A negative semantic similarity effect on short-term order memory: Evidence from recency judgements

Abstract

Speeded recency judgements were used to examine whether judgements concerning the relative temporal order in which two items were presented on a short eight-item, study list would be facilitated or hindered when the two items were from the same semantic category (related, R, items), relative to when they were from different categories (unrelated, U, items). Semantic similarity led to slower and less accurate recency judgements. To evaluate the extent that item memory contributes to recency judgement performance, I used speeded pair recognition and speeded two-alternative forced-choice (2AFC) recognition trials, in which the stimuli and designs were the same as in the recency judgement test. The type of test trial was manipulated between subjects (recency judgement vs pair recognition) or within participants (recency judgement and 2AFC recognition). Regardless of the between- or within-participants manipulation of recognition vs order memory tests, the pattern of recognition performance was very different from the pattern of recency judgement performance, indicating that semantic similarity had a negative effect on order memory per se. The explanation of the current findings offered by position-based order memory theories is discussed.

Keywords:

• Order memory
• Recency judgement
• Semantic similarity

Acknowledgements

Part of this research was supported by a Dissertation Research Fellowship Award granted by the University at Albany, State University of New York and was conducted in partial fulfillment of the requirements for a PhD degree at the University at Albany, State University of New York, August 2007. I thank the chair, Jim Neely, and the other two dissertation committee members, Jeanette Altarriba and Tram Neill for their comments on my dissertation and Ian Neath, Jim Neely, and Jean Saint-Aubin for their comments on an earlier version of this manuscript and Nicole Becker, Katelyn DiLorenzo, Maria Emi, Yuki Fong, Amanda Long, and Mary Unger for their help with data collection.

Notes

¹The study modality×semantic similarity effect interaction and the open/closed set×semantic similarity effect interaction may be worth pursuing further. However, due to the potential methodological concerns associated with serial recall and order reconstruction tests (to be discussed later), this interaction should be interpreted with caution and be replicated in a single experiment using the recency judgement test, which as discussed later, shows a more robust semantic similarity effect in order memory than do serial recall and order reconstruction tests.

²The difference in overall performance in Experiment 2 and Experiment 1 was probably due to the sampling difference: whereas participants in Experiment 1 were recruited from a large university with a more diverse student population, those in Experiment 2 were recruited from a, highly selective university.

³There is an interesting contrast between the negative semantic similarity effect in short-term order memory and the positive semantic similarity effect in long-term order memory, as reported in studies that supported the reminded-conjoint-rehearsal theories. When longer (e.g., 60-item) study lists are used, semantic similarity facilitates, rather than hinders, recency judgement performance. These theories proposed that when participants encode items in the study list, they continuously rehearse them and this rehearsal automatically codes the temporal order information for the items in the study list, such that the ordered links can be established between the earlier and later presented items. While the relative order of two R items and two U items can be encoded by this cumulative rehearsal, the relative order of two R items could also be encoded by study-phase retrieval. That is, the encoding of an item presented later in the study list can remind participants of the earlier item and this would be much more likely when the two items are semantically similar (e.g., under the same category) than when they are not. Hence, this additional “reminding” mechanism for R items facilitates the recency judgement of R pairs and thus yields the positive semantic similarity effect in long-term order memory. Relevant to the current study, the reminded-conjoint-rehearsal theories assume that the study-phase retrieval occurs whether the two items are presented back-to-back or farther apart as long as the later item successfully triggers the retrieval of the earlier one (e.g., Winograd & Soloway, 1985). Thus, these theories may still predict R pairs would show better recency judgement performance than U pairs, inconsistent with the negative semantic similarity effect in recency judgement I obtained for eight-item study lists. Given a clear contrast between interfering and facilitatory effects of semantic similarity in recency judgements following short and long study lists, respectively, future studies should test the interactive effect of study-list length and semantic similarity more systematically.