The claim that Smith actually thinks "I am not Smith" in this case might be inaccurate, however, for Smith might not use the nominative first-person singular pronoun in relation to himself: he might customarily use his first name instead (as I once did). What he thinks, in this case, might be accurately rendered by "Bruce is not Smith." Although the point is not vital for my argument here, I think it is worth observing that Anscombe's remarks about the peculiarities of the indirect reflexive do not, in fact, succeed in identifying a special use or role for "I" or any other pronoun. In a reflexive statement such as "Sam said that he is a philosopher," the pronoun occurs in the context of an indirect quotation; and what some philosophers take to be a peculiarity of pronouns in such constructions is actually a peculiarity of the sentential context. The point can be seen by reflecting on the sentence "Sarah said that she would see him today." The that-clause reports what a person, Sarah, supposedly said, but it gives no indication of how she identified herself, the man to whom she spoke, or the day represented as "today."
I spoke earlier of the entity to which Smith's "I" presumably refers because of Anscombe's contention that the pronoun is not actually a referring term. Her reasons for this implausible contention are related to her belief that, if an expression were a referring term, a person using it would have to have some conception of the entity to which, in using it, he intends to refer. If the expression were "I," it would also have two kinds of guaranteed reference, she said: the object or thing to which it refers must exist when it is so used, and that thing must be what the user conceives it to be: if the user conceives it to be a thing X, it must be X (p. 30). Since, in Anscombe's opinion, a use of "I" could carry this stringent guarantee only if its referent were freshly defined for the occasion and also remained in view so long as the reference remained in effect, she thought the only plausible candidate for such a reference would be a Cartesian Ego or, more exactly, a "stretch" of one: the entity would be "the thinking that thinks the thought." This conception of the referent of "I" is authentically Cartesian, she says (p. 31), because Descartes described the essence of res cogitansas nothing but thinking, and a stretch of this process is an episode or activity. Thinking of the referent this way is an intolerable outcome, however; the idea that "I" is actually a referring term should be abandoned.Why did Anscombe believe that an episode of thinking is the only plausible candidate for the referent of the first-person pronoun? The answer is "Because the plausible alternative, that `I' stands for a demonstrated body, is even more problematic." As she put it (ibid):
Imagine that I get into a state of "sensory deprivation." Sight is cut off, and I am locally anaesthetized everywhere, perhaps floating in a tank of tepid water: I am unable to speak, or to touch any part of my body with any other. Now I tell myself "I won't let this happen again!" If the object meant by "I" is this body, this human being, then in these circumstances it will not be present to my senses; and how else can it be "present to" me?So many unanswerable questions arise here that she turned almost with relief to the Cartesian conception. But she saw that as hopeless, too. Apparently forgetting her claim that an authentic Cartesian res cogitans is an act or episode of thinking rather than a persisting thing, she asked (ibid): "How do I know that `I' is not ten thinkers thinking in unison?"The case Anscombe made for her position here is very dramatic, but it is certainly not successful. We must remember that in describing her state of sensory deprivation she was not defending a thesis restricted to aberrant uses of “I” produced by subjects who cannot detect their bodies: she was defending a general thesis covering standard uses of this pronoun. But this general thesis is certainly not supported by the imaginary situation she described. As far as the actual world is concerned, the conditions she imposed on “I” as a genuinely referring expressing are regularly satisfied. One condition was that persons using “I” referentially must have some conception of the entity to which, in using it, they intend to refer. Actual persons using “I” (including Anscombe herself) clearly satisfy this condition: they invariably conceive the referent of their “I”s as themselves—people with certain names, who live in certain places, have (or had) certain friends, and so on. The other conditions she mentioned are also satisfied in actual cases. The person referred to always exists when the pronoun is so used, and that person is always the person the speaker takes the referent of the pronoun to be: when the speaker takes the referent to be X, it always is X. Since these conditions are satisfied when “I” is used in actual cases, even Anscombe should accept it as a referring term in those cases.
