The Nature and Impact of Working Memory

Working memory has been recognised as one of the most important cognitive constructs linked to effective teaching and learning to emerge from cognitive psychology (Alloway, Doherty-Sneddon, & Forbes, Reference Alloway, Doherty-Sneddon and Forbes2012; Dehn, Reference Dehn2011). However, it remains difficult to arrive at a single cohesive definition of working memory, with continued development and debate over its meaning and function. Many definitions of working memory characterise it as a cognitive system responsible for the immediate storage and manipulation of information (e.g., Alloway, Reference Alloway2011; Baddeley, Reference Baddeley2003). Gathercole describes working memory as ‘several interacting subsystems that include specialised stores for verbal and visuo-spatial material, and an attentional component that controls activity within working memory’ (2008, p. 382). Although working memory is often presented as a cognitive capacity that has attentional components, it can also be conceptualised as an observable set of skills or strategies used by students to ensure effective cognitive and academic functioning (Davis, Sheldon, & Colmar, Reference Davis, Sheldon and Colmar2014; Gathercole, Reference Gathercole2015). A working memory index has featured in both the Wechsler Intelligence Scale for Children — Fourth Edition (WISC–IV) and the new WISC–V, acknowledging the important role of working memory as part of cognitive functioning. Although there are continuing definitional issues and debates about the nature and measurement of working memory (Alloway & Alloway, Reference Alloway and Alloway2013; Baddeley, Reference Baddeley2003), these are beyond the scope of this article, which focuses on working memory interventions, particularly those that are appropriate for younger students.

The traditional view of working memory development argues that working memory capacity develops throughout childhood and early adulthood, before levelling off and eventually declining during adulthood, with evidence that this developmental trajectory is underpinned by naturally occurring neurological changes in brain structures and activation (Casey, Giedd, & Thomas, Reference Casey, Giedd and Thomas2000). Further, working memory and academic achievement are closely associated, and this association may be reciprocal and bidirectional, with time spent in schooling making a positive impact, such that increases in working memory have been observed within the duration of a school year, independently of age (Roberts et al., Reference Roberts, Quach, Mensah, Gathercole, Gold, Anderson and Wake2015).

The development of working memory capacity is thought to be of critical importance to academic outcomes and the development of complex cognitive skills, such as language, literacy, and mathematical ability (DeMarie & López, Reference DeMarie, López, Bauer and Fivush2013). An expanding body of evidence has demonstrated a link between working memory and the performance of specific academic abilities (Aronen, Vuontela, Steenari, Salmi, & Carlson, Reference Aronen, Vuontela, Steenari, Salmi and Carlson2005; DeMarie & López, Reference DeMarie, López, Bauer and Fivush2013; Gathercole, Pickering, Knight, & Stegmann, Reference Gathercole, Pickering, Knight and Stegmann2004). For instance, one study by Monette, Bigras, and Guay (Reference Monette, Bigras and Guay2011) showed that when children begin school, their working memory predicts their later academic ability over and above pre-existing differences in family, affect and ability variables. Other researchers have shown a positive relationship between working memory capacity and the development of reading and literacy (Normand & Tannock, Reference Normand and Tannock2014; St Clair-Thompson & Gathercole, Reference St Clair-Thompson and Gathercole2006) and mathematics (Bull, Espy, & Wiebe, Reference Bull, Espy and Wiebe2008; Bull & Scerif, Reference Bull and Scerif2001). Importantly, working memory is believed to be of critical importance when learning new information and novel concepts (Pickering, Reference Pickering, Alloway and Gathercole2006).

