The conditions under which stimulus over-selectivity occurred were studied using a matching-to-sample procedure with non-autistic adults. A matching-to-sample discrimination learning task with a number of sample-comparison retention intervals was used. The results demonstrated that an increase in retention interval increased the degree of stimulus overselectivity displayed. In addition, it was shown that the matching-to-sample procedure is suitable for eliciting overselectivity in a non-autistic adult sample.