I am not being unfair to Anscombe here. She made it clear in the course of her article that she actually thinks of herself as a thinking animal, and “a certain thinking animal” (one that is a philosopher and has the name “Elizabeth Anscombe”) could certainly serve as a conception by which she might mentally "latch onto" a referent for her first-person pronouns. Such a conception is certainly not rendered inadmissible by her counterfactual supposition that she is floating, "locally anaesthetized everywhere," in a tank of tepid water. By hypothesis, it is a thinking animal that is floating in this way. It is true that, in imagining herself so floating, she might imagine that she would wonder if she even had a body. This hypothetical uncertainty might well alter the conception she would have of what she means when, “locally anesthetized everywhere,” she speaks of herself; it might even render nonsensical the words she would then utter. But the possibility of her being in this imagined state of mind has no implications for her actual self-concept (let alone the actual self-concepts of others) and for the referring or non-referring use of actually used pronouns. Just think of the analogous possibility: if I were dazed by an automobile accident, I might answer the question “Are you all right?” by saying “Yes, you think so.” In giving this answer, I would not (I stipulate) be using “you” as a proper second-person pronoun, but this hypothetical misuse on my part does not show that, when I use the word in a normal state of mind, I do not refer to an appropriate “second” person.
I now turn to John Mackie, who included Anscombe in his group of philosophers who have drawn false or extravagant conclusions from premises involving the first person singular pronoun. Like Chisholm, some of these philosophers have thought that the pronoun refers to a Cartesian ego; others (Anscombe was Mackie's example) claim that it doesn't refer to anything; and others still say that it refers to a "bare form of consciousness," a "mere frame in which any picture fits" (p. 15). Mackie rejected such conclusions himself, arguing that "I" is not only a regularly referring term, but one regularly referring to the human being (the particular rational animal) who uses it. This sort of reference is secured, he argued, by two different rules which render the pronoun ambiguous, involving two distinct uses that normally coincide but can, in imagined cases such as the one Anscombe offered, diverge. He was inclined to think that the ordinary concept of I is, on account of one of these rules, partly Cartesian but that a minor conceptual reform, which he supplied, can easily remove the Cartesian taint.
Mackie did not formulate the two rules explicitly; his claim was merely that the first rule "links" the pronoun "I" to the human being who uses it, whereas the other (perhaps as he reformed it) links the pronoun to the subject, "whatever it may be," of certain thoughts and experiences. (Anscombe, had she known of it, might have used this rule to link the pronoun to an attenuated concept that might "latch on" to a subject in her imagined case.) Although the subject specified by Mackie's second rule is, he said, "contingently identifiable" in normal cases (that is, identifiable on contingent grounds) with a certain human being, it may have a divergent identity in imagined cases. Thus, a Cartesian may imagine that he (the referent of his "I") may exist without a physical body, and Zeno Vendler said he can imagine himself being Hannibal, sitting on a horse, and seeing the battle of Cannae through his own eyes. These thoughts or imagined identities would not make sense if "I" were used in accordance with Mackie's first rule, for a human being could not exist without a body and Zeno Vendler could not be identical with a Carthaginian general who died more than two thousand years ago.
Astute as his discussion is, I think Mackie went astray in claiming that "I" is ambiguous. As I see it, "I" is simply an indexical (or token-reflexive) expression tokens of which standardly refer, on each occasion of their use, to the subject producing them. When I use this pronoun in a normal, assertive way, I always refer to myself, a male human being living in the 20th century. This use of "I" accords, obviously, with Mackie's first rule, but it diverges from his second, which is supposed to account for, or render coherent, the cases of personal "transference" imagined by Vendler and others. I regard this divergence as a virtue because such cases are far more problematic than Mackie allows.
If, sitting at my desk in my house in the New England village of Montague Center, Massachusetts, I imagine that I am Hannibal sitting on a horse and observing the battle of Cannae, I am imagining something that I know to be impossible: a person living in the twentieth century cannot be an ancient general watching an ancient battle. From a semantical point of view, the identity statement cannot be true at a single possible world. This semantic impossibility need not come to mind, however; and I may coherently entertain the thought "I am Hannibal sitting on my horse, etc." if I imaginatively disregard many of my current beliefs. (The trick is to ignore at least one of every pair of conflicting beliefs, so that nothing contradictory or incoherent is before one’s mind.) To draw any interesting conclusions from the identity hypothesis I thus entertain, I must obviously hold in mind more than a few disorderly fragments of my normal beliefs about myself and Hannibal; I must have some reasonably coherent picture of what I am imaginatively taking myself to be. Different pictures involving different assumptions are clearly possible here, but the conclusions I draw will be acceptable only in relation to one or the other of them: they will not be acceptable generally. Since the interesting conclusions Mackie and Vendler can reasonably draw from the unusual suppositions they make have this relative or dependent status, I want to say a bit more about the sort of dependency they will possess.
