In recent years, one of the tasks, which frequently has been used to study cognitive functions is recognition memory. In this task, visual or auditory stimuli is presented to subjects at an interval of time. Then, exposed stimuli mix with similar stimuli, and subjects are asked to recognize the "old stimuli". Traditionally, statistical analysis of these tasks are based on the number of correct responses between the two or more control and experimental groups.
Recent research has shown that using the number of correct responses, called as response sensitivity (d'), could not be the only criterion to evaluate the data of such a research and could lead to wrong results. Thus, it is proposed that, in this case, the researcher should use signal detection theory (SDT) before using usual statistical methods. Based on this theory, in addition to response sensitivity, there is another important factor, called "response bias" ß, which is the subject's tendency to say "yes" or "no" whatever the task is. Accordingly, a high d' value indicates high sensitivity of subjects to discriminate between old stimuli from new ones and a high ß value would correspond to a conservative criterion for accepting recognition of old words.
In sum, the usefulness of this method has been confirmed by many researchers (e. g. Holly et al, 1994). In this paper, after presenting a definition of recognition memory, the scientific measurement of the task is discussed. Then, the main elements of the SDT which are False Alarm, Correct Rejection, Hit and Miss are introduced. Also, the calculation of d' and ß are explained.