I believe that Russell's change of heart was due to his discovery of the work by Frege [On denoting 483]. He looked at what had been done by Meinong and how his own work in Principles of Mathematics was along the same lines, at least in terms of allowing for the existence of things that did not really exist. After discovering Frege, he reconsidered and was able to see that there were some issues with his own work. He then set to improve on Frege's work and correct the mistakes in his own, hence On Denoting. Russell points out that if we allow that a denoting phrase will "express a meaning and denote a denotation" there is trouble where the denotation is not present. The missing denotation makes the proposition false, rather than nonsensical. This is fine as Russell says that denotation is not that which "is concerned in propositions which contain denoting phrases [484]."

Quine suggests that Russell had originally adopted a conception that allowed every word to refer to either a thing or an entity. This allowed for an inflated universe full of non-existent impossibles. When Russell published On Denoting, he was able to cast out this problem by allowing singular descriptions as the means. In this way, according to Quine, "reference to fictitious objects can be simulated in meaningful sentences without our being committed to the objects [659]." It appears this is suggesting that Russell changed his mind based on the ability to get rid of unwanted clutter rather than based on the realization of any potential absurdities that came with the original method detailed in Principles.

Russell gives some insight into his own evolution of thought by explaining the role of the theory of descriptions and how it helped to get around the difficulties with non existing entities 'this object has the odd property of not existing' [x]. He mentions that the next step was to get rid of classes. These, and the cardinal numbers, became 'symbolic or linguistic conveniences, not genuine objects' [x]. He mentions that Whitehead convinced him to throw out 'points of space, instants of time, and particles of matter, substituting for them logical constructions composed of events [xi].' He further reports that by taking logical constants to be part of the language and not what the language is about, logic is found to be more linguistic than he had thought it to be originally [xi]. Yet, 'logical propositions are true in virtue of their form. He mentions that he does not know how to explain clearly what that means, but that it is what must be solved to properly define logic [xii].' Then he brings us to the "question of the contradictions and the doctrine of types."

-- BenjaminZamzow - 09 Oct 2006