1. Conversational Frames and Overcoming the Apparent Limits to Human Rationality John C. Thomas IBM T. J. Watson Research Laboratory PO Box 704 Yorktown Heights, New York 10598 email@example.comWe argue that many of the apparent deviations from optimal decision making andrationality exhibited in laboratory experiments may be due to differences in the apparentconversational frameworks between investigators and subjects. Examples of suchdeviations include neglect of base rates in probability judgments, failure to find optimalsolutions in “hidden profile” tasks, and the influence of path history on decisions. In thefirst paradigm, people are typically asked to make probability judgments after reading adescription. For example, a subject is told that someone walks into a room and that theyare over 6’2” tall, have a thick neck and enormous biceps. Now, the subject is asked tojudge the chances that this person is a professional (American) football player or anaccountant. People typically judge that this person is much more likely to be aprofessional football player and this judgment ignores the fact that there are many moreaccountants than professional football players. The second paradigm refers to a situationin which members of a small group are given partially overlapping information aboutsome decision task such as finding the best employee for a given job. The “optimal”answer requires the group to use and combine some of the information which is uniquelyshown only to individuals while what people typically do instead is focus on theinformation common to the group and thereby chose a non-optimal answer. In the thirdparadigm, people make a decision such as whether to buy or sell a stock. Optimally, thedecision depends only on the probable future performance, but people are actually muchinfluenced by the history. There are many other situations wherein people in experimentsbehave in a non-optimal fashion (Kahneman & Tversky, 1979).These deviations from optimal behavior might be explained by looking at differences inthe conversational frameworks held by the experimenters and the subjects in thesesituations. Various attempts have been made to provide “decision aids” for people thatwould influence them to move toward greater optimality. However, since none of theseaddress the underlying problem of mismatching conversational frames, it is not surprisingthat such attempts have typically had moderate or no success. Instead, we suggest aids tohelp surface the conversational frames to the parties involved. In the base rate neglectparadigm, for instance, the experimenters may see themselves as portraying someobjective information to subjects in a non-selective way. However, from a subject’spoint of view, this is an extremely odd conversational frame. Most commonly in telling astory, the teller selects a small number of details meant to convey or foreshadowsomething of importance to the story. We hypothesize that making these conversationalframes and their associated assumptions explicit may reduce many of the apparently non-optimal behaviors of human decision makers.Kahneman, D. and A. Tversky (1979): Prospect Theory: An Analysis of Decision UnderRisk, Econometrica 47 263-291.