According to Keynes (1936) uncertainty plays a crucial role for holding liquidity, especially money. Uncertainty is understood not as risk which could be presented as a singleton probability measure on the set of events, but as fundamental uncertainty about the underlying structures, economic relationships, the inferences that could be drawn from past experience, etc. Such fundamental uncertainty arises due to the uniqueness of an event, the novelty of an economic activity or technology as well as to the lack of knowledge about the underlying economic causal relationships or the fact that knowledge is inconclusive for probabilistic inferences. Some investment decisions are unique in the sense that they are not repetitive and have no long record of experience regarding the distribution of returns. Most investment decisions incorporate a specific “new” element so that accumulated knowledge is of limited use to form expectations about future outcomes. The agent has simply no objective basis to determine reasonable probability measures. In a similar fashion also Knight (1921) called for fundamental uncertainty in the sense that the agent doubts his own probability measures since they are based on vague and subjective knowledge. It is widely discussed that this fundamental uncertainty as opposed to risk and also to ambiguity requires a new route in the theory of decision making (for example Dequech (2000a), Dequech (2000b), de Carvalho (1988), Fontana and Gerrard (2004), Rosser (2001), Wray (2006)).

Some economists claim that fundamental uncertainty is an omnipresent and unavoidable phenomenon since the economic process evolves in historical time, the unknown underlying economic structures may change over time and produce therefore a non-ergodic trajectory of data (Davidson (1987), de Carvalho (1988)). This would exclude any Bayesian rationality. It should be noted, however, that nonergodicity is not proven to be an overall empirically relevant phenomenon and that it should not be invoked as a “nihilistic” argument against any form of expectation formation (Rosser (2001)). Also Keynes’ main point was to think about how a rational agent behaves in presence of uncertainty, not to disprove any rationality. But the point is not (only) the question of non-ergodicity. It is typical that among economic agents as well as among economists there exist different and partially conflicting views about reality, and the existing empirical data does not clearly rule out most of them and does not give a clear evidence for only one point of view. Thus, beliefs are dispersed which reflects fundamental uncertainty, and it is a matter of rationality that agents will respond to this fact in some way.