Abstract
This study examined whether naturally acquired expertise in object classification provides an advantage on an unfamiliar detection task. Ten radiologists with a range of clinical experience, and forty novice subjects performed detection of textures in noise on two consecutive days. The textures were bandlimited noise patterns of low, medium and high spatial frequency content shown in two levels of external Gaussian noise. Subjects performed a yes-no task in which signal contrast was varied using the method of constant stimuli, and signal present probability was 50%. A fixed set of five textures were used on both days in each spatial frequency condition. Spatial frequency and noise conditions were blocked, with a total of 1008 trials per session (168 trials × 3 spatial frequency × 2 external noise conditions). Sensitivity (d’), criterion location (c), and relative criterion location (c’), were calculated in each condition on both days. Contrary to previous work showing domain-specific effects of expertise, the radiologists showed superior performance to novices on both days in all conditions. Both groups improved from day 1 to day 2, with larger improvement at the medium and high spatial frequencies in low noise. Perceptual strategies differed between groups: experts were more liberal than novices at baseline, and showed a conservative shift with practice in low noise, whereas practice did not alter the novice group’s criterion in any condition. These group differences were less pronounced in the relative criterion. Overall, the results suggest that expertise can benefit perceptual judgements outside the trained domain, and that the expert advantage may include a strategic component that is modified through experience.
Acknowledgement: University Research Board (URB) grant, American University of Beirut