The accuracy with which observers could judge whether two visual stimuli were the same or different was measured with the rating method of detection theory. For judgments of whether two pictures referred to natural or manufactured things, the shape of the obtained receiver operating characteristic (ROC) was consistent with the observers adopting an optimal decision strategy. A similar result was found for judgments of complex but meaningless visual patterns. For judgments of whether two colours that differed along a simple sensory dimension were the same or different, however, the resulting ROC was consistent with the observers adopting a suboptimal differencing strategy. The accuracy of the judgments did not depend on the visual field to which the stimuli were presented.

Three experiments were conducted to determine the effects of practice on the Adjusting-Paced Serial Addition Task (Adjusting-PSAT) (Tombaugh, 1999) and the Computerized Tests of Information Processing (CTIP) (Tombaugh & Rees, 2000). The Adjusting-PSAT is a computerized modification of the Paced Auditory Serial Addition Test (PASAT) (Gronwall, 1977) that makes the interval between digits contingent on the correctness of the response. This titration procedure permits a threshold value to be derived that represents the shortest presentation interval in which a person can process the digits to produce the correct sum. The CTIP consists of three reaction time tests that are progressively more difficult. Results showed that robust practice effects occurred with the Adjusting-PSAT, with the greatest increase in performance occurring on the first retest trial. Practice effects were equally prominent regardless of whether the first retest trial occurred 20A min, 1 week, or 3 months after the first administration. These gains were maintained for periods up to 6 months and were independent of modality of presentation (visual or auditory) and type of number list (easy or hard). In contrast to the findings with the Adjusting-PSAT, only minimal practice effects were observed with the CTIP. The major clinical implication of the study is that the high reliability coefficients for the CTIP, the lack of anxiety associated with its administration, and its insensitivity to variables such as numerical and verbal ability make the CTIP ideally suited for the serial evaluation of cognitive status. These characteristics also make the CTIP a viable alternative to the Adjusting-PSAT or PASAT for measuring speed of information processing. If the Adjusting-PSAT is administered repeatedly in clinical evaluations, a "dual baseline" or "run in" procedure should be used, with the second administration serving as the baseline measurement.

Background Drug administration errors occur in every aspect of clinical practice. Using a novel task-relevant Medication Recognition and Confirmation Test (MRCT), we investigated the effects on performance of working night and day shifts and labelling different drug formats. Methods Anesthesia trainees (n = 18) participated in one of two experiments during an 8-12 hr day shift and an 8-12 hr night shift. In Experiment-1 (n = 10), we compared standardized colour-coded labels with pictures of ampoules. In Experiment-2 (n = 8), we compared colourcoded labels with black and white labels. Sleep was measured with wrist actigraphy during both day and night shift runs over seven to eight days. The MRCT outcome measures were reaction times and drug errors.Results In the two experiments, colour-coded labels were recognized (and therefore selected) more quickly than pictures of conventional ampoules (mean difference 332 msec, 95% confidence interval (CI) 242-422 msec; P \ 0.0001) and faster than black and white labels (mean difference 96 msec, 95% CI 46-146 msec; P \ 0.0001). Participants obtained less sleep while working night shifts than while working day shifts (mean difference 57 min, 95% CI 0:15-1:39 hr; P = 0.013). Mean confirmation reaction times were slower during night shifts than during day shifts (mean difference 60 msec, 95% CI 1-120 msec; P = 0.048). No differences in error rates were observed between shifts or among drug label types. Conclusions Label format influenced recognition and confirmation reaction times to representations of drugs in this study, and we found some evidence to suggest that performance is better during day shifts than during night shifts. The task-relevant test evaluated here may have further application in measuring performance in the wider clinical setting. RésuméContexte Les erreurs d'administration me´dicamenteuse surviennent dans toutes sortes de situations de la pratique clinique. A`l'aide d'un nouveau Test de d'identification et de confirmation des me´dicaments (MRCT) pertinent a`la taˆche, nous avons examine´les effets de quarts de travail de nuit et de jour sur les performances, ainsi que les effets de diffe´rents formats d'e´tiquetage de me´dicaments. Méthode Des stagiaires en anesthe´sie (n = 18) ont pris part a`l'une de deux expe´riences pendant un quart de

A modified computer version of the PASAT (Adjusting-PSAT; ) is described that measures speed of information processing and working memory by means of a temporal threshold rather than number of correct responses. This is accomplished by making the duration of the interval between numbers depend on the correctness of responding-a correct response decreases the interval between digits and an incorrect response increases the interval. Modality of presentation (visual and auditory) was factorially combined with problem difficulty (answers between 2-10 or 2-18). Performance of 60 healthy student volunteers on the Adjusting-PSAT was compared to that obtained on several traditional neuropsychological measures (Digit Span, Trail Making Test, and Symbol Digit Modality Test) and on a test of basic addition skills. The visual version of the test produced a lower threshold than did the auditory version, but problem difficulty did not produce a significant effect. Of the neuropsychological tests, Trails-B (TMT-B) was most highly correlated with thresholds. However, regression analyses revealed that math ability accounted for more variance than did TMT-B. The clinical implications of these finding are discussed.

The accuracy with which observers judged whether two words belonged to the same semantic category was determined from a detection-theoretic analysis of same-different judgments. In Experiment 1, one word was presented centrally and the other word in either the left visual field (LVF) or the right visual field (RVF); in Experiment 2, both words were presented to either the LVF or the RVF. In order to obtain receiver-operating characteristics (ROCs) of performance, observers were asked to rate their confidence that the two words belonged to the same semantic category. Two models of the decision strategy were fitted to the obtained characteristics: a differencing model, in which the decision variable was the difference between the two observations; and an optimal model, in which each observation was judged in relation to a criterion. In both experiments, the optimal model provided a better fit than the differencing model to the obtained characteristics. Maximum-likelihood estimates of both the criterion-free parameter, d', and the area under the operating characteristic, peA), were greater for words presented in the RVFthan for those presented in the LVF.The same-different task has enjoyed wide usage in investigations of cognitive functions, including investigations of hemispheric asymmetries, perceptual matching, and the processes involved in stimulus comparisons. The predominant focus in cognitive psychology has been on reaction times rather than accuracy. Typically, errors in same-different studies are of secondary interest, since the conditions are usually such that few errors occur (Farell, 1985), and most investigators attempt to minimize them.In contrast to this approach, we report here on the accuracy of same-different judgments rather than on reaction time. We present findings concerning the accuracy with which people can judge whether two words belong to the same semantic category or not, and also concerning the question of whether the accuracy of their judgments depends on the visual field in which the words are displayed. By so doing, we hope to provide a different perspective on the processes involved in same-different judgments of this kind. Thus, we propose that an appropriate measure of accuracy can reveal the underlying decision processes in judgments of identity or difference, and that these processes are fundamental to understanding such judgments. One way in which this can be accomplished is by analyzing the same-different receiveroperating characteristic (ROC), which provides a We thank Michael Corballis, Carole Ernest, and Michael Hautus for helpful comments on an earlier draft of this paper, and Michael Corballis for advice on Experiment 2. We also thank the editor and three anonymous reviewers for their useful and challenging comments. Experiment I is included in part ofa doctoral dissertation by M .A.F. Requests for reprints should be sent to R.