Journal of Experimental Child Psychology 119 (2014) 112–119
Contents lists available at ScienceDirect
Journal of Experimental Child Psychology journal homepage: www.elsevier.com/locate/jecp
Brief Report
Visuospatial bootstrapping: Implicit binding of verbal working memory to visuospatial representations in children and adults Stephen Darling a,⇑, Mary-Jane Parker a, Karen E. Goodall a, Jelena Havelka b, Richard J. Allen b a b
Division of Psychology and Sociology, Queen Margaret University, Edinburgh EH21 6UU, UK Institute for Psychological Sciences, University of Leeds, Leeds LS2 9JT, UK
a r t i c l e
i n f o
Article history: Received 13 May 2013 Revised 8 October 2013 Available online 25 November 2013 Keywords: Working memory Verbal memory Spatial memory Visual memory Bootstrapping Episodic Buffer
a b s t r a c t When participants carry out visually presented digit serial recall, their performance is better if they are given the opportunity to encode extra visuospatial information at encoding—a phenomenon that has been termed visuospatial bootstrapping. This bootstrapping is the result of integration of information from different modalityspecific short-term memory systems and visuospatial knowledge in long term memory, and it can be understood in the context of recent models of working memory that address multimodal binding (e.g., models incorporating an episodic buffer). Here we report a cross-sectional developmental study that demonstrated visuospatial bootstrapping in adults (n = 18) and 9-year-old children (n = 15) but not in 6-year-old children (n = 18). This is the first developmental study addressing visuospatial bootstrapping, and results demonstrate that the developmental trajectory of bootstrapping is different from that of basic verbal and visuospatial working memory. This pattern suggests that bootstrapping (and hence integrative functions such as those associated with the episodic buffer) emerge independent of the development of basic working memory slave systems during childhood. Ó 2013 Elsevier Inc. All rights reserved.
⇑ Corresponding author. E-mail address: [email protected] (S. Darling). 0022-0965/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jecp.2013.10.004
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
113
Introduction There is considerable evidence that verbal and spatial information is handled by different shortterm memory processes (see Baddeley, 2000). This sits alongside considerable evidence of substantial shared variance within measures that involve the processing, rather than the simple storage, of remembered items (Engle, Tuholski, Laughlin, & Conway, 1999). Evidence of both types underlies the development of theoretical approaches that clearly differentiate process-dependent memory systems from storage systems where information is stored but not manipulated. One such approach is the working memory model (Baddeley, 1986; Baddeley & Hitch, 1974), which proposed the interaction of a central executive (CE) processing component alongside modality-specific passive storage systems (the phonological loop and visuospatial sketchpad). An alternative model (Engle, 2010; Engle et al., 1999) proposed that storage functions labeled as short-term memory (STM) may be supported by distinct subsystems, whereas processing functions labeled as working memory (WM) require a common integrated set of processes associated with a substantial shared variance. Even theoretical approaches that deemphasize modality-specific systems (Cowan, 2005) still accommodate the idea that basic storage processes may be differentiable on the basis of task demands. Despite this, STM subsystems are not entirely discrete and information stored in separate modalities can be linked (Mate, Allen, & Baqués, 2012: Morey & Cowan, 2004, 2005; see Baddeley, 2000, for a review of earlier evidence). Recently, the integration of visuospatial, verbal, and long-term memory has been demonstrated in verbal serial recall. When to-be-remembered digits were presented in a familiar visuospatial array—the standard numeric keyboard used in mobile telephones—memory was facilitated compared with when digits were presented in a single location (Darling & Havelka, 2010). Participants were asked to attend to only a single stimulus dimension (digit serial recall), but the performance improvement associated with keypad presentation indicated that they were able to extract visuospatial information and integrate it with the verbal material. Hence, this pattern was described as visuospatial bootstrapping (Darling & Havelka, 2010) because verbal memory performance was bootstrapped by the integration of redundant visuospatial information that was present in the keypad condition and not in the single item condition. A subsequent study has replicated visuospatial bootstrapping (Darling, Allen, Havelka, Campbell, & Rattray, 2012), also showing that the availability of a compatible representation in long-term memory (LTM) is necessary for observing bootstrapping. Visuospatial bootstrapping represents an effect where nonverbal memory processes interact with verbal memory; a parallel result has recently been reported, where visuospatial memory is differentially impaired by concrete and abstract verbal load (Mate et al., 2012). Taken together, these patterns are inconsistent with the possibility that WM subsystems are fully independent of each other. They also suggest that interactions between WM systems take place on an implicit level—or at least without explicit instruction to combine information from different sources. The episodic buffer (Baddeley, 2000) was proposed as an additional WM component capable of maintaining cross-modality bindings between information in LTM and STM, with a role in the encoding of specific episodes. Unlike the CE and other theoretical mechanisms targeted at understanding how information in WM is integrated (e.g., the focus of attention described by Cowan, 2005), the episodic buffer is now thought to require neither executive nor attentional resources to function, instead operating in an efficient, automatic, and rule-governed manner (Baddeley, Allen, & Hitch, 2011). Given this, it seems plausible that the potentially implicit visuospatial bootstrapping data might be explained with reference to the episodic buffer (Darling et al., 2012). Aside from the episodic buffer model, there are alternative accounts of WM that can potentially offer a theoretical account of visuospatial bootstrapping. Unsworth and Engle (2007) proposed that effective WM performance entails maintenance in primary memory alongside effective search of secondary memory, and it is plausible that the bootstrapping effect occurs because keypad displays facilitate this search of secondary memory due to the provision of additional cues. There is also robust evidence that reconstructive processes facilitate temporary memory (Cowan et al., 2003; Towse, Cowan, Hitch, & Horton, 2008; Towse, Hitch, Horton, & Harvey, 2010). Given that WM for spatial configurations is certainly affected by aspects of long-term spatial memory (Brown & Wesley, 2013), it is plausible that richer spatial arrays at presentation lead to more effective reconstructions.
