Does semantic similarity affect immediate memory for order? Usually not, but sometimes it does

Abstract

Recall performance in working memory (WM) is strongly affected by the similarity between items. When asked to encode and recall list of items in their serial order, people confuse more often the position of similar compared to dissimilar items. Models of WM explain this deleterious effect of similarity through a problem of discriminability between WM representations. In contrast, when lists of items that are all semantically similar are compared to lists of dissimilar items, semantic similarity does not negatively impact order memory, questioning the idea that semantic information is part of the WM content. This study reports four experiments in which semantic similarity was manipulated using lists composed of multiple semantic categories. These experiments revealed two main patterns. First, semantic similarity can increase, rather than decrease, order memory. Second, semantic knowledge reliably constrains the way items migrate; when migrating, items tend to do so more often toward the position of other similar items, rather than migrating toward other dissimilar items. These results challenge the way current models of WM represent similarity. The plausibility of different theoretical accounts and mechanisms is discussed.

Semantic Representations in Working Memory: A Computational Model

“…Subsequent experiments have replicated these results in the phonological and visuospatial domains (Kowialiewski, Lemaire, et al, 2022). A recent work of our own (Kowialiewski, Majerus, et al, 2023) showed that the effect replicates regardless of presentation modality (i.e., auditory, written), and test method (i.e., serial recall, order reconstruction). Furthermore, the increased of within-category transpositions is not due to participants developing long-term memory knowledge or expectations about the semantic list structure during the experimental setup, as the effect persists when participants cannot predict the lists' semantic structures, suggesting a non-strategic origin.…”

“…The effect was only half as large in the interleaved condition. Results from a recent replication (Kowialiewski, Majerus, et al, 2023, Experiment 1) are illustrated in Figure 5. As can be seen, the separation effect has a beneficial effect on item recall (left panel).…”

“…In the separation effect, semantically similar items are better recalled if encoded at close vs. distant serial positions. We fitted the model to the data reported in Kowialiewski, Majerus, et al (2023), Experiment 1. The experiment involved three conditions: grouped, interleaved, dissimilar.…”

“…Sometimes, a detrimental effect was found (Saint-Aubin et al, 2005;Tse, 2010;Tse et al, 2011), and some have argued that these discrepancies might be explained by the metric used to manipulate semantic similarity (Ishiguro & Saito, 2020). Recent studies have showed multiple times that semantic similarity has no negative impact on order recall (Kowialiewski, Krasnoff, et al, 2023;Kowialiewski, Majerus, et al, 2023), even when different similarity metrics are used (Neath et al, 2022), including the metric recently proposed by Ishiguro and Saito (Ishiguro & Saito, 2024;Kowialiewski, Krasnoff, et al, 2023). As similarity effects have been previously taken as evidence that WM relies on a particular kind of information, the null effect of semantic similarity on order recall could be taken to imply that WM does not encode semantics.…”

“…How can this contradiction be resolved? Here we provide a solution based on the idea that WM encodes only the semantic features shared by the items (Kowialiewski, Majerus, et al, 2023). By encoding only the features shared between memoranda, the model not only predicts an item recall advantage for semantically similar items, but also an absence of a detrimental effect on order recall, because the encoded features are common to all items.…”