Estimating d' from extreme false-alarm or hit proportions (p = 0 orp = 1) requires the use of a correction, because the z score of such proportions takes on infinite values. Two commonly used corrections are compared by using Monte-Carlo simulations. The first is the l/(2N) rule for which an extreme proportion is corrected by this factor before d' is calculated. The second is the log-linear rule for which each cell frequency in the contingency table is increased by 0.5 irrespective ofthe contents of each cell. Results showed that the log-linear rule resulted in less biased estimates of d' that always underestimated population d'. The 1/(2N) rule, apart from being more biased, could either over-or underestimate population d'.

No abstract

In a recognition memory test, subjects may be asked to decide whether a test item is old or new (item recognition) or to decide among alternative sources from which it might have been drawn for study (source recognition). Confidence-rating-based receiver operating characteristic (ROC) curves for these superficially similar tasks are quite different, leading to the inference of correspondingly different decision processes. A complete account of source and item recognition will require a single model that can be fit to the entire data set. We postulated a detection-theoretic decision space whose dimensions, in the style of Banks (2000), are item strength and the relative strengths of the two sources. A model that assumes decision boundaries of constant likelihood ratios, source guessing for unrecognized items, and nonoptimal allocation of attention can account for data from three canonical data sets without assuming any processes specifically devoted to recollection. Observed and predicted ROCs for one of these data sets are given in the article, and ROCs for the other two may be downloaded from the Psychonomic Society's Archive of Norms, Stimuli, and Data, www.psychonomic.org/archive.

No abstract

In recognition memory experiments, the tendency to identify a test item as "old" or "new" can be increased or decreased by instructions given at test. The effect of such response bias on remember-know judgments is to change "remember" as well as "old" responses. Existing models of the remember-know paradigm (based on dual-process and signal detection theories) interpret this effect as a shift i nresponse criteria, but differ on the nature ofthe dimension along which t he changes take place. W e extendedthe models to account simultaneously for remember-know and confidence rating data and tested them using old-new (Experiment 1) and remember-know (Experiment 2) rating designs. Quantitative fits show that the signal detection models provide the best overall description of the data.