Previous research has suggested that poor working memory function is linked to poor academic performance in primary school children (Gathercole & Pickering, Reference Gathercole and Pickering2000). For example, this includes poor attention as a key necessary element of working memory, and working memory overload when, for example, students cannot manage multiple instructions. It is suggested that about 10% to 15% of children in regular classrooms have working memory difficulties and that this negatively impacts on their academic learning and progress (Gathercole & Alloway, Reference Gathercole and Alloway2008). Given that successful attention, listening to, and following the teachers’ instructions are key elements of effective functioning in most classrooms from middle primary school onwards, it is unsurprising that addressing and remediating the effects of working memory problems have become increasingly important concerns for school psychologists and educators (e.g., Minear & Shah, Reference Minear, Shah and Pickering2006; St Clair-Thompson & Gathercole, Reference St Clair-Thompson and Gathercole2006).

Working Memory Interventions

Typically, interventions aim to assist children with working memory difficulties by targeting one of three key pathways:

1. 1. Training the teacher to make appropriate adaptations within the classroom context (e.g., Elliot, Gathercole, Alloway, Holmes, & Kirkwood, Reference Elliott, Gathercole, Alloway, Holmes and Kirkwood2010);

2. 2. Ensuring the students are provided with personal access to strategies to manage tasks using attention and working memory skills independently within the classroom context (e.g., Colmar et al., Reference Colmar, Davis and Sheldon2016; Davis et al., Reference Davis, Sheldon and Colmar2014); and

3. 3. Training working memory capacity directly, typically using a computerised working memory program, with Alloway, Robinson, and Frankenstein (Reference Alloway, Robinson, Frankenstein, Strobach and Karbach2016) describing narrow-scope and broad-scope training programs, where the former aims to train skills similar to those used in the pre- and post-tests, and the latter trains working memory in the context of broader abilities, including attention and learning skills.

This paper will now briefly and critically review the evidence for the efficacy of working memory interventions both within the classroom and using computerised working memory interventions, particularly with regard to their effectiveness at bringing about measurable benefits to students’ academic performance and classroom behaviour and engagement. However, it is first necessary to establish the criteria by which the interventions will be judged, and this is particularly important given that different expectations about what constitutes a successful working memory intervention have complicated the interpretation of intervention studies (Birney, Reference Birney2015). With regard to working memory interventions, researchers and psychologists are typically interested in three levels of effect, which are:

1. 1. near transfer effects, which are improvements on a similar task to the trained task but using different stimuli or response mode;

2. 2. far transfer effects, which are gains on a task dissimilar to the trained task, such as a measure of fluid intelligence; and

3. 3. improvement in everyday functioning, academic performance and behaviour, which is known as generalisation.

School counsellors, school psychologists, and teachers are particularly interested in the extent to which working memory interventions can improve student outcomes on measures of academic performance, such as standardised tests (e.g., Elliot et al., Reference Elliott, Gathercole, Alloway, Holmes and Kirkwood2010). For our purposes, we will consider evidence at each level of effect, but regard them hierarchically; that is, transfer to everyday classroom functioning is more persuasive than far transfer, which is in turn more persuasive than near transfer.

Working Memory Interventions in the Classroom

Logically, given the strong relationship between working memory and academic achievement, classroom-oriented interventions should provide effective conditions for optimal transfer or generalisation, as the context and the students’ learning and academic requirements within the classroom are closely related. To date, most classroom-based working memory interventions have focused on training the teacher to support students by providing key adjustments within the classroom. A number of books and articles, summarising key adaptations for children with working memory difficulties, and written for teachers, parents and school psychologists, have been published (Alloway, Reference Alloway2006, Reference Alloway2011; Dehn, Reference Dehn2008; Elliott et al., Reference Elliott, Gathercole, Alloway, Holmes and Kirkwood2010; Gathercole & Alloway, Reference Gathercole and Alloway2007, Reference Gathercole and Alloway2008). For example, ensuring initial attentiveness by engaging the whole class, and the simplification of the instructions and their presentation in more than one mode, such as spoken and written, are relevant adaptations that most teachers can easily make. Another key role in training teachers to successfully assist students, particularly those with working memory difficulties, is to ensure that teachers are observant of the behavioural indicators of working memory overload. The teacher can then immediately adapt within context and provide additional support, thus ensuring the student does not switch off or become disengaged from the classroom task. DeMarie and López (Reference DeMarie, López, Bauer and Fivush2013) sum up the teacher's role succinctly when they state that ‘instead of teaching students only procedures of what to do, teachers need to spend instructional time helping students to understand how to study, and why using particular strategies will help them to learn different types of material’ (p. 858).

