Similarity relations in recognition: A model to explain Tulving's data

Abstract Tulving tested memory for pictures by two-choice recognition tests. Recognition was higher but less confident when the target and distractor of a pair were highly similar than when they were dissimilar but the distractor was highly similar to an untested target. Tulving was able to partially explain these results by distinguishing two types of similarity, physical and ecphoric, and by postulating interactive processing of target and distractor. A model is described which explains the above results without assuming interactive processing. It also accounts for the results he obtained with other types of test pairs.