Argument from Analogy Is Not an Argument for Other Minds

The most common and for many the most forceful objection to the argument from analogy for other minds is that it must rely on a single case, namely, one’s own. The argument is held to be weak because it must generalize from that one case to a vast number of other cases. A standard reply (used by Ayer and many others) is that we do not need to proceed from a single case of one mind correlated with one body, that we can proceed instead from an enormous number of correlations between instances of behavior and experiences. The experiences and the instances of behavior all belong to the same person, but this does not preclude there being an enormous number of correlations between the two sorts of event. I think that this response is basically correct although it has been thought to suffer from certain weaknesses which I shall consider in some detail.

Another objection to the argument from analogy is that whether or not it proceeds from a single case in the attempt to show that experiences are associated with other bodies, it must proceed from a single case in attempting to defend the more ambitious claim that at least one mind is associated with every other active human body. It is claimed that the only analogical argument that could support this conclusion is an analogical argument based on the fact that at least one mind has been associated with one’s own body throughout its existence. For the argument from analogy to take us all the way to the conclusion that every other body has at least one mind, it would indeed have to rely in its final stage on a single case; but I will try in the first section to show that it isn’t necessary to rely on the argument to establish that much. When properly understood, the argument from analogy is not in itself an argument for other minds.

I

I believe that once I can show by multiple-case analogical arguments that it is very highly probable that other bodies have experiences and that it is probable that at various times they are having certain specific sorts of experiences (particularly ostensible memories), I can justify the hypothesis that at least one person is associated with every other body during its active existence (hypothesis PB) in much the same way as I can justify the similar hypothesis in regard to myself. And this is a method which does not involve any form of the argument from analogy and therefore a fortiori does not involve a single-case argument from analogy.

How then do I justify the hypothesis in regard to myself? At any given time, I have many memories of experiences which I seem to have had in the past. None of these memories associate these experiences with other bodies, and though some memories may not be associated with any body, I have other memories which make it reasonable for me to suppose that all of my experiences are causally dependent on my body in a way in which they are not dependent on the body of anyone else. I have had, of course, many more experiences than I can remember but there is no reason to suppose that the experiences I cannot remember differ from the ones that I can in a way that would justify me in thinking that the former were associated with some other body rather than with my own. On the basis of these considerations it is reasonable for me to infer that all of my experiences have been associated with the same body, and therefore that at least one person has animated my body throughout its existence.

When I turn to other bodies, I need only the suppositions (obtained, as I shall argue, through analogical arguments from a multiplicity of cases) that (i) those bodies have been associated with mental events throughout their lives, and (ii) that some of those mental events are ostensible memories that resemble my own, in order to defend a hypothesis for other bodies similar to that which I formed about myself. I am not offering here a single-case argument from analogy to the effect that since my body has been associated with a person, other bodies must be associated with persons. I am instead presenting a hypothesis in regard to those other bodies—namely, hypothesis PB, and defending it in the same manner as I have defended the similar hypothesis about myself— except that in the case of other bodies I must ascertain first by analogical arguments relying on a number of cases that those bodies are associated with mental events, some of which are ostensible memories similar to my own. I shall call this defense of hypothesis PB the memory-of-same-body defense.

