Transitivity revisited (Q1813375)

A survey is given of the research on transitivity and it also discusses the relation between transitivity and rationality. First the different forms of transitivity are considered (full transitivity, negative transitivity, semi-transitivity etc.) It has been frequently observed that not only groups, but also individuals are not very consistent in their choices. This has given rise to probabilistic theories of choice, with stochastic versions of transitivity. Experimental testing of these concepts has appeared to be problematic. Intransitivity has been mostly studied by considering intransitive triples, though it might as well occur among larger numbers of alternatives. The possibility of intransitivity of indifference has long been recognized, however strong preferences are generally assumed to be transitive. Much energy has been expended in justifying transitivity of weak or strong preference. But also the transitivity of strong preference may be doubted. Sometimes choices result from a judgement rather than from a choice. In other cases agents are ignorant of the exact state of their preference. Preference judgement is not always very precise. This calls for a probabilistic model. The author distinguishes true preferences and true choices from observed choices. Intransitivity may be due to error, if the true preferences are transitive, but they may also result from intransitive true preferences. The possibility of true or ``genotypic'' intransitivity calls for a theory to explain it. The literature on this topic is surveyed and the author presents his own model, where preferences are generated by a function \(\varphi(x,y)\) on tuples of alternatives, each alternative being a vector \((x_ 1,x_ 2,\dots,x_ n)\) of attributes. Under certain assumptions on \(\varphi\) this model allows for intransitivity. Finally the rationality issue is discussed.