DECISION-MAKING IN A DYNAMIC ECOLOGICAL-S YSTEM

The formal model of Dynamic programming (Mangel & Clark, 1988) describ es optimal behavior in behavioral ecology. We apply this model to huma n decision making. The relevant variables (payoff, probability of payo ff, current state of the system, time horizon) are incorporated in an experimental gamble and the influence of important characteristics (e. g., expected value, number of options, probability, payoff) on decisio n making is tested. For the different, but related, dependent variable s Proportion of optimal choices, Deviation from optimal fitness, and A mount won, we get surprising results. The complexity of decision situa tions as manipulated by the number of options, the difference in expec ted values of options, and the difference in probabilities, was negati vely related to overall fitness. That is, the less complex a situation is, the bigger are the deviations from optimal behavior. But, somewha t perplexing, the mean amount won raised with complexity. Correlations between final capital and optimal behavior as required by the model w ere negligible. These results were interpreted by reference to the res pective goal that a decision maker may strive to attain. Different goa ls may be used in gambling situations, which may appear suboptimal wit h respect to the model of dynamic programming.