Research within classroom contexts has been very limited. Elliott et al. (Reference Elliott, Gathercole, Alloway, Holmes and Kirkwood2010) focused on training teachers how to provide an appropriate classroom environment to support students with working memory difficulties. The intervention period of eight months was implemented across three groups: intervention, control, and academic direct instruction. No differences were found between the groups across several outcome measures, including working memory, reading, and mathematics. Importantly, they noted that the teachers’ use of supportive adjustments, independent of group allocation, were effective, suggesting that teacher strategies per se can make a positive difference.

Students may be able to make greater gains in their working memory capacity and use their working memory skills more often and with greater and broader facility by following a specific classroom-integrated intervention program targeted directly at the students, facilitating their independent use of and ownership of the strategies. Recently, Colmar and colleagues (Colmar et al., Reference Colmar, Davis and Sheldon2016; Davis et al., Reference Davis, Sheldon and Colmar2014; Sheldon, Davis, & Colmar, Reference Sheldon, Davis and Colmar2015) developed a unique classroom-based working memory intervention called Memory Mates that addresses the two classroom-oriented approaches simultaneously: (1) training the teacher to become aware of and to provide adjustments to facilitate students’ attention and working memory, and (2) providing the students individually with strategies to use independently to manage attention and working memory difficulties.

Memory Mates consists of 10 student-relevant attention and working memory icons, with short explanations derived from the research literature (Davis & Sheldon, 2013, as cited in Colmar et al., Reference Colmar, Davis and Sheldon2016; see Figure 1), and a booklet for teachers outlining how to implement the Memory Mates’ strategies as a classroom intervention (Sheldon et al., Reference Sheldon, Davis and Colmar2015). Students are taught and supported in using the 10 strategies focused on how to best attend, learn, and engage in classroom activities across all academic areas.

FIGURE 1 Memory Mates icons and explanations.

In an initial article involving targeted students within one class, major improvements in engagement were observed, with positive feedback from the teacher, and this research work assisted in the further development of the specific strategies (Davis et al., Reference Davis, Sheldon and Colmar2014).

In a small-scale evaluation of the Memory Mates project with two classes in a single school, one functioning as the intervention classroom and one as the control, no significant impact on academic outcomes was noted (Colmar et al., Reference Colmar, Davis and Sheldon2016). However, teachers’, students’ and parents’ feedback was very positive, with interesting meta-cognitive insights from students about how they used the strategies at school and home. Further, children particularly enjoyed having a peer memory coach who could assist them when necessary (see Figure 1). The short period of intervention, which was 5 weeks, was the most likely reason for the inconclusive findings.

In a comprehensive evaluative study (Colmar et al., Reference Colmar, Double, Davis, Sheldon, Phillips, Cheng and Briddon2017), teachers of six classes of Year 4 students implemented Memory Mates for 8 months, with seven classes functioning as a waitlist control group. Results confirmed significant changes in mathematics and spelling in the Memory Mates group compared with no-contact controls. Reading comprehension and academic engagement were not significantly different, although reading engagement increased in both groups over the period of intervention. In addition, students responded well and again showed meta-cognitive insights about their own attention and working memory processes. These significant and interesting findings suggest that directly providing young students with child-appropriate strategies is an important and currently under-researched approach.

