Three paradigms are identified as crucial to interpretations of results in most studies of heuristics and biases in probabilistic thinking. The paradigms are criticised as being so limited and inadequate that generalisations from current research on heuristics and biases cannot be justified. In particular, the view of people as ‘intellectual cripples’, who exhibit severe and systematic biases in making judgements, is shown to be a value judgement on the part of the investigator. The implicit acceptance of the paradigms is shown to have created four problems in current research. An alternative perspective, the generation paradigm, emphasizes the role of problem structuring, in particular the subject's internal representation of the task, within which information is processed. The generation paradigm sees decision making and the forming of judgements as dynamic, generative processes, conducted interactively between people within a social and cultural context. From this perspective, research would attempt to discover the conditions under which people can do well, aiding rather than de-biasing procedures would be investigated, and expected utility theory would be seen as neither a normative nor descriptive model, but rather as a guide to thinking that polices consistency within a small-world representation of a problem.