This paper questions evidence for the thesis that causal reasoning of older children is more logical than that of younger ones, and describes two experiments which attempted to determine (1) whether there are true developmental differences in causal reasoning, and (2) what explanations for developmental differences can be supported. In the first experiment a problem that was completely unfamiliar to the subjects was used, so that subjects could not base their responses on previous statements of parents or teachers. Results indicated a clear developmental difference. In order to generate plausible interpretations of this finding, an information processing model was constructed to describe the steps involved in performing the experimental task. Various developmental differences were tested using an interactional strategy. Siegler and Liebert's regularity-without-contiguity task was analyzed and different versions presented to children of different ages. Competing explanations of the developmental difference were tested. Findings indicated that the difference lay only in the subjects' ability to overcome the influence of the blinking lights of the computer used in the experiment. The use of such an interactional design is recommended as a procedure for directly comparing the strength of factors which are hypothesized as central by different theories. (GO)