114
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
In the current study, we asked whether there was evidence of multimodal WM processing in samples of children and adults by assessing evidence for visuospatial bootstrapping across the developmental time course. Although separate verbal and visuospatial short-term memory systems are known to be observable in children (e.g. Pickering, Gathercole, & Peaker, 1998), subsequent research (e.g., Alloway, Gathercole, & Pickering, 2006; Bayliss, Jarrold, Gunn, & Baddeley, 2003) has further suggested that storage functions of verbal and visuospatial memory systems are easily segregable, whereas processing functions are more unified, consistent with the idea that STM is differentiated in children, whereas WM processing functions are more easily understood in a domain-general framework. Research specifically probing the episodic buffer in children (Alloway, Gathercole, Willis, & Adams, 2004) has suggested its emergence as early as 4 years of age (see also Kapikian & Briscoe, 2012). However, the task used to assess it (sentence recall) was restricted to the verbal modality and, thus, does not fully characterize the multimodal aspects of the proposed episodic buffer (Baddeley et al., 2011). Other studies of multimodal WM in children have examined binding between color and location (Cowan, Naveh-Benjamin, Kilb, & Saults, 2006a) and names and locations (Cowan, Saults, & Morey, 2006b) in 8-year-old children and found reduced accuracy relative to adults, but these studies did not consider how information from different domains is integrated with long-term knowledge. The specific benefits of assessing visuospatial bootstrapping with the goal of understanding the integration of information across WM components are threefold. First, bootstrapping is superficially an implicit phenomenon and, hence, should not recruit central attentional processes. Second, there is no requirement to manipulate any information; participants should benefit from keypad-type displays by integrating visuospatial information, verbal information, and LTM knowledge to form an STM trace that is more resilient than an unintegrated one, but there is no need to judge, reorder, or process the information to obtain this benefit. Third, any bootstrapping effect is undeniably multimodal, bridging the two best-documented modality-specific STM systems. Although there is also evidence of a gradual capacity increase in WM as age increases (Gathercole, 1999), specific critical periods of change in relationships between modality-specific STM systems might be expected to occur somewhere after 6 or 7 years of age as visuospatial STM (arguably) becomes differentiated from the CE (Alloway et al., 2006) and as verbal rehearsal becomes available as a mnemonic method (Gathercole & Hitch, 1993). We recruited groups of 6- and 9-year-old children to span these changes. We also recruited a sample of young adults to serve as a comparison sample. To minimize the complicating effects of verbal span improvement as age increases (Alloway et al., 2006; Pickering et al., 1998), participants initially had their capacity assessed in a standard digit span task; subsequently, they were tested under different display conditions at their span. Three display configurations were used. The Single display involved presenting all to-be-remembered digits in a single location. The Typical display presented a representation of a standard numeric keypad within which the to-be-remembered digits were indicated by highlighting numbers in the display. The Novel display was similar but used a grid where digits were located randomly in order to assess the need for a preexisting LTM representation (see Fig. 1 in Darling et al., 2012, p. 260, for illustrations of these displays).
