The role of stimulus type in list length effects in recognition memory

Abstract

While many studies have investigated the list length effect in recognition memory, few have done so with stimuli other than words. This article presents the results of four list length experiments that involved word pairs, faces, fractals, and photographs of scenes as the stimuli. A significant list length effect was identified when faces and fractals were the stimuli, but the effect was nonsignificant when the stimuli were word pairs or photographs of scenes. These findings suggest that the intrastimulus similarity is what dictates whether list length has a significant effect on recognition performance. As is the case with words, word pairs and photographs of scenes are not sufficiently similar to generate detectable item interference.

Sources of interference in item and associative recognition memory.

“…As we have mentioned previously, experiments with novel nonlinguistic stimulus classes uncovered small detrimental effects of list length and list strength, which are consistent with item noise models but inconsistent with context noise models (Kinnell & Dennis, 2012;. However, as noted by , the detrimental effects of list length and list strength were quite small in magnitude compared with what would be expected from a pure item noise model, and they suggested that both item noise and background noise are relevant to understanding recognition memory performance for nonlinguistic stimuli.…”

“…Kinnell and Dennis (2012) posited that the differences associated with the differ ent stimulus classes may be attributed to different levels of item noise, with word stimuli and natural scenes being exempt from item noise at retrieval whereas faces and fractals suffer from item noise, possibly because of having more distributed item representations (we return to the issue of how different stimulus classes can suffer from different degrees of item noise in the General Discussion). Included in the model fit are five experiments using the list length paradigm: the dataset of Dennis et al (2008) along with the four experiments by Kinnell and Dennis (2012). All of the exper iments use two list length conditions that all use a 1:4 list length ratio from the short list to the long list.…”

“…More recently, Turner, Dennis, and Van Zandt (2013) compared the REM (the original pure item noise version) and BCDMEM models in their ability to account for data from the list length paradigm (specifically the datasets of Dennis et al, 2008 andDennis, 2012) using hierarchical Bayesian methods. BCDMEM consistently exhibited lower values of the deviance information criterion (DIC), a Bayesian model selection measure that measures goodness of fit relative to the degree of model complexity.…”

“…Kinnell and Dennis (2012) tested novel nonlinguistic stimuli, specif ically images of fractals, faces, and natural scenes, in single item recognition using the same controls used in the other investigations. Significant effects of list length were found for fractals and faces but no effect of list length was found for natural scenes.…”

“…Following the investigation of Kinnell and Dennis (2012); tested images of fractals, faces, and natural scenes for the presence of a list strength effect while simultaneously using the list length controls advocated by Dennis and colleagues.6 A significant list strength effect was found for fractals with both yes/no and 2AFC testing whereas null effects of list strength were found for faces and scenes (although a significant list strength effect was found using artificial faces by Norman et al, 2008). used this finding to support the hypothesis of Kinnell and Dennis (2012) that novel nonlinguistic stimuli may be more susceptible to the effects of item noise.…”