....Previous

The Principle of Transmissibility does not require that the evidence which confirms the consequent must be the very same evidence which confirms the antecedent.³⁰

Thalberg, as we have seen, questions the PDJ principle and challenges the genuineness of Gettier's counterexamples. In this latter task he finds a comrade in C. G. New.³¹ New makes his point by referring not to Gettier's original counterexamples but to one of Lehrer's Gettier like counterexamples which we shall have occassion to examine in greater detail later on. The situation is: S believes that someone in his office owns a car on the basis of his wrong belief that Mr. Nogot owns a car. The conclusion however is true because another of S's colleagues, namely, Mr. Havit owns a car.

The following, according to New, is the basic structure of Lehrer's Gettier like counterexample if we lake 'A' to stand for Mr. Nogot and 'B' for Mr. Havit:
(i) I believe that A owns a car
(ii) I am justified in believing that A owns a car
(iii) I deduce from the proposition "A owns a car" the proposition "Someone owns a car"
(iv) A does not own a car
(v) B owns a car.
(vi) I do not believe that B owns a car
(vii)  "Someone owns a car" is true because B owns a car
(viii) I have justified true belief that someone owns a car but cannot be credited with that knowledge.

One of the essential conditions of Lehrer's counterexample is that the inference from "A owns a car" to "Someone owns a car" is a valid inference. C. G. New contests this claim. He holds that the inference is valid only under a certain interpretation of the term 'someone'. The proposition "Someone owns a car" admits of three different interpretations. They are:

This objection, as Gilbert Harman observes, is misleading. It seems that here New is confusing between 'entailment' and 'mutual entailment'. When two propositions mutually entail each other they are logically equivalent. If a proposition is inconsistent with one of the logically equivalent propositions, then it must be inconsistent with the other proposition too. But a proposition may be inconsistent with the premise of an inference but may quite well be consistent with the conclusion of the inference. Thus if we derive p v q from p.r then — r may be inconsistent with the premise but not with the conclusion.

Further, it is not clear why New thinks that the inference is valid only if the first of the three interpretations of the conclusion "Someone owns a car" is accepted. In fact, if we take the first interpretation then the inference amounts to: A owns a car; therefore, A owns a car. It amounts to saying that the rule of existential generalisation is invalid. For the element of generalisa­tion is ruled out, if we accept the first interpretation. This is exactly the charge that is brought against New by James Smith.³²

Faced with objections raised by Harman and Smith, New modified his argument. He says that he docs not question the validity of the rule of existential generalisation. He only specifies the contexts under which it is valid. In one context, the premise may express one statement; in another context, it may express quite a different statement. The same is true of the conclusion. Hence it is quite possible that in one context it is valid while in the other it is not. Thus, according to New, the validity of the rule of existential generalisation is context bound. But, as far as we know, the validity of this rule, as would be testified by all logicians, is context free. It is valid in any and every context. It is valid by virtue of its form.

New further argues that when we use the expression 'someone' we may mean some identifiable person or some unidentifiable person.³³ The word 'someone' is thus ambiguous. The ambiguity is cleared up when some qualifying clauses are added. Thus we may say, "Someone took the book, I must find out who" or "Someone took the book, I must not tell who". In the former sentence 'someone' means an unidentifiable person, while in the latter 'someone' refers to an identifiable person. Accordingly, New holds that the sentence "Someone owns a car" may mean either of the following three sentences:
(i) Some identifiable person owns a car
(ii) Some unidentifiable person owns a car
(iii) Someone, an identifiable or an unidentifiable person, owns a car. (And New further asserts that only in the first interpretation the inference is valid, in the other two interpretations it is not.)

However, it is not clear why New thinks that in the third interpreta­tion the argument is invalid. It is possible that he thinks that it is psychologically impossible to pass from a state of certainty about the person who is said to own a car to a state of uncertainty about the person who is said to own a car. But psychological impossibility, even if it is a fact, has nothing to do with the logical validity of an inference.

Next..