The ‘true probability’ problem

As scientists and as technologists we should discard the idea of a ‘true’ or ‘objective’ probability. Instead, we should think of probability judgements as the result of an individual's feelings of uncertainty, translated into a numercial response by internal decision processes. Many factors, both internal and external to the assessor, may influence the feelings of uncertainty, or the decision processes, or both. From this point of view, a probability cannot be wrong; it can, however, be more or less related to stimulus or task characteristics, and it can to varying degrees be affected by memory and cognitive processes, prior experience and information, social and cultural norms, personality, and cognitive styles. In some applied settings it may be desirable to train assessors to base their judgements only on certain factors, and to the extent that we agree on these factors it is possible to assess the relative ‘goodness’ of assessors. But effective training can be designed only when we know how these factors influence the naive person's judgements. This should be one focus for future research.