CCNCS Seminar Details
Decision by Sampling [Joint with Reasoning Group]
|Speaker:||Prof Neil Stewart|
|Date/Time:||Wednesday 9 March 2011, 3.30pm|
Decision by sampling (DbS) is a model of economic decision making. Attributes, such as an amount of money, a chance of winning, or the capacity of an iPod, are valued in a series of binary, ordinal comparisons with a small sample of attribute values. The sample comprises attribute values from the immediate context in which a decision is made and from the distribution of attribute values in memory, which is assumed to represent the distribution of attribute values in the real world. For example, a target value of $12 might be compared against other amounts on offer, say $5 and $20, and recent values recalled from long-term memory, say $8 and $35. In this example, $12 has a subjective value of 1/2, because half of the comparisons make $12 look good (i.e., those to $5 and $8) and half make $12 look bad (i.e., those to $20 and $35). Thus, in DbS, the subjective value of an attribute is effectively its rank position within the sample. This talk reviews the evidence to date for each of the claims in the DbS model. I review how one can use the real-world distribution of attribute values to explain why utility functions, probability weighting functions, and delay discounting functions take the forms they do. I review evidence that moment-to-moment fluctuations in individual choice follow from variation in the flux of recently sampled attribute values. I review evidence that experimentally manipulating the distribution of attribute values in the immediate context can reverse people's preferences, showing a causal link between attribute distributions and preferences. I show how classic results from the decision under risk literature follow from DbS and compare this account to other models of decision under risk.