A COMPARATIVE STUDY OF THE INTERACTIVE EFFECTS OF DIGITIZED AND PHOTOGRAPHIC IMAGE MODES AND COLOR REALISM IN A PICTORIAL RECOGNITION MEMORY TASK (SIGNAL DETECTION, RESEARCH)

The present study was designed to investigate the interactive and relative effects of color mode (B&W and color) and imagery mode (digitized and photographic) on pictorial recognition memory (PRM) as modeled and measured by Signal Detection Theory (SDT) and Receiver Operation Characteristic (ROC) indices: d', A', A(,G), Delta(M), A(z), d'(,e), and d(,a). A non-additive two-way analysis of variance model was assumed and tested by using SDT and ROC indices as dependent variables measuring PRM.;The original pictorial materials were 140 photographic realistic color 35 mm slides. These slides were then manipulated photographically to produce 24 x 24 mm('2) slides in both realistic color (PRC) and black & white (PWB). The same materials were processed by the digital image processing system at the Pattern Recognition Laboratory (PRL) at the University of Pittsburgh producing comparable (24 x 24 mm('2) slides) digitized black & white (DBW) and digitized pseudocolor (DPC) treatments. A random division of these slides resulted in 66 slides to the old set (signal + noise in SDT) and 74 to the new set (noice only in SDT). PRM test was constructed from a random mixture of both these sets. A microcomputer (Super-PET) was interfaced to drive slide projectors in the PRM task at 500 ms in displaying the old set and at 10 seconds during the PRM test. A combined method of both old-new (yes-no) and confidence-rating was employed. Subjects made old-new responses and their confidence-rating in a scale of six levels (positively old, very-likely old, probably old, probably new, very-likely new, and positively new). The first three levels were summed as yes/old and the last three summed as no/new in yes-no SDT analysis. A computer program was developed to compute yes-no SDT indices (d', A' & A(,G)). The RSCORE program by Dorfman and Alf (1968, 1969) was modified and used to compute the Maximum Likelihood Estimation of Delta(M), A(Z), d'(,e), and d(,a) of ROC analysis from confidence rating data. . . . (Author's abstract esceeds stipulated maximum length. Discontinued here with permission of author.) UMI;Ninety-six graduate and undergraduate students at the University of Pittsburgh werw subjects in the study. Subjects were randomly assigned to four groups in 2 x 2 treatment-by-treatment factorial design of equal cells of 24 subjects each. None of the subjects were color-blind.