The focal point of the problem of the transformation of Keynes's method is determining whether or not he eventually changed the philosophy developed in his 1921 Treatise on Probability. One faction contends that Keynes's method changed and the other that it was constant. In this paper I posit that both viewpoints are at least partially true. In his early stage, Keynes applied his theory of probability both to the researches of economists and the behavior of economic agents. But upon reading Ramsey's critique, ‘Truth and Probability’ (1926), Keynes noticed the danger to which the application of his theory of rational inference to the conduct of economic agents could lead. Subsequently, he adopted the assumption of conventional behavior as seen in his General Theory of Employment, Interest and Money (1936). But he maintained his rational inference as a methodology of economics. We can verify this by reviewing Keynes's critique of Tinbergen's method of macroeconometrics (1939). Considering that Ramsey's optimal growth theory and his expected utility theory are roots of stochastic models of optimal growth theory and real business cycle theory, it is very interesting to notice that the contemporary methodological confrontation between Keynesians and neoclassicists dates back to the birth of macroeconomics.