One of the key issues facing classroom-based working memory interventions is the need to address attention directly, rather than solely targeting working memory. Inattentiveness can be a key indicator of working memory difficulties in the classroom context (Gathercole, Lamont, & Alloway, Reference Gathercole, Lamont, Alloway and Pickering2006), as well as being critical in ensuring a student gains information fully, ready to encode it into working memory. It has been postulated that inattentiveness occurs and persists because students can become overloaded with instructions and information. Therefore, they are unable to cope well with the educational demands of a classroom, or to sustain academic engagement and academic progress (Gathercole & Alloway, Reference Gathercole and Alloway2007). Further, these students may present with behavioural difficulties and disrupt other students’ learning. In the Memory Mates approach, attention is specifically targeted in the first two strategies (see Figure 1).

Computerised Interventions

To date, most working memory research and training has focused on improving students’ working memory capacity outside the classroom, predominantly using computerised programs; for example, Cog Med (Klingberg et al., Reference Klingberg, Fernell, Olesen, Johnson, Gustafsson, Dahlstrom and Westerberg2005) and Jungle Memory (Alloway, Reference Alloway2012; Alloway & Alloway, Reference Alloway and Alloway2013). Most computerised working memory programs focus on increasing working memory capacity, with an assumption that transfer or generalisation to everyday contexts, including learning in classrooms, will take place. However, while these commercial programs make optimistic claims about their effectiveness, overall support for the efficacy of such programs within the academic literature is more equivocal. It is particularly difficult to reach strong conclusions about these interventions given that they have often been evaluated on atypical populations, where working memory issues are considered central to the child's impairment; in particular, in relation to attention deficit hyperactivity disorder (ADHD; e.g., Chacko et al., Reference Chacko, Bedard, Marks, Feirsen, Uderman, Chimiklis and Zwilling2014; Egeland, Aarlien, & Saunes, Reference Egeland, Aarlien and Saunes2013; Gray et al., Reference Gray, Chaban, Martinussen, Goldberg, Gotlieb, Kronitz and Tannock2012).

A number of studies have found promising benefits for cognitive abilities as a result of computerised cognitive training. Klingberg et al. (Reference Klingberg, Fernell, Olesen, Johnson, Gustafsson, Dahlstrom and Westerberg2005) found that children with ADHD who underwent adaptive computerised working memory training experienced significant improvements in working memory, complex reasoning, and response inhibition compared with a control group who completed the same training but in a non-adaptive fashion. Other researchers have found benefits to fluid intelligence, which has a well-established role in educational attainment (Deary, Strand, Smith, & Fernandes, Reference Deary, Strand, Smith and Fernandes2007). For example, Jaeggi, Buschkuehl, Jonides, and Shah (Reference Jaeggi, Buschkuehl, Jonides and Shah2011) found that children who improved on a computer-game-like working memory task showed benefits to untrained measures of fluid intelligence. Similarly, Bergman Nutley et al. (Reference Bergman Nutley, Söderqvist, Bryde, Thorell, Humphreys and Klingberg2011) found that compared to a non-adaptive control condition, children trained on a non-verbal reasoning task significantly improved on measures of fluid intelligence. In the same study, children trained in working memory made improvements in working memory tasks, but with no impact on fluid intelligence.

Ultimately, school psychologists, school counsellors, and educators are most interested in programs that can demonstrate direct benefits to academic outcomes, and a number of studies have shown significant benefits of computerised training to academic performance (Dahlin, Reference Dahlin2011; Kroesbergen, van't Noordende, & Kolkman, Reference Kroesbergen, van ’t Noordende and Kolkman2014). Holmes, Gathercole, and Dunning (Reference Holmes, Gathercole and Dunning2009) reported that children with working memory difficulties, who were trained on a range of adaptive working memory tasks for 20 days, showed significant gains on a standardised test of mathematics at a 6-month follow-up; however, given that the control groups were not retested, it is difficult to separate this effect from natural development. In addition, there is some evidence that cognitive training benefits reading processes in children (Loosli, Buschkuehl, Perrig, & Jaeggi, Reference Loosli, Buschkuehl, Perrig and Jaeggi2011), including children with special needs (Dahlin, Reference Dahlin2011). For instance, Loosli et al. (Reference Loosli, Buschkuehl, Perrig and Jaeggi2011) found that when compared to a matched no-contact control group, a working memory training group showed significant improvements in text and word reading.

