Abstract:
Economic intuitions concerning rational behaviour in interactive social situations are shaped by idealized models which are regarded as "approximately true". But ideal models cannot be meaningfully deemed approximately true unless asymptotically convergent processes imply them as limit cases. We illustrate by various examples - infinitely patient customers on durable monopoly markets, homogeneity of commodities, super-games etc. - how this necessary methodological requirement is almost routinely neglected. On this basis we draw some conclusions concerning the continuity between abstract and less abstract models on the one and the world modelled by them on the other hand.