Does short-term memory develop?

Abstract

Such is the consistency by which performance on measures of short-term memory (STM) increase with age that developmental increases in STM capacity are largely accepted as fact.However, our analysis of a robust but almost ignored findingthat span for digit sequences (the traditional measure of STM) increases at a far greater rate than span for other verbal materialfundamentally undermines the assumption that increased performance in STM tasks is underpinned by developmental increases in capacity. We show that this digit superiority with age effect is explained by the relatively greater linguistic exposure to random sequences of digits versus other stimuli such as words. A simple associative learning process that learns incrementally from exposure to language accounts for the effect, without any need to invoke an STM mechanism, much less one that increases in capacity with age. By extension, using corpus data directed at 2-3 year old children, 4-6 year old children, and adults, we show that age-related performance increases with other types of verbal material are equally driven by the same basic associative learning process operating on the expanding exposure to language experienced by the child. Our results question the idea that tests such as digit span are measuring a dedicated system for the temporary maintenance and manipulation of verbal material, and as such have implications for our understanding of those aspects of typical and atypical development that are usually accounted for with respect to the operation of such a system.

Verbal Working Memory as Emergent from Language Comprehension and Production

“…Thus, meaning does not seem to be necessary for the effect. Jones and Farrell (2018) further demonstrated that people are more likely to recall sentence-like lists in an order consistent with syntactic knowledge and that errors are more likely to conform to prior syntactic knowledge than expected by chance (for corpus analyses tying language experience to memory performance, see Perham et al, 2009;Jones et al, 2020). In each case, inter-item information affected memory for order via long-term knowledge of language syntax, suggesting that memory for items and their order interact to support each other.…”

“…Whereas some researchers have considered poor VWM performance as a cause of poor language skills, potentially ameliorated by working memory training (e.g., Ingvalson et al, 2015), our language-emergent VWM view suggests that poor VWM performance is a symptom associated with poor language skill. In other words, the abilities to encode, maintain, and order verbal information are skills that emerge from language use, and individuals who have higher language skills have richer LTM representations and more practiced comprehension and production processes (see also Jones et al, 2020). Thus, we can view tasks that are described as VWM tasks not as assessments of a separate VWM capacity but rather as measures of a person's skill in encoding and maintaining verbal information.…”

Under-resourced or overloaded? Rethinking working memory deficits in developmental language disorder

“…The challenge for mechanistic accounts arguing that apparent working memory capacity limitations are the consequence of shortfalls in long-term language knowledge is, of course, to explain how and why speech encoding is deficient without appealing to a primary working memory capacity bottleneck. Along these lines, computational modelling of variance in non-word repetition and span task performance among typically developing individuals has appealed to the notions of input frequency and regularity (Jones, 2016;Jones et al, 2007Jones et al, , 2008Jones et al, , 2020MacDonald & Christiansen, 2002;Jones et al, 2020;MacDonald & Christiansen, 2002). Here, the idea is that the ability of an artificial neural network to accurately process any given speech sequence relates directly to the quality of the network's established, analogous representations, which is higher when the relevant input previously received is frequent and structurally consistent.…”

“…This indicates that a separate buffer system which hypothetically varies in capacity between individuals (e.g., a phonological loop) is not required to explain variance in task performance; variance in the frequency of stimulus exposure and therefore the quality of long-term encodings (i.e., more frequently encountered, regular stimuli are better encoded) can parsimoniously account for the data at hand. The long-term encoding benefits of high frequency and regularity of exposure clearly boost performance for certain stimuli in working memory tasks, and may more broadly explain why working memory capacity appears to increase during infancy and childhood (Jones et al, 2020). Simply, as implicit in the state-based framework of working memory, task performance may improve as children become increasingly adept at deploying their mounting long-term language knowledge in the moment, not, as is commonly argued, because of developmental capacity increases that are independent of the quality of long-term representations (Gathercole et al, 2004).…”

Investigating the relationship between syntactic and short-term/working memory impairments in children with developmental disorders is not a straightforward endeavour

“…They review evidence for non-word repetition performance being linked to knowledge of phonological patterns and vocabulary, and they argue that non-word repetition tests measure the quantity and quality of language skill and experience relevant to the specific demands of this particular task, rather than a separate memory capacity. Similarly to G. Jones et al’s (2020) argument for span tasks, Schwering and MacDonald (2020) argue that what non-word repetition tests measure is language ability.…”