Judgement and Decision Making in Young Children: Probability, Expected Value, Belief Updating, Heuristics and Biases

ion in weighting tasks may allow for earlier application of the multiplicative rule. Schlottmann and Christoferou (in preparation) tested this possibility by giving fourto tenyear-olds a conjunctive area task and two weighting tasks, an EV task and a novel sourcemessage task. As predicted, the younger children multiplied in both weighting tasks, but added in the area task, supporting the view that cognitive complexity may differ for different kinds of multiplication concepts in formally isomorphic situations. Thus, the multiplicative nature of children’s EV judgements appears clearly established. A second point is that children’s probability understanding is remarkably abstract. Even the youngest children’s judgements co-vary with probability, not the physical quantity used to represent it, in contrast to the magnitude estimation hypothesis (Hoemann and Ross, 1982). Children clearly understand that probability is a relative, not an absolute quantity. They also interpret the physical probability cues selectively, adapting their meaning to the value structure of the situation. In Figure 4, for instance, judgements of the same physical game reverse, depending on which outcome carries the larger prize. Thus, children do not just associate physical targets and responses, but clearly know that the meaning of a physical cue can differ, depending on how desirable and probable the outcome is. Across-outcome EV: Subadditivity caused by a serial strategy? So far we have highlighted children’s remarkable intuitive competence. However, while it seems clear that the within-outcome integration is multiplicative, as normatively required, there is doubt about the additivity of the across-outcome integration, even though addition seems mathematically simpler than multiplication. Adults (Shanteau, 1975) and primary-school-aged children as well (Schlottmann, 2001) often show subadditivities in games with two probabilistic outcomes, as in the task of Figure 3, in which both blue and yellow outcomes could carry prizes. Normatively, overall EV is the sum of the component EVs, implying that the effect of one prize is independent of