An instructive starting-point here is Quine's problem-sentence beginning, "If Verdi had been French..."6 How might we complete this sentence if we are thinking about La Forza Del Destino? Some philosophers might advise us to inspect the "closest" possible worlds in which Verdi is French, but I should prefer to attend to the beliefs I have about the actual Verdi and the actual opera. If I focus my attention on the belief that La Forza Del Destino was written by an Italian, I might complete the sentence with the words "Verdi would not have written La Forza Del Destino." If, on the other hand, I attend to the fact that Verdi did, after all, write this opera, we can conclude with "La Traviata would have been written by a Frenchman." Obviously, if I have no beliefs about La Forza del Destino, I can draw no conclusion about it. What I do conclude about Verdi and the opera depend, therefore, on the beliefs I have or, more generally, on the assumptions I make. No relevant conclusion is acceptable absolutely: the ones I reach in the cases just considered are, in fact, false. Essentially the same considerations apply, I believe, to the imagined identities Mackie discussed. They make sense only in relation to special sets of suppositions, and the conclusions they warrant--like the conclusions warranted by Anscombe's supposition that she has no awareness of a physical self--have a purely relative significance.
The implications of all this for the subject of "the self" are easily developed. Although the pronoun "I" standardly refers (as I believe) to the person actually uttering it and not some minimal subject of mental attributes, we can conceive of its referent more or less abstractly. If we think of such a referent very abstractly--as, say, a mere thinker of this or that thought--we set aside much of what we know or believe about that subject. A good deal of this setting aside must occur in the imagined cases Mackie described, for they make sense only on this condition. On the other hand, his cases warrant interesting conclusions only in relation to special sets of suppositions. Thus, if I seriously imagine myself to be Hannibal observing the battle of Cannae, I must ignore my belief that I am in Montague Center, Massachusetts, that I am a professor at the nearby university, and that I am living in the twentieth century. Of course, to distinguish the thought "I am Hannibal" from the thought "Hannibal is Hannibal" and to draw relevantly interesting conclusions from my imaginative hypothesis, I must explicitly attach some of my beliefs about myself to Hannibal (e.g. that he is perceiving these trees, that he has these hands, this nose, and these thoughts), and I must, in turn, disregard some of my beliefs about Hannibal (e.g. that he is not now alive). As in the case of Quine's "belief-contravening" supposition about Verdi,7 any interesting conclusion I draw from my imagined identity (e.g. that I am a general or that I have a hundred elephants) is acceptable only in relation to a body of suppositions; it is not warranted by the imagined identity itself.
A little more detail, and a change of perspective, may be helpful here. When people imagine things about existing beings, they commonly ignore much of what they know about those beings. This is true not only of philosophers making suppositions but of children at play and ordinary adults watching soap operas or Greek tragedies. Plato, struck by the fact that mimesis tends to “bewitch” an audience, temporarily robbing spectators of true beliefs about what is actually before them, concluded that guardians of an ideal polis should have minimal exposure to mimetic art.8 Observers of dramas often object to Plato's conclusions, insisting that their knowledge is increased rather than diminished by serious drama; yet if they feel pity and terror on seeing an actor with a wooden mask feign emotion before them, their sense of reality is at least temporality disrupted, and they are ignoring facts that would ordinarily be obvious to them. (No one is really hurt or dying on the stage; some actors are merely pretending to suffer or pretending to die.) Something similar happens when a child imagines he is a pirate with a real sword rather than a yardstick and when a philosopher imagines he is an ancient general. The philosopher's sense of reality may not (possibly unlike that of the child) be temporarily distorted, but many of her beliefs are, for the moment, ignored (put out of mind) and others are imaginatively attached to new subjects.
