ABSTRACT
This chapter discusses the nature of recognition tests in contrast to recall tests. It shows that a recognition test typically yields superior performance for three reasons. First, guessing probability is higher in a recognition test. Second, correct recognition may be carried out simply because incorrect alternatives can be easily rejected. And finally, recall involves a memory search, whereas recognition probably does not require such a search. A very old and stable finding is that, given some information stored in memory, a recognition test on the information leads to better performance than a recall test. Recognition may be viewed as a relatively “pure” measure of whether some information has been stored in memory; it is a test that is relatively uncontaminated by variations in the retrieval process. More evidence suggests that recognition tests a different kind of memory than does recall. The chapter describes two of these models: the “all-or-none” and the “signal-detection” models.
