Objective Difficulty and Input History as Factors in Sequential Decision Making

Abstract

Amount of displayed input history and level of problem difficulty were varied in a four-alternative, sequential decision task. Problem difficulty, defined by the number of inputs necessary to achieve one standard deviation separation between input sequences with different growth rates, was varied by decreasing growth-rate differences and by increasing input variability. The results indicated that information taking decreases (without increasing wrong decisions) with increasing input history. It was found that the relative amount of information taken decreases and the number of wrong decisions increases with increasing problem difficulty. Further, information taking was found to increase more with decreasing input-sequence growth-rate differences than with increasing input variability to statistically equivalent levels. When scored in the same way as terminal decisions and as compared to them, riskless estimates showed poorer performance. Finally, the results indicated that information taking increases on trials following wrong decisions.