CAUSAL INFERENCE AS A COGNITIVE STRATEGY

This paper reports on ongoing research about how humans process probabilistic or frequentistic information. It suggests that humans depart from the probability calculus not only when they engage in the kind of reasoning that cognitive psychologists like Daniel Kahneman and Amos Tversky label as 'erroneous', but also when they choose to give weight to causal rather than to frequentistic information and thereby in effect to ignore the odds. Finally, it argues that these results complicate the notion of rationality that should be employed by AI researchers.