What then of my knowledge of the minds of others? On Locke's view there can be only one answer: since what I know directly is the existence and contents of my own mind, it follows that my knowledge of the minds of others, if I am to be said to possess such knowledge at all, has to be indirect and analogical, an inference from my own case. This is the so-called "argument from analogy" for other minds, which empiricist philosophers in particular who accept the Cartesian account of consciousness generally assume as a mechanism for avoiding solipsism.
Observing that the bodies of other human beings behave as my body does in similar circumstances, I can infer that the mental life and series of mental events that accompany my bodily behavior are also present in the case of others. Thus, for example, when I see a problem that I am trying unsuccessfully to solve, I feel myself becoming frustrated and observe myself acting in a particular way. In the case of another, I observe only the first and last terms of this three-term sequence and, on this basis, I infer that the "hidden" middle term, the feeling of frustration, has also occurred.
There are, however, fundamental difficulties with the argument from analogy. First, if one accepts the Cartesian account of consciousness, one must, in all consistency, accept its implications. One of these implications, as we have seen above, is that there is no logically necessary connection between the concepts of "mind" and "body;" my mind may be lodged in my body now, but this is a matter of sheer contingency. Mind need not become located in body. Its nature will not be affected in any way by the death of this body and there is no reason in principle why it should not have been located in a body radically different from a human one. By exactly the same token, any correlation that exists between bodily behavior and mental states must also be entirely contingent; there can be no conceptual connections between the contents of a mind at a given time and the nature and/or behavior of the body in which it is located at that time.
This raises the question as to how my supposed analogical inferences to other minds are to take place at all. How can I apply psychological concepts to others, if I know only that they apply to me? To take a concrete example again, if I learn what "pain" means by reference to my own case, then I will understand "pain" to mean "my pain" and the supposition that pain can be ascribed to anything other than myself will be unintelligible to me.
If the relationship between having a human body and a certain kind of mental life is as contingent as the Cartesian account of mind implies, it should be equally easy - or equally difficult - for me to conceive of a table as being in pain as it is for me to conceive of another person as being in pain. The point, of course, is that this is not so. The supposition that a table might experience pain is a totally meaningless one, whereas the ascription of pain to other human beings and animals that, in their physical characteristics and/or behavioral capabilities, resemble human beings is something which even very young children find unproblematic.
How is this to be accounted for? It will not do, in this context, to simply respond that a table does not have the same complex set of physical characteristics as a human body or that it is not capable of the same patterns of behavior as a human body. Because the Cartesian position implies that there is no logical connection between the mental and the physical, between the possession of a body of a particular kind and the capability for consciousness. Physical differentiation can and must be acknowledged, but it can play no role in any explanation of what it is to have a mental life.
I am surrounded by other bodies, some of which are similar to mine, and some of which are different. On Cartesian principles such similarities and such differences are irrelevant. The question as to whether it is legitimate for me to ascribe psychological predicates to entities other than myself, which the argument from analogy is designed to address, cannot hinge on the kind of body that I am confronted at a given time.
Assuming the validity of the Cartesian position, we have to infer that it makes as much or a little sense, on these premises, to attribute any psychological predicate to another human being as it does to attribute it to a table or a rock.
On these premises, it makes no sense to attribute consciousness to another human being at all. Thus on strict Cartesian principles, the argument from analogy will not do the work that is required of it to bridge the gulf between my conscious states and putative conscious states that are not mine. Ultimately, it must be confessed that on these principles I know only my own mental states and the supposition that there are mental states other than my own ceases to be intelligible to me. It is thus that solipsism comes to seem inescapable.
If the above argument is valid, it demonstrates that the acceptance of the Cartesian account of consciousness and the view that my understanding of psychological concepts derives, as do the concepts themselves, from my own case leads inexorably to solipsism. However, it may fairly be said that the argument accomplishes more than just this. It can, and should, be understood as areductio ad absurdum refutation of these Cartesian principles. Viewed from this perspective, the argument may be paraphrased as follows:
If there is no logical connection between the physical and the mental, if the physical forms no part of the criteria that govern my ascription of psychological predicates, then I would be able to conceive of an inanimate object such as a table as having a soul and being conscious. But I cannot attach any intelligibility to the notion of an inanimate object being conscious. It follows therefore that there is a logical connection between the physical and the mental: the physical does form part of the criteria that govern my ascription of psychological words.
