The Hypothesis of Rational Expectations:

In the General Theory (Keynes, 1936) we noted that the state of expectations was taken as given. There was, in addition, explicit recognition that changes in other independent variables including policy variables could lead to changes in expectations. However, nothing could be said in general about the nature and extent of such shifts without specific knowledge of the prevailing psychology of the economic agents, which would be influenced by prevailing social and political circumstances.

In a model that aims to provide a probability distribution for dependent variables using data on objective measurable variables, however, changes in expectations cannot be assumed to be dependent on non-quantifiable factors outside the model. Expectations themselves must therefore be explainable in terms of objective measurable variables. 

Suppose economic agents are assumed to be rational in the sense that they seek to Rational Expectations and best achieve their objectives, subject to external constraints on their choice of actions. Economic Suppose also that the degree of 'correctness' of the expectations on the basis of which economic agents act is a sufficiently important for determining their welfare. Then individual decision makers too, like the economists who make predictions on the basis of these objective conditional probability distributions, will try to learn about and make decisions on the basis of these objective conditional probability distributions.

The above is the hypothesis of rational expectations. It implies that the subjective(individual)) probability distributions that individual economic agents are assumed to use in making their decisions in an economic model are consistent with the objective conditional probability distribution implied by the model.

In most economic models, it is assumed that the decisions of economic agents are dependent only on one or two parameters of the subjective probability distribution they have for future values of relevant variables and not on the entire distribution. Often under the assumptions of a model, only the mathematical expectation or expected value of this probability distribution is relevant for decision-making. In this me, instead-of assuming that the subjective probability distributions that economic agents have coincide with the objective probability distribution plied by the model, it is sufficient to assume that the expectations of these distributions are equal.

The latter case may therefore be called the weak version of the rational expectations hypothesis in contrast to the strong version, which assumes that the entire objective probability distribution is known.