To make it doubly clear that the memory-of-same-body defense of hypothesis PB is not a single-case argument from analogy, consider a case in which a person would not be in a position to support hypothesis PB by a single-case analogical argument yet could still support it by the memory-of-same-body defense. A is suffering from amnesia; he can remember no events prior to the last six months. He is therefore not in a position to determine that his own experiences have always been associated with the same body: for all he knows, they might have been associated with another body before the onset of his amnesia. Thus he can’t defend hypothesis PB by a single-case argument based on himself. He can, however, still defend that hypothesis by the memory-of-same-body defense. After using a multiple-case argument from analogy to determine that other persons have ostensible memories, he can use their memory reports to justify the claim that whether or not he himself has always been associated with the same body, there are good grounds for supposing that every other body has had at least one person associated with it throughout its existence. (The same would hold true in the far-fetched case where A has memories according to which his experiences were associated with one body during the first half of his life and with another during the second half. In this case, an argument from analogy would in fact suggest that hypothesis PB was false, but the claim that it was true would still be supported by the memory-of-same-body defense.) [It has been suggested to me that although multiple-case arguments from analogy aren’t suitable to the defense of the claim that there are other minds if that claim is construed traditionally as the claim that for every other human body there is one mind or at least one mind, it can be used to show that there are other minds (leaving aside the question of whether they are associated with just one body). But even in this case a multiple-case argument from analogy wouldn’t take us far enough. After it was determined by a multiple-case argument that various sorts of experiences are associated with other bodies, another sort of argument is needed to show that these experiences are joined together into minds. If it were claimed that a single-case argument from analogy is needed for this purpose, I can remind my opponent that I have already shown that such a single-case argument is not needed to support the conclusion that there are other minds associated with a single body, and therefore, a fortiori, is not needed to show that there are other minds.]

II

If my preceding contentions are correct, once I have used this sort of multiple-case argument from analogy to justify the claim that ostensible memories similar to my own are associated with other bodies, I don’t need an additional argument from analogy (and, a fortiori, not a single-case argument from analogy) to justify the claim that at least one person is associated with each human body. But it has been claimed that multiple-case arguments from analogy can’t justify the contention about ostensible memories or indeed justify any claim that experiences are associated with other bodies. It is commonly supposed that although one can correlate vast numbers of one’s own experiences with instances of one’s own behavior, all correlations after the first are redundant so that one must in effect rely on an argument from a single case and that such a basis is unsatisfactory. Budlong argues (Analysis 35.3) that just as it would be redundant for a person testing a bottle of milk for freshness to take a second glass to serve as a supplement to the first, it would also be redundant to use more than one correlation between your own mental state and your behavior. But let’s suppose that you really tried to make the argument from analogy work on the basis of a single case. You would have to imagine that you had forgotten all of your previous experiences and that you had observed only a single correlation of an instance of behavior with a mental event. Perhaps you would be entitled to infer that there were associated mental events when other bodies exhibited behavior that was just like your behavior in the case you had observed, but any inference that went beyond this either about your own body or about the bodies of others would be on a very shaky foundation indeed. I grant that you don’t need the vast multitude of correlations that one could find by a careful and thorough use of memory, or by, for example, systematically recording your experiences and the accompanying behavior, but you obviously need much more than a single correlation.

III

If I am right, multiple-case arguments from analogy are not redundant, but perhaps even with a multiplicity of correlations the argument won’t work. I shall consider a number of attempts to show that this is the case.

1. It has been urged that even if it were granted that a number of correlations between my mental states and my behavior are better than just one, the class of actions where I have observed this correlation is only a minute fraction of the class of actions about which I want to make an inference. My sample class is far too small to allow me to have anything like the confidence that we are entitled to have about the existence of mental states in other persons.

But as a matter of fact we make confident judgments about the behavior of sub-atomic particles on the basis of observations of a much smaller fraction of sub-atomic particles. Further, most opponents of the argument from analogy would remain dissatisfied even if we were only making an inference about one other person; or for that matter, even if we were only inferring that a single action of another person was accompanied by a single experience.

2. It could be argued that the behavior of other people differs too much from one’s own behavior for the latter to serve as the basis for inference. But even if everyone had exactly the same sort of body and behaved in exactly the same way, opponents of the argument from analogy would remain dissatisfied. And no one suggests that the probability of the claim that other persons have experiences varies to any significant extent depending on the similarity of those other persons to oneself. The objection seems to be rather that my sample does not include anyone who is not myself. When it is reduced to this, it carries little weight. It seems more important than it is simply because it is logically impossible to verify the claim that this particular difference is, as it would seem to be, unimportant to the analogy. If this is the true source of the objection, it comes down to no more than the objection that the conclusion of the argument is unverifiable, an objection that Hyslop and Jackson have shown to be ineffectual [A. Hyslop and F. C. Jackson, "The Analogical Inference to Other Minds," American Philosophical Quarterly, vol. 9 (1972), pp. 168-176].

