1987 Constructing Institutions rejected or modified by others.Prefer- Preferences may be rationalized from ences can and do come sideways,from the top down,specific applications being identifications,experiences,and conver- deduced from general principles.But sations.What matters is not how prefer- complexity of the causal chains invoked ences are first proposed (many are called leaves people who lack a capacity for but few are chosen)but how they are abstract thought unable to form prefer- ultimately disposed through the presence ences.Reasoning in steps is also slow. or absence of social validation.It is not Without social validation at each step, the lone individual,after all,who creates moreover-which is difficult to achieve- what is called ideological constraint ("one the chain of reasoning may snap.For- thing entailing another")among prefer- tunately,faster methods are available. ences but social interaction among adher- People can know what they believe or ents of a particular culture in contrast to whom they trust without knowing how other cultures whose identifiers have dif- the belief is derived.Sniderman,Hagen, ferent preferences Tetlock,and Brady (1986)agree that such bottom-up processes operate on white Heuristics attitudes toward blacks.In their view Brady and Sniderman,in pursuing a It may be nearer the mark to say that citizens,so closely related question,"How ..can far as their reasoning about policy is affect- citizens make sense of groups-that is, driven,start at the beginning of the chain,taking know which is relevant to which issue and account of their feelings toward blacks.Then, rather than working their way along the chain which stands for what-without having hierarchically,from general to specific,they skip to know a great deal about them?"(1985, over the intermediate links of the chain and go 1073), straight to its end.Having reached the end of the chain,they work their way backwards and fill in focus on the operation of an affective calculus, the missing links.That is to say,not only do they or,as we call it,a likability heuristic.This reason forwards,from general to.specific;they calculus is organized around people's feelings also reason backwards,from specific to general toward groups such as liberals and conserva- And,because they can reason both forwards and tives.Clearly,many in the mass public lack a backwards,with affect guiding them,they can firm understanding of political abstractions.All indeed figure out what they think about ques. the same,many know whom they like,and, tions,such as the reasons for racial inequality, equally important,they also know whom they they may not ordinarily think about.(p.33) dislike.If coherent,these likes and dislikes can supply people with an affective calculus to figure Mediating their perceptions through their out the issue positions of strategic groups.We cultures,people can grab on to any social suggest that in this way many in the mass public handle to choose their preferences.All can figure out who wants what politically with- they need are aids to calculation. out necessarily knowing a lot about politics.(pp. "How,"Paul Sniderman and his col- 1061-62) leagues ask,"do people figure out what The more people are able to choose sides they think about political issues,given -ours versus theirs-"the more they ap- how little they commonly know about preciate the differences between the issue them?"(Sniderman et al.1986,2).They positions of the two sides.What counts, state that "three heuristics are of par- then,is not how people feel toward ticular importance:affect (likes and dis- groups,one by one;rather it is how they likes);ideology (liberalism/conservatism); feel toward pairs of opposing groups"(p. and attributions of responsibility (the so- 1075).It is precisely this pairing or,more called desert heuristic)"(p.2).The desert accurately,this triangulation of rival heuristic is a version of system versus in- cultures,I believe,that enables people to dividual blame through which adherents position themselves in political life. of political cultures seek to hold others 9