I was recently re-reading an article by Nicholas Kaldor and J. Trevithick  and I came across this fine description of rational expectations:
The main plank of the monetarist school has hitherto been that inflation is invariably ‘demand induced’: it can result only from an excessive demand for goods which, however, can manifest itself in the prevalence of excess demand in the labour market [footnote i]. In either case, any consequential increase in output or any fall in unemployment below its natural rate can occur only temporarily.
This latter view, which was shared until recently by the great majority of monetarist economists, is now contradicted by a more radical group of monetarists who developed the notion of ‘rational expectations’ and applied it to the study of inflationary processes. Their position is an extension of the argument that the increase in employment induced by monetary and fiscal policy is the result of some form of ‘cheating’ since workers had expected higher real wages than they actually received, whereas employers had expected to pay a lower real wage than they ultimately had to pay.
It is now claimed by this group of American monetarists that the above theory assumes that expectations are formed on an irrational basis, whereas it is in the interests of all economic agents to form a ‘correct model’ of how the economy functions. The proper cognition of the economy enables rational expectations to be formed which will prevent all but ‘surprise’ departures from an equilibrium path and will, therefore, render nugatory any attempt to reduce unemployment below its ‘natural’ level even in the short run. The centrepiece of this argument is that both workers and employers realise that the quantity theory of money is correct and that wages and prices must rise in the same proportion as the money supply. As a result, it is argued that increased expenditure will cause increases in wages and prices directly without affecting real variables such as output, employment or the real wage rate. They contend that they will base their expectations not on a projection of past trends in the price level or one of its time derivatives (such a procedure would usually be ‘irrational’) but on the ‘correct’ understanding of the economy which takes changing trends into account. Although the mechanism through which prices and wages rise is unclear, this school by-passes the traditional mechanism by which they rise under the pull of excess demand only. The corollary of this hypothesis is that inflation can be reduced far more painlessly than was thought by early monetarists, for, provided that the government can convince the public that it has a firm intention to get the money supply under control, the price level and the level of money wages will respond with only a very short lag: it does not require appreciable restriction of demand in real terms or any abnormal fall in employment even for a temporary period. [footnote ii]
This rational expectations theory goes beyond the untestable basic axioms of the theory of value, such as the utility-maximising rational man whose existence can be confirmed only by individual introspection. The assumption of rational expectations which presupposes the correct understanding of the workings of the economy by all economic agents—the trade unionists, the ordinary employer, or even the ordinary housewife—to a degree which is beyond the grasp of professional economists is not science, nor even moral philosophy, but at best a branch of metaphysics.
[footnote i: Harry G. Johnson, ‘What is Right with Monetarism’, Lloyds Bank Review, April 1976]
[footnote ii: It is well known that in the last five years, many Western countries have experienced the phenomenon of rising unemployment coupled with accelerating inflation. This appears to undermine the validity of the traditional natural rate hypothesis and, a fortiori, the rational expectations version. Professor Friedman (‘Inflation and Unemployment’, Institute of Economic Affairs, 1977), has acknowledged this divergence between monetarist theory and empirical observation, but he is hard-pressed to explain it.]
… orthodox macroeconomists came to conflate ‘being rational’ with thinking like an orthodox economist. What this implied was that agents knew the one and only ‘true model’ of the economy (which conveniently was stipulated as identical with neoclassical microeconomics) …
 Kaldor N. and Trevithick J. 1981. A Keynesian Perspective On Money, Lloyd’s Bank Review. (reprinted in Collected Economic Essays, Vol. 9)