3. It has been argued that (i) since I only observe that most of my manifestations of pain- behavior are accompanied by pain, I can only infer that (ii) many instances of behavioral states are associated with mental states and that it does not follow from this that (iii) the common sense belief that every body which behaves in a manner similar to my own will have many instances of mental states associated with its behavioral states is correct.

But despite the fact that it is only in most cases (but not in all) that my own pain behavior is accompanied by pain, it is incorrect to infer from this that my observations entitle me only to conclude that many rather than all instances of behavioral states are associated with mental states. On those occasions when I engage in pain behavior but am free from pain, my pain behavior is nonetheless associated with some other mental state. Thus one is entitled to infer that, for every body which behaves much as one’s own, that many instances of its behavioral states are associated with mental states.

So far so good, but we expect the argument from analogy to yield not merely the conclusion that all other human bodies are associated with mental events but also that they are associated at some time in their histories with certain sorts of mental event —with, for example, pains and ostensible memories (henceforth memories). Now as I observed before, I do not always feel pain when I exhibit pain behavior. Could it be argued on this basis that although the argument from analogy entitles me to suppose that every human body has associated mental events, it does not entitle me to suppose that every human body at some time or other has pains associated with it? Consider the argument just criticized with the modification that we stick to pains throughout rather than switching from pains to mental events halfway through.
  1. Usually I feel pain when I exhibit pain- behavior.
  2. Therefore many instances of pain behavior are associated with pain.
  3. It is nonetheless possible that when certain human bodies exhibit pain behavior, this pain behavior is never accompanied with pain.
  4. Conclusion: A multiple-case argument from analogy fails to show that it is certain that all human bodies are sometimes accompanied with pain.
The trouble with this is that its conclusion isn’t particularly damaging to multiple-case arguments from analogy. It can be admitted that we can’t be certain that for every single human body some instances of its pain-behavior are associated with pain, so long as the possibility is sufficiently slight. It might be objected that the plain man wouldn’t admit that there is even an extremely slight possibility of such exceptions. However, a philosopher needn’t be unduly disturbed if this turns out to be another case where we can only hope for moral certainty, even if the plain man thinks (or is supposed to think) that we can have absolute certainty.

But perhaps the argument can be revised so as to provide a sufficiently strong conclusion to create a real problem. Consider the following:

I observe a number of correlations between my own pain and my own pain behavior. However, I also note that there are a small but significant number of occasions when I exhibit pain behavior but feel no pain. Accordingly, whenever I observe pain behavior in another person, an analogy with myself would lead me to conclude that there is a small but significant chance that he is not in pain. It follows from this that if I were to observe all the pain behavior of everyone who exists, statistical considerations would make it fairly probable that there are some human bodies such that no instance of their pain behavior is accompanied by pain. Therefore I am not entitled by a multiple- case argument from analogy to be as sure as I am in fact entitled to be that every other body is at some time associated with pain.

This may look plausible but it isn’t; the multiple-case argument from analogy can be buttressed with too much additional evidence for the conclusion to survive. In the first place, I could note that there are only certain circumstances in which I exhibit pain behavior when I am not in pain, and if I could observe other persons long enough (which an objector to the argument from analogy would obviously allow) I could surely find occasions for any other person when he exhibited pain behavior and these circumstances didn’t pertain. More important, I needn’t limit myself to correlations between pain and pain behavior. I can also consider correlations between pain and pain stimuli because in the case of certain kinds of pain stimuli, there is a far greater correlation between my being subjected to the stimuli and feeling pain than there is between my exhibiting pain behavior and feeling pain. Finally, if one went on to talk about neural responses to stimuli, a real doubter could be led to find correlations between his own neural states and feelings of pain which would presumably be even tighter than the correlations between pains and pain-stimuli. Any doubts that remained after this would I think be perfectly tolerable.

