By the time My Philosophical Development was published in 1959, forty seven years after The Problems of Philosophy, Russell’s ‘realist’ and ‘atomist’ phases had given way to a form of empiricism grounded in the first-hand experience of common sense but informed by Hume’s philosophy and complemented by some of the conclusions of modern psychology. The late Russell continued to object to Kant for placing epistemological knowledge inside the subject, through the ‘categories’ while leaving the extra-mental in the object itself. His analytic realism had been an attempt to put Kant’s philosophy on its head by conferring objectivity on the abstract world and subjectivity on the sensible one.

In the final stage of his philosophical career, Russell still agreed with Descartes on the importance of analytic reasoning but also with his definition of the brain as consisting of ‘thoughts’, defined as ‘all that of which we are conscious as operating in us.’ (‘Principle IX). ‘Thoughts’ in the brain may become ‘events’ of the mind, the latter being connected by memory-chains. As for Berkeley, after expressing strong objections about the absence of direct physical causes in his theory of mental events, Russell eventually came to accept his fundamental contention that everything we know is ‘mental’, an argument which, back in 1927, he judged ‘unsound and Berkeley’s conclusion improbable, though not certainly false.’

In the same way that ‘sense-data’ had lost their reliability when objects were no longer regarded as dependent on them, universals and particulars also shed their initial appeal after more rigorous critical scrutiny. His final clarification on this point reflects an admission of error: ‘I spoke about universals and our knowledge of them with a confident assurance which I no longer feel, though I have not any new opinions on the subject which I feel prepared to advocate with equal confidence.’ Having dropped a purely logical interpretation of particulars, he was now prepared to define them in terms of ‘qualities’ and spatio-temporal relations. He almost fell in agreement with Leibniz’ to the extent, that only our ignorance make names for necessary complex matters, a need which would cease if our knowledge were complete.’

Russell’s views did not change regarding our perception of the external world insofar as it remained clear to him that ‘what we directly experience cannot be the external world with which physics deals, and yet it is only what we directly experience that gives us reason to believe in the world of physics.’ In The Problems of Philosophy, he argued that physical objects are inferred entities which are the causes of our experience. However, nearly fifty years later, Russell’s theory of perception was still impeded by the fact that any inference beyond sense-data is actually incapable of being empirically tested.

Embedded in language, belief lies at the heart of Russell’s theory of knowledge and his late views on such a central notion also shows a retreat from his earlier certainty: ‘Belief is not a precise concept … it is a certain state of body or mind or both’ which does not necessarily require a direct experience as in the person being told that ‘a car is coming’ and who may be showing a physiological or psychological reaction of fear, anxiety or indifference. Russell’s theory of belief evolved beyond the strict analysis of the relations between words in complex propositions when he abandoned the notion of subject and realised that beliefs could be identified in bodily states and not attached to linguistic statements: ‘truth and falsehood both belong primarily to beliefs and only derivatively to propositions and sentences.’