Other research studies have reached less optimistic findings with regard to training outcomes. A number of studies have demonstrated that working memory training was able to lead to improvements in memory (or at least scores on cognitive task designed to measure memory), but frequently these gains fail to generalise to academic outcomes (Banales, Kohnen, & McArthur, Reference Banales, Kohnen and McArthur2015; Chacko et al., Reference Chacko, Bedard, Marks, Feirsen, Uderman, Chimiklis and Zwilling2014; Henry, Messer, & Nash, Reference Henry, Messer and Nash2013; van der Donk, Hiemstra-Beernink, Tjeenk-Kalff, Van Der Leij, & Lindauer, Reference van der Donk, Hiemstra-Beernink, Tjeenk-Kalff, Van Der Leij and Lindauer2015). For instance, Dunning, Holmes, and Gathercole (Reference Dunning, Holmes and Gathercole2013) compared improvements in a group of low working memory children trained on an adaptive working memory taskFootnote ¹ with a group trained on a non-adaptive task. They found that adaptive training led to selective improvements on untrained working memory tasks; however, classroom and academic tasks did not improve relative to non-adaptive difficulty controls, concluding that ‘gains do not translate into capacity improvements on ecologically valid measures of working memory or to gains in academic progress’ (p. 923). A recent large-scale Australian study following up the impact of CogMed after 2 years, examining the maintenance of near transfer effect and generalisation to academic outcomes, found it had no sustainability or academic impact (Roberts et al., Reference Roberts, Quach, Spencer-Smith, Anderson, Gathercole, Gold and Wake2016). Further, in a cost benefit analysis, these researchers concluded that CogMed was also expensive and time consuming, with a concomitant loss in classroom learning time.

Meta-analyses confirm that while there may be studies that have found transfer effects of working memory training, overall these effects tend to be short-lived and rarely generalise to academic abilities. A number of reviews have reached sceptical conclusions about the effectiveness of cognitive training (Melby-Lervag & Hulme, Reference Melby-Lervag and Hulme2013; Shipstead, Redick, & Engle, Reference Shipstead, Redick and Engle2012). For example, a recent meta-analysis examining working memory intervention research completed with children diagnosed with ADHD concluded that the studies were often methodologically poor and most showed little impact on behavioural, cognitive, or academic skills (Rapport, Orban, Kofler, & Friedman, Reference Rapport, Orban, Kofler and Friedman2013). Similarly, in their review of the literature, Melby-Lervåg and Hulme (Reference Melby-Lervag and Hulme2013) concluded that there is currently no evidence that the benefits of computerised working memory training generalise to other skills or are maintained over the long term. However, it should be noted that interpreting these studies is difficult, as currently much of the research into the effectiveness of working memory training has been carried out on atypical groups of students (in particular, ADHD samples), and the extent to which these findings generalise to normally functioning children is unclear. It is also difficult to interpret many meta-analytical reviews of the literature, as there is often a large degree of heterogeneity in the studies being analysed, particularly in terms of their sample characteristics, the cognitive ability being trained and the training task.

The inconsistent findings reviewed above make it difficult to draw definitive conclusions about the effectiveness of computerised working memory training for children. The contradictory evidence can be explained by a number of factors, including differences in the tools used to measure working memory, the research methodology used, and the theoretical approach taken to training (Birney, Reference Birney2015; Lampit, Hallock, & Valenzuela, Reference Lampit, Hallock and Valenzuela2014; Morrison & Chein, Reference Morrison and Chein2011). In particular, Morrison and Chein (Reference Morrison and Chein2011) draw a distinction between strategy and capacity or core training, and they also note that evaluating the success of a computerised program is highly dependent on the researchers’ definitional and conceptual stance, the areas of working memory that they target, and their outcome measures. They further suggest that strategy training may impact on working memory capacity even if temporarily (see also Witt, Reference Witt2011).

