Extensive Interference Attenuates Reinstatement in Human Predictive Judgments

An experiment assessed the impact of varying levels of interference on reinstatement in human causal learning. Participants studied fictitious customer files to learn relationships between foods and gastric illness in acquisition. During interference training, a new relationship was learned between the same foods and a different illness over 12, 15, or 18 trials. Prior to the test, presentations of either outcome in the absence of information about the food led to losses of the second-learned information and recovery of that learned first. This effect was reduced as the number of interference trials increased. Results are discussed in terms of their implications for theories of reinstatement and of the parallels with animal studies on renewal