Different Paths to the Same Result Bianco 1998

Here was my response to this paper

Seth Frey 9/19/09 Memo on Bianco’s 1998 Account of Impression Formation

“Different Paths to the Same Result”, William Bianco's 1998 comparison of a psychological and a game-theoretic account of voter impression formation, can be summarized as follows: Assuming that the game-theoretic account is indistinguishable from the psychological account, the game-theoretic account is indistinguishable from the psychological account. This mapping hinges on at least two assumptions made in the paper, the first of which gives excited humans unlimited rationality, and the second of which assumes that stereotypes reflect perfect information. Given these assumptions, the two accounts of voter decision-making become identical and lead to exactly the same conclusions.

According to Bianco, “piecemeal integration is described as driven by the fit of new information with available categories” (1069). This statement quietly underwrites an assumption that voting agents, when they get motivated enough to seriously consider two candidates, can be modeled as having unlimited information-processing capability. While defensible, this assumption cannot safely be interpreted as having existed in the original psychological model, and its addition constitutes a reorientation of the theory to the more specific conditions where it holds.

The second step in merging the two theories is to make human stereotypes reflect perfect information. Bianco proposes that, for a special case of his formalization, “initial impressions may be more accurate than those formed by … strategic actions” (1075). By the next page, this speculation on a special case has become a general result: “impression formation by the continuum model ... is consistent with the efficient use of information.” While there are many conditions under which initial impressions may be more accurate, Bianco provides only one mechanism to make this event likely enough to justify the more general, summary, claim (1076). The mechanism adds to the psychological account an expectation in the voter of strategic behavior by the candidate. This new machinery makes the key parts of Figure One’s parameter space much more likely to be accurate reflections of the voter’s decision process. However, this addition of game-theoretic rationality to heuristic categorization is made by assumption, not by reason. Without strategic behavior, perfect initial impressions are too special a case to support any more general claim.

Though it is presented as a general recasting of two unmodified theories, Bianco’s “Different Paths” actually presents a new model of voter behavior that, given a few additional assumptions, can be seen equally clearly as a special case of either of them. While it is entirely likely that real-world conditions exist where this combination of the psychological and game-theoretic models will make the best predictions of a real voter's behavior, restricting focus to only the conditions where they are indistinguishable has serious consequences: “Insofar as real-world voters are motivated tacticians, they will reason to the same impressions as those formed by a rational actor behaving according to Bayes’ Rule.” (1076)

This conclusion is a double-edged sword. On one hand, it proposes heuristic techniques as optimal. On the other hand, it proposes much more general conditions for the ecological validity of the “infinitely-rational” assumption. Beyond domains where the necessary assumptions hold, one must forfeit the luxury of the former so as not to concede the latter.