Abstract
Purpose. In a recent study Magnusson, Greenlee, Aslaksen and Kildebo (2003) reported high-fidelity long-term memory in a spatial frequency discrimination task with retention intervals up to 24 hr. Each trial was conducted with a different observer to avoid the influence of criterion-setting processes (Lages & Treisman, 1998). In two studies we tried to establish whether observers do employ a visual long-term memory for the discrimination of spatial frequency. Methods. The first study was conducted under similar conditions as reported in Magnussen et al. (2003). Images subtended 12 visual angle on a calibrated CRT monitor at a viewing distance of 57 cm. Stimuli were vertical sine-wave gratings in a Gaussian envelope displayed at 30% contrast for 5 sec. Each observer from a total sample of N=60 participated in a single trial only. Supported by a chin- and head-rest they viewed a reference grating of 3 c/deg. After a retention interval of 1–2 hr a test stimulus was displayed that differed in spatial frequency from the reference by ±10 or ±20%. Observers had to indicate whether the test stimulus appeared “thicker” or “thinner” than the reference. Data were pooled across observers to establish a psychometric function. In a second study we tested observers repeatedly in several trials with a retention interval of 24 hr. The observer viewed reference and test stimuli monocularly through a pinhole aperture to exclude peripheral size and depth cues. Results. We were not able to confirm Magnussen et al.'s results. Both studies indicate no high-fidelity long-term memory for spatial frequency. Discrimination performance was close to chance level for longer retention intervals. Conclusions. The results suggest that visual long-term memory for spatial frequency is weak if not absent. Peripheral size and depth cues in combination with the stimulus may account for previously reported long-term effects.
Royal Society of London and EPSRC, UK