IV

I have argued that we can establish with a reasonable degree of certainty that experiences including ostensible memories are associated with other bodies and that once this is granted, we don’t need an additional argument from analogy—this time a single-case argument—to show that there is at least one mind associated with every body. It might still be objected that it is unnecessary to use the argument from analogy because we have in any case a better reason for supposing that other bodies are ‘inhabited’ by minds: we are justified in hypothesizing the existence of other minds in order to provide a causal explanation for the rational behavior of other bodies. But this explanation would be false if epiphenomenalism were true. Probably epiphenomenalism is false but the probability of that does not appear to be as great as the probability that there are other minds or the probability afforded by the argument from analogy. Therefore it is unsatisfactory to rest the claim that there are other minds exclusively or even primarily on any argument that relies on the falsity of epiphenomenalism.

It has been suggested in response to my last remarks that there is no reason at all to give any weight at all to any analogical argument unless it involves the premiss that certain events cause other events; and, specifically, that an analogical argument for the existence of ‘other’ mental events absolutely requires the premiss that some of one’s own behavior is caused by associated mental events; and thus of course to require the falsity of epiphenomenalism. I agree to the extent of granting that the argument must claim that there are causal relations between mental events and physical events; but in terms of the argument it doesn’t matter which way the causal relations go. Interactionism would be fine, but from the point of view of the argument it would be equally satisfactory if epiphenomenalism were true for it too would guarantee that mental events accompany certain sorts of physical events; and it would also for that matter be satisfactory if mental events caused physical events but physical events never caused mental ones. I should add that the question of whether the truth of epiphenomenalism is compatible with the success of the argument from analogy is of more than academic interest: Shaffer, certainly one of the most sophisticated writers on the philosophy of mind, thinks that there is a considerable chance that epiphenomenalism is true.

Alternatively, it can be argued that we are justified in supposing that there are other minds because the hypothesis that all living human bodies are accompanied by minds is in some sense simpler than the hypothesis that you are an exception to an otherwise universal rule that such bodies are not accompanied by minds. The hypothesis that there are other minds does not claim that mental events cause physical events or even that some psychophysical events cause others so its truth could be more probable than the claim that epiphenomenalism is false.

However, in the last analysis it relies itself on an argument from analogy that would seem to be weaker than the one we rely on in the argument from analogy itself. When confronted with two hypotheses both of which account for the facts, the best reason for accepting the simpler one as correct is the fact that in the past further investigations have confirmed the simpler of two rival hypotheses more often than the more complex one. But the simplicity of the hypothesis that there are other minds is not just like the simplicity of the various other simple hypotheses that have been confirmed in the past so that we are relying here on no more than an analogy between the simplicity of the hypothesis that there are other minds and that of various other hypotheses. Further, I feel more confidence in my intuition that its being other persons is irrelevant to the analogy in the case of the argument from analogy than in my intuition that the difference between the simplicity of the hypothesis that there are other minds and the simplicity of the various other hypotheses is irrelevant. Accordingly, here again it seems to me that the argument from analogy provides the best support for our belief that experiences are associated with other bodies.

Summary

The argument from analogy should not be construed as an argument for other minds; it is rather an argument that certain sorts of experience are associated with other bodies. It is a multiple-case argument. The multiplicity of cases do not make it redundant and the number of cases is adequate to establish the conclusion of the argument with a reasonable degree of certainty. Once it has been established that ostensible memories similar to my own are associated with other bodies, I am entitled to infer without being forced to use an additional argument from analogy (this one proceeding from a single case) that at least one person is associated with every other active human body. The reports of ostensible memories that I am given would entitle me to make this hypothesis about other bodies even if my own ostensible memories did not entitle me to make it about myself.

[I am indebted to Jonathan Bennett for comments and criticisms.]


The Argument from Analogy Is Not an Argument for Other Minds
Author(s): R. I. Sikora
Source: American Philosophical Quarterly, Vol. 14, No. 2 (Apr., 1977), pp. 137-141
Published by: University of Illinois Press on behalf of North American Philosophical Publications
Stable URL: http://www.jstor.org/stable/20009660