One issue that cognitive training programs face is developing training regimes that are both engaging and effective, which is particularly important for children. Specifically designed cognitive training programs have typically been developed based on redesigns of traditional cognitive assessments (e.g., Stroop test, n-back). While it has been assumed that these activities are the most effective ways to improve cognitive abilities (or at least scores on measures of cognitive abilities), they are not typically engaging over longer sessions or repeated plays (Deveau, Jaeggi, Zordan, Phung, & Seitz, Reference Deveau, Jaeggi, Zordan, Phung and Seitz2014). Specifically, computerised training is autonomous in nature; it is usually completed at a learner's own pace, often in their home and at the learner's discretion. Learning within such a context is therefore likely to require effective self-regulation on behalf of the learner (Sitzmann & Ely, Reference Sitzmann and Ely2011) in order for them to train regularly and engage with the training task. However, as working memory is highly involved in self-regulation (Hofmann, Friese, Schmeichel, & Baddeley, Reference Hofmann, Friese, Schmeichel, Baddeley, Vohs and Baumeiste2011), children who have working memory difficulties are more likely to struggle with the independent and autonomous nature of computerised training.

Recommendations for Practice and Future Research

This article has reviewed the current evidence base for a number of interventions for children with working memory difficulties. Although we have separated these interventions into classroom-based and computer-based, this should be regarded more as a convenient division for looking at working memory interventions, particularly within the current literature, rather than a static dichotomy. However, importantly, despite the publication of a number of books and articles about working memory in children and adolescents, very limited research has focused on training teachers to support students with working memory difficulties. The single exception is Elliot et al. (Reference Elliott, Gathercole, Alloway, Holmes and Kirkwood2010). Further, only one team (Colmar, Davis, and Sheldon's work on Memory Mates; Colmar et al., Reference Colmar, Davis and Sheldon2016; Davis et al., Reference Davis, Sheldon and Colmar2014; Sheldon et al., Reference Sheldon, Davis and Colmar2015) has focused on developing independent attention and working memory strategies and skills in the students themselves. Many students experience difficulties within classrooms, both cognitive and academic, with a negative impact on their wellbeing. Working with teachers and providing strategies to students to enhance and support classroom learning, through attention and working memory strategies, is a key role that school psychologists and counsellors ideally need to know about and to engage with.

To the authors’ knowledge there have been no intervention studies that have evaluated the extent to which classroom and computerised interventions could complement one another when used together, particularly in their impact on working memory capacity, the use of, and the frequency of use of a variety of working memory strategies, and academic performance. Future research could explore the differential outcomes of using computerised and classroom interventions together, compared with just one form of intervention.

Children appear to learn better strategies for utilising their attention and working memory more effectively as they get older, and as they adapt to increasing academic demands (Gathercole, Reference Gathercole2008; Witt, Reference Witt2011). A program such as Memory Mates aims to enhance this developmental progression in functionally relevant contexts by targeting students’ attention and working memory within the classroom context (Colmar et al., Reference Colmar, Davis and Sheldon2016; Davis et al., Reference Davis, Sheldon and Colmar2014), and with teacher training and support (Sheldon et al., Reference Sheldon, Davis and Colmar2015). Memory Mates aims to increase students’ use of a greater variety of working memory skills, the frequency of use of newly learned and existing skills, and students’ working memory capacity.

Overall, it is recommended that school psychologists and counsellors support educators to take a multilayered approach by creating a classroom environment that minimises working memory demands and facilitates developing attention and working memory strategies in students, with teacher support. Importantly, students need to be provided concurrently with working memory strategies, such as those presented in the Memory Mates program, that they can use independently, and thus develop and self-regulate their own working memory skills and capacity.