If I imagine that I am Hannibal sitting on a horse and watching a battle, I shall continue to think of Hannibal as a Carthaginian general--perhaps as the famous general who used elephants in crossing the Alps and attacking the Romans. In thinking of myself as Hannibal, I shall therefore be thinking of myself as this military commander. (In doing this I shall be making tacit use of the logical principle that if a=b and Fa, then Fb.) Since, apart from my thoughts about Hannibal, I believe that I am sitting on something, x, and looking out on a field, y, with various trees, my thought that I am Hannibal makes me imaginatively disregard the fact that x is a chair and y is located in twentieth-century Massachusetts and think of them as, respectively, a horse and a plain in ancient Italy. I may also think of the river near y, which I know to be the Connecticut, as the Aufidus, and I may think of the village I know to be at the foot of the hill on which y is located as Cannae. In this way I gerrymander the beliefs that come to mind regarding Hannibal and the person I normally take myself to be. By disregarding illusion-shattering facts about both Bruce Aune and Hannibal (e.g. that the former was not alive in ancient times and that the latter is now dead), and tacitly making generous use of identity logic in relating other facts about myself and Hannibal, I work out a consistent scenario about myself being Hannibal watching the famous battle.
What I take to be common to imaginative feats such as this last one and the philosophical arguments I have cited by Chisholm, Anscombe, and Mackie is that well-known or at least generally accepted facts about a subject are imaginatively disregarded and various suppositions are considered in their place. Although I, in my imaginative identification with an ancient general, tacitly drew some erroneous conclusions about myself, I did not endorse those conclusions: I knew they were merely fanciful. The same is not true of Chisholm, Anscombe, and Mackie, however: they did endorse the conclusions they drew about the nature of selves or the significance of certain pronouns referring to them. This, I contend, is a mistake. Instead of falsely concluding from their suppositions that "I" is ambiguous or nonreferential, Anscombe and Mackie should have acknowledged that, figurative uses aside, "I" is actually a referring term that standardly refers to the person uttering it. And instead of falsely concluding from a special set of Cartesian suppositions that the referents of that pronoun may actually lack physical bodies and thus be (in a fundamental respect, at least) something less than what we normally conceive them to be, Chisholm should have admitted that those suppositions merely allow one to conceive of those referents in abstraction from the physical attributes that we normally take them to possess: the suppositions provide no evidence that those referents are not, in fact, thinking animals and could exist without physical bodies.
These last remarks disclose the error, recurrent in philosophical psychology, that I want to expose in this paper. Although it occurs in discussions devoted to selves or persons, it is really a logical error involving the presumption that, because a certain proposition is reasonably inferred from or shown to be consistent with a set of assumptions that one can coherently suppose or imagine to be true (or, as in the case of Chisholm, take to be basic necessities), it represents a metaphysically significant possibility or the sort of live option that Chisholm hoped to defend. The presumption here is clearly erroneous since propositions so related to such assumptions in some cases represent patent impossibilities--a contemporary philosopher (Zeno Vendler or myself) being an ancient general watching an ancient battle and a man I have seen and spoken to (Roderick Chisholm) being an immaterial monad. I confess to serious uncertainty about how metaphysically significant possibilities are properly identified, but like live options they cannot be ascertained merely by a criterion of formal consistency or intuitive plausibility. Live options must at least be epistemic possibilities--things that, for all we know or (better) have good reason to believe, may well be true. To identify this sort of possibility, obviously we must pursue a strategy very different from the one involving the error that I have been discussing. Instead of restricting ourselves to the implications of some special set of assumptions or presuppositions, we must consider everything we know (or reasonably believe) to be pertinent to our subject.
To accuse another philosopher of making a definite error is a serious matter, and a cautious reader might protest that, in the case of Chisholm at least, no identifiable error was actually made. Chisholm made it clear, the reader could say, that he was taking a Cartesian approach to the subject and that he was presupposing (among other things) that "our mental properties provide us with the most assured information that we have about any individual thing or substance." Since he wanted to show only that a Cartesian view of the self is a "live option," an epistemic possibility in which he, at least, has a vital interest,9 it was perfectly reasonable for him to support that option by demonstrating its consistency with "the most assured information that we have." Our most assured information; is what we are most entitled to say we know, and epistemic possibilities (in the strict sense) are things consistent with this "most assured" information.
