Structural priming is supported by different components of nondeclarative memory: Evidence from priming across the lifespan.

Abstract

Structural priming is the tendency to repeat syntactic structure across sentences and can be divided into short-term (prime to immediately following target) and long-term (across an experimental session) components. This study investigates how nondeclarative memory could support both the transient, short-term and the persistent, long-term structural priming effects commonly seen in the literature. We propose that these characteristics are supported by different subcomponents of nondeclarative memory: Perceptual and conceptual nondeclarative memory respectively. Previous studies have suggested that these subcomponents age differently, with only conceptual memory showing age-related decline. By investigating how different components of structural priming vary across the life span, we aim to elucidate how nondeclarative memory supports 2 seemingly different components of structural priming. In 167 participants ranging between 20 and 85 years old, we find no change in short-term priming magnitude and performance on perceptual tasks, whereas both long-term priming and conceptual memory vary with age. We suggest therefore that the 2 seemingly different components of structural priming are supported by different components of nondeclarative memory. These findings have important implications for theoretical accounts of structural priming.

Structural priming persists for (at least) one month in young adults, but not in healthy older adults

“…It could be that older adults are more vulnerable to being biased towards infrequent syntactic structures. A similar trend was found in the performance of older adults in a statistical learning task (serial reaction time task; Heyselaar et al, 2020), suggesting that this effect could be due to their declining implicit memory system and may not be language-specific. Patients with amnesia also show a more robust passive priming magnitude after a single session (Heyselaar et al, 2017).…”

“…For the analysis, we used a generalized additive mixed model (GAMM), using the mgcv package (version 1.8-22; Wood, 2017) as previous experiments have shown that cumulative priming mirrors a growth-curve more than a linear correlation over the length of the experimental session (Heyselaar et al, 2015(Heyselaar et al, , 2020Segaert et al, 2016). Unlike ANOVAs or generalized mixed-effects regression (GLMER), GAMM does not assume linearity (although it can find a linear form if supported by the data).…”

“…Recent studies have suggested that healthy, older adults may show a significantly less robust structural priming effect, within single sessions (Heyselaar et al, 2017), and have correlated this effect with a decline in implicit learning (Heyselaar et al, 2020). As the major theoretical models proposed to underlie the structural priming effect involve unconscious learning (Chang et al, 2012;Jaeger & Snider, 2013;Reitter et al, 2011), it was an open question whether healthy older adults would also show longterm priming.…”

“…1 As such, if components of structural priming are indeed supported by implicit memory, one would expect a decrease in structural priming ability with increasing age. A recent study indeed showed a significant decrease in cumulative priming in healthy, older adults, which correlated with their performance in standardised implicit learning tasks (Heyselaar et al, 2020). However, this study only looked at the cumulation of structural preferences within a single session (i.e., cumulative priming), and hence to what extent older participants will show a decline in the retention of these structural preferences across sessions (i.e., long-term priming) is still unknown.…”

“…In the present study we investigate: 1) Whether structural priming effects persist for one week and/or four weeks, 2) If the persistence effect is similar for young and older, healthy adults. Recent studies have shown a disconnect in the structural priming effects between these two age groups (Heyselaar et al, 2017(Heyselaar et al, , 2020, which we will further elaborate on below. Additionally, we will investigate whether 3)…”