DEBIASING BY INSTRUCTION - THE CASE OF BELIEF BIAS

The study is concerned with the question of whether robust biases in r easoning can be reduced or eliminated by verbal instruction in princip les of reasoning. Three experiments are reported in which the effect o f instructions upon the belief bias effect in syllogistic reasoning is investigated. Belief bias is most clearly marked by a tendency for su bjects to accept invalid conclusions which are a priori believable. Ex periment 1 attempted to replicate and extend an experiment reported by Newstead, Pollard, Evans and Allen (1992). In contrast with their exp eriment, it was found that belief bias was maintained despite the use of augmented instructions which emphasised the principle of logical ne cessity. Experiment 2 provided an exact replication of the augmented i nstructions condition of Newstead et al., including the presence of pr oblems with belief-neutral conclusions. Once again, significant effect s of conclusion believability were found. A third experiment examined the use of elaborated instructions which lacked specific reference to the notion of logical necessity. The use of these instructions signifi cantly reduced the effects of belief on the reasoning observed. Taking the current findings together with the experiment of Newstead et al., the overall conclusion is that elaborated instructions can reduce the belief bias effect in syllogistic reasoning, but not eliminate it. Th is conclusion is discussed with reference to (1) the practical implica tions for improving thinking and reasoning via verbal instruction and (2) the nature of the belief bias phenomenon.