Portal de Periódicos da CAPES

Using recognition in multi-attribute decision environments

2013; Wiley; Volume: 35; Issue: 35 Linguagem: Inglês

1551-6709

Don van Ravenzwaaij, Ben R. Newell, Chris P. Moore, Michael Lee,

Cognitive Science and Mapping

Using Recognition in Multi–Attribute Decision Environments Don van Ravenzwaaij (d.vanravenzwaaij@unsw.edu.au) , Ben R. Newell (ben.newell@unsw.edu.au) , and Chris P. Moore (christophermoore@gmail.com) School of Psychology, University of New South Wales, Sydney 2052, Australia Michael D. Lee (mdlee@uci.edu) Department of Cognitive Sciences, University of California Irvine, Irvine, CA, 92697–5100, USA Abstract An experiment examined the effect of ‘pure’ recognition — in the absence of concomitant evaluation — on inferences. In the first stage of the experiment, participants indicated whether they recognized a number of Italian and US cities. In the sec- ond stage, they decided which of two cities had the larger population. Crucially, names of the cities were not available in the second stage, but participants could find out whether they had recognized them (yes/no) in the first stage of the ex- periment (i.e., pure recognition). Additional predictive cues (e.g., presence/absence of a university) were also available. Participants used the recognition cue about 50% of the time, rarely examined it first, and used it differently as a function of whether recognition information was binary or continuous. Furthermore, participants used the recognition cue more often if they recognized more items, irrespective of its predictive va- lidity. Implications for theoretical frameworks that view infer- ence as driven by discrete heuristics or processes of evidence– accumulation are briefly discussed. Keywords: Inference, heuristics, recognition, decision mak- ing. Humans are decision makers. Throughout our lifes, we are constantly confronted with situations that force us to make a choice. Whether it is a preference decision, “Do I take the car or do I walk to work?”, or a knowledge decision, “Which soc- cer team scored more goals last season, Borussia Dortmund or Bayern M¨unchen?”, we are evaluating alternatives. Gigerenzer and colleagues have proposed a number of rel- atively simple heuristics that could help us making such de- cisions. In this paper, we will focus on one of the most prominent examples: the recognition heuristic (Goldstein & Gigerenzer, 1999, 2002; Gigerenzer & Goldstein, 2011). In the original conceptualisation of the recognition heuris- tic, called take–the–best (TTB; Gigerenzer & Goldstein, 1996), the first step in deciding which of two response op- tions to choose was to use recognition. So, if a decision maker knows the Bayern M¨unchen soccer team, but has never heard of Borussia Dortmund, then respond that Bayern M¨unchen scored more goals last season. When both teams are rec- ognized (thus disabling the use of recognition) the heuristic consults relevant information, or cues, in memory that are in- dicative of the number of goals scored (e.g., “What was the team’s final standing in the national competition?”). These cues should be consulted in descending order of informative- ness, starting with the cue that will be most indicative of the criterion of interest (i.e., number of goals scored). Cue search stops when the decision maker examines a cue that points in one direction (i.e., Borussia Dortmund was first last sea- son, Bayern M¨unchen was second, so respond Borussia Dort- mund). This proposal for a simple mechanism based on recogni- tion sparked a wide ranging debate about the plausibility, em- pirical validity, and generality of the recognition heuristic (for recent examples see the papers in the three special issues of the Journal of Judgment and Decision Making — Vol 6 (1) & (5), 2011; Vol 5 (4), 2010). Much of the debate revolves around some key assumptions about the nature and operation of recognition in inferential judgment. In the paper that introduced the recognition heuristic as a stand–alone ‘tool’ (i.e. not just the first step in Take–the– Best), Goldstein and Gigerenzer (2002) assume, firstly, that recognition is binary. That is, we either recognize something, or we do not, and there is no room within the heuristic for the distinction between something being vaguely familiar and something being very familiar. Secondly, recognition is as- sumed to be noncompensatory. That is, when we recognize one option, but do not recognize the other, then we should always go with the recognized option, regardless of any addi- tional information. Lastly, Goldstein and Gigerenzer (2002) make a distinction between familiarity and recognition: “The term familiarity is typically used in the literature to denote the degree of knowledge (or amount of experience) a person has of a task or object. The recognition heuristic, in con- trast, treats recognition as a binary, all–or–none distinction; further knowledge is irrelevant.” (pp. 77). Thus, according to a strict interpretation of the (2002 version of the) recognition heuristic, when deciding whether an Italian city you know has a larger population than an Italian city you do not know, it makes no difference whether the city you do know is Rome or Pisa. All three of these assumptions have been roundly chal- lenged in the literature on both empirical (e.g., Pohl, 2006; Newell & Shanks, 2004; Newell & Fernandez, 2006) and theoretical grounds (e.g., Hilbig, 2010; Newell, 2011). Re- sponding to some of these critiques, Gigerenzer and Gold- stein (2011) recast the adaptive use of the recognition heuris- tic as involving a two–step process: first recognition (“Do I