A recent study, using a perceptual task, indicated that two heads were better than one provided that the members could communicate freely, presumably sharing their confidence in their judgments. Capitalizing on recent work on subjective confidence, I replicated this effect in the absence of any dyadic interaction by selecting on each trial the decision of the more confident member of a virtual dyad. However, because subjective confidence monitors the consensuality rather than the accuracy of a decision, when most participants were in error, reliance on the more confident member yielded worse decisions than those of the better individual. Assuming that for each issue group decisions are dominated by the more confident member, these results help specify when groups will be more or less accurate than individuals.:
Tuesday, May 08, 2012
Further work on when two heads are better or worse than one.
Studies that compare the accuracy of individual and group decision yield somewhat inconsistent results. The key to benefiting from other minds is to know when to rely on the group and when to walk alone. To follow up a thread started in two previous posts (here and here) on when two heads are better or worse than one, I pass on this work by Koriat. He shows that in an inference task involving two alternatives, optimal results are obtained with the simple heuristic of selecting the response expressed with the higher—or in the case of more than two heads, highest—degree of confidence. Here is the abstract: