Inverted list-strength effects in recognition.

Abstract

If some list items are studied strongly and others weakly, many memory models predict the effect of strength on memory will be larger when strengths are mixed within a list than between pure lists of a single strength: a list-strength effect. In explaining why list-strength effects were elusive in old/new recognition, Shiffrin et al. (1990) introduced differentiation. This gave the model a way to produce an inverted list-strength effect, which they thought was usually offset by the coexisting expected "upright" list-strength effect. Alternatively, attentional subsetting theory (Caplan, 2023; Caplan & Guitard, 2024b) predicted inverted list-strength effects in some circumstances by considering how the dimensionalities of attended feature spaces might differ for strong and weak items. Inversions were indeed found in manipulations of stimulus duration (e.g., 500 ms vs. 2,000 ms study time/word). Here we replicated the pattern when display time was equated (Experiment 1) and with massed repetition (Experiment 2), ruling out the relevance of vision-locked features and the number of stimulus onsets. Both theoretical accounts of inverted list-strength effects, however, miss the fine structure of the data, namely, reduced hit rates for weak items in pure than mixed lists and the reverse effect (albeit less robust) for strong items. Model fits suggested the critical factor is that list composition parametrically influences the number of deep features processed at test combined with participants' response bias adapting to list composition. In sum, inverted list-strength effects are robustly found in manipulations of item study time and point to differential processing of probe features depending on list composition, compatible with most models. (PsycInfo Database Record (c) 2025 APA, all rights reserved).