Oskar Morgenstern, Nature's Attitude and Rational Behavior (1966)

Morgenstern’s paper examines scientific method, rational action, and moral responsibility through a reconsideration of Jacques Rueff’s account of physical and social science. It distinguishes the scientific standards economics should adopt from the physical models it should not simply imitate. Intentional interaction and uncertain consequences demand a theory of decision connecting economics with ethics. The question of nature’s possible hostility tests assumptions underlying reliable knowledge without establishing a cosmological doctrine.

The opening section credits Walras’s simultaneous equations with a major advance but questions mechanics’ continuing authority as economics’ prototype. Physics itself has changed:

The rise of relativity theory and quantum mechanics brought profound changes.

An analogy with physics cannot therefore provide economics with a permanent foundation. What economics can learn is less a particular model than a disciplined practice of inquiry:

There is tradition and experience in theory formation, in critical evaluation of data, of experiments, and profound knowledge of the difficult role of measurement.

Mathematical techniques also develop in relation to specific scientific problems. Success in one domain does not establish adequacy elsewhere; Morgenstern even entertains the possibility that different sciences might require different logics. Economic concepts must consequently be examined before mathematical tools are selected. Utility illustrates this need:

For example, "utility" has for centuries been conceived of as some kind of an inherent quality of goods.

The movement toward preferences changes what economic theory describes. More fundamentally, economic agents cooperate and conflict intentionally: outcomes depend on one another’s choices. Game theory addresses these relations directly, whereas ordinary maximization can obscure them. Morgenstern accordingly treats Walrasian equilibrium as a limiting case with little empirical relevance.

The second section separates a theory’s rationality—its availability for mathematical analysis and predictive testing—from rational behavior as the pursuit of optimal action. Such action requires anticipating consequences under alternative choices by others. Statistical uncertainty about nature differs from strategic uncertainty about an opponent: treating an adversary as a passive source of statistical regularities may produce disastrous decisions. Economic action combines these uncertainties rather than permitting either to be discarded for analytical convenience.

The third section asks what warrants confidence that nature permits reliable discovery. Einstein’s belief in a nonmalicious universe, divine veracity, and Poincaré’s interpretation of the Michelson–Morley experiment supply contrasting approaches. Human vulnerability suggests nature’s harshness, while scientific discovery suggests its accessibility; neither establishes an underlying attitude. The practical danger is that physical knowledge grows faster than humanity’s ability to manage its consequences. Successful inquiry might enable collective destruction. Morgenstern leaves unresolved both nature’s supposed hostility and whether humanity belongs within the nature being assessed. Random processes offer an alternative to personification, not a conclusive answer.

The fourth and longest section develops the ethical implications of uncertain prediction. Moral judgment must anticipate consequences, especially when novel situations exceed inherited rules. Legal interpretation demonstrates why an exhaustive catalogue of permissible acts cannot replace judgment. Ethical theory must address uncertainty fundamentally: moral commitment may have to be approximate where consequences are only statistically knowable. Morgenstern criticizes Bentham and Mill for insufficient attention to uncertainty, novelty, and the difficulty of comparing subjective experiences. Collective intelligence likewise does not guarantee that the state can foresee outcomes better.

Time, distance, and scale complicate responsibility. Deforestation and resource depletion expose conflicts between present purposes and distant harms. Responsibility depends on available knowledge, but evaluating future consequences also introduces discounting. Economics requires a moral framework, while describing that framework requires economic concepts—a reciprocal dependence supporting a unified theory of decision. Remote weapons and individually small contributions to pollution or population growth similarly obscure the relationship between action and aggregate harm.

The final section returns to inquiry itself. If nature were hostile, investigation might require indirect strategies resembling those used with responsive human subjects. Problems of observation in anthropology and physics unsettle a simple separation between observer and observed. Morgenstern thus joins methodological criticism to the problem of sustaining rational action and moral responsibility when knowledge is incomplete, other agents respond strategically, and discovery generates dangers beyond our capacity to govern.