To deal with these last observations, I must clarify one matter and be a bit more precise about another. The matter needing clarification is the criticism I made of Chisholm. As far as his specific argument is concerned, my criticism was that one of his premises (the third one) is highly dubious if not simply false. Conceivably, Chisholm might wish to defend that premise, but but if he does so, he will undermine his second premise, for the considerations he aduced in favor of his second are at odds, as I explained, with his third. Apart from my criticism of his specific argument, I have tacitly criticized the basic strategy he employed in supporting his belief that a Cartesian view of persons is still a live option. This latter criticism is, in my view, far more significant than the former. In attempting to show that the option he favors is viable, Chisholm focused his attention on a limited number of alleged metaphysical certainties and thereby neglected much (perhaps most) of what we actually know or have good reason to believe about the nature of living human beings. This strategy (or argumentative procedure) has been common in philosophy since the time of Descartes, and I have been principally concerned to discredit it.
The items whose treatment requires a bit more precision are the notions of an epistemic possibility and a live option in philosophy. Normally, one thinks of an epistemic possibility as something consistent with what we (or someone) knows.10 Yet if we think of knowledge very strictly, so that something is known only if it satisfies a conception of knowledge to which there are no plausible counter-instances, we will have to accept as epistemically possible all sorts of things that we may believe on good scientific grounds to be too unlikely (too improbable or too far-fetched) to be worth taking seriously. The example of a person dropped from a high-flying airplane without a parachute provides a case in point: the probability that she will live to tell the tale is so low that it is hardly worth thinking about; yet there have been cases of survival, and until the outcome of a particular fall has been ascertained, one cannot be said to know what it will be. Since many epistemic possibilities are unlikely in the extreme, we cannot reasonably characterize a live option by reference to an epistemic notion as weak as that of strict epistemic possibility (as it might be called). A live option should not be (in relation to available evidence) too far-fetched to be worth taking seriously.
Given what I have just said, we cannot identify live options by an epistemic condition focused on "the most assured information that we possess." What we are most assured of is what we are certain about, and the sphere of epistemic certainty does not include the high probabilities we attach to factually well-founded opinions. The conjectured defense of Chisholm given above is therefore ill-considered. If we take the totality of our well-founded opinions into account, the idea that we are monads, though perhaps a strict epistemic possibility, is too far-fetched to be taken seriously at the present time. 11
Bruce Aune
UMass at Amherst
--------------------------------------------------------------------------------NOTES
[1] John Mackie, "The Transcendental `I'," in Mackie, Selected Papers, vol. II (Oxford: Clarendon Press, 1985), pp. 15-27.
[2] See Chisholm, "On the Simplicity of the Soul," in James E. Tomberlin, ed., Philosophical Perspectives, Vol. 5 (Atascadero: Ridgeview, 1991), pp. 167-181.
[3] His principal definition concerns "qualitative" properties. For a statement of this definition, see Chisholm, p. 174.
[4] Chisholm has often made use of Bishop Butler's distinction between "loose and popular" and "strict and philosophical" senses of words. See, for example, his book Person and Object (London: Allen and Unwin, 1976), Ch. 3.
[5] Reprinted in volume 2 of her collected philosophical papers, Metaphysics and the Philosophy of Mind (Minneapolis: University of Minnesota Press, 1981), pp. 21-36.[6] See W.V.O. Quine, Methods of Logic, third ed. (New York: Holt, Rinehart, and Winston, 1972), p. 21.
[7] This is Nicholas Rescher's expression; see his discussion of such suppositions in his book, The Coherence Theory of Truth (Oxford: Clarendon Press, 1973), ch. XI.
[8] At Republic 413c Socrates speaks of pleasure and fear as things that bewitch us and rob us of true opinions. The context makes it clear that he thinks mimetic poetry (of which drama is a prime example) does the same thing.
[9] For James' discussion of a live option, see "The Will to Believe" in Albury Castell, ed., Essays in Pragmatism by William James (New York: Hafner, 1951), pp. 88-109.
[10] See the article "Possibility" in the Encyclopedia of Philosophy, vol. 5 (New York: Collier Macmillan, 1967), pp. 419-424.
[11] For helpful comments on earlier drafts of this paper I thank Gareth Matthews, Lynne Baker, Ernesto LePore, and especially an anonymous referee for this journal.