Decision rules and correlated features

Most research on decision making attends more to cognitive rules than to the situations in which these rules are employed. One characteristic of these situations is the correlation among the features or cues they reveal. We simulated thousands of decision situations that varied in 1) the number of alternatives, 2) the number of features, and 3) the correlations among features. Six, simple decision rules were then used to choose an alternative and their choices were compared to those generated by a mathematically optimal rule. Results show that all rules, including some that do very poorly when features are uncorrelated, greatly increase their chances of making optimal decisions as feature correlations rise. We discuss some implications of these results for the use of simple decision rules in the real world.