Method Participants Two groups of 15 children attending an after-school club in Peebles, United Kingdom, took part: a group of 6-year-olds (mean age = 80 months, SD = 4, 6 girls and 9 boys) and a group of 9-year-olds (mean age = 111 months, SD = 5, 8 girls and 7 boys). The parents of all child participants gave informed consent in writing, and the children themselves also gave verbal consent. None of the children had any known additional learning needs or sensory impairments. Child participants took part in a quiet room in their after-school club. An adult comparison sample was recruited among students and staff of Queen Margaret University. The sample size was 18 (mean age = 25.17 years, SD = 8.87). All adult participants gave informed
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
115
1 Single Typical Novel
Proporon correct
0.8
0.6
0.4
0.2
0 Younger children (~6Y)
Older children (~9Y)
Adults
Age group
Fig. 1. Memory performance (proportion of trials remembered correctly) across display condition and age group. Error bars represent standard errors of the mean.
consent in writing. Adult participants took part in a laboratory at the university. This research was approved by the research ethics committee of the university. Design On the first day, participants carried out a quick assessment of their knowledge of the standard telephone keypad. Following this, their digit serial recall span was assessed using a single digit display. On 3 subsequent days, participants took part in sets of trials in each of three display conditions: Single, Typical, and Novel. In these different conditions, participants were tested with digit sequence lengths of the maximum length correctly recalled in the span task. The order of these conditions was counterbalanced. Participants in the adult condition took part in all conditions on the same day and omitted the keypad knowledge check. Materials and procedure Keypad knowledge assessment Children were shown a blank grid containing 10 empty squares in the equivalent positions to the keys on a standard numeric telephone keypad. They were asked to write in the digits 0 to 9 in the correct positions that they appeared on a telephone. Experimental tasks Stimuli were presented using a standard personal computer. The display was set to a resolution of 1024 768 pixels. Each trial was initiated by participants pressing a key, and then a fixation cross was displayed for 500 ms, followed by a blank screen interval of 250 ms and then by digit sequence presentation. Digits were always presented in Arial font, point size 18, and presented centrally within squares of side 60 px. Once all digits in a given trial had been shown, the screen cleared for a 1000ms retention interval and then the message ‘‘RECALL’’ was presented in the middle of the screen. At this point, participants attempted to repeat the digits aloud in the correct order. Span assessment To assess maximum span, participants were asked to remember random sequences of digits using Single displays; digits were presented in the middle of the screen, inside a square with a green
116
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
background. Each digit was shown for 1000 ms, and between each digit in the sequence there was a 250-ms interval during which the screen was blank. Trials were presented in pairs. In the first 2 trials, participants were shown and then recalled a single digit. If they correctly recalled at least 1 trial, the sequence was increased in size to two items and a further pair of trials was shown. If participants recalled at least one sequence correctly, then a further pair was presented with a sequence of three. This incrementing of sequence length by one item was repeated until participants failed to recall any sequences correctly at a given sequence length. Span was defined as the maximum sequence length at which participants could recall at least one sequence correctly. Performance at span: Single digit display Participants undertook a set of 24 trials. In each trial, they needed to remember a random sequence of digits. Sequence size was set at individual span. Digits were presented in a single green square in the middle of the screen. Timing parameters were the same as in the span task. Performance at span: Typical keypad display In the Typical condition, the digits 0 to 9 were presented in the same array used in a traditional telephone keypad, aligned centrally on the screen. Detailed dimensions of the array were the same as those used by Darling and colleagues (2012). The digit sequences were presented by highlighting individual digit backgrounds in green. Timing of the highlighting of the digits followed the timing used in the Single condition, with each digit being highlighted for 1000 ms and a blank screen interval of 250 ms being imposed between sequence items. Performance at span: Novel static keypad display The Novel condition was identical to the Typical condition except that digits were presented in nontypical locations so that on beginning the study the relationship between digits and locations was unfamiliar. Digit locations were consistent throughout all trials. Results Keypad knowledge Of the younger children, 7 correctly completed the grid, whereas 8 did not. In the older group, 13 responded correctly and 2 did not. This association between age and successful completion of the keypad task was significant, v2(1) = 5.40, p = .020. Span The mean span was 4.47 items (SD = 0.64, min = 3, max = 5) for the 6-year-old group, 5.13 items (SD = 0.74, min = 3, max = 6) for the 9-year-old group, and 6.72 items (SD = 1.32, min = 5, max = 9) for the young adult participants. The main effect of digit span was significant, F(2, 45) = 22.45, p < .001, g2p = .51. Individual t tests indicated that the adults displayed significantly higher digit span than the 9-year-olds, t(31) = 4.14, p < .001, d = 1.49, whereas span was higher in the 9-year-olds compared with the 6-year-olds, t(31) = 2.63, p = .01, d = 0.95. Effects of age and display configurations on memory The proportions of the 24 trials in the three display conditions that were correctly recalled are summarized in Fig. 1. Data were assessed using a mixed-design 3 (Age Group) 3 (Display Condition) analysis of variance (ANOVA), indicating no significant main effect of age group, F(2, 45) = 1.53, p = .228, g2p = .06. There was a nonsignificant trend toward a main effect of display, F(2, 90) = 2.57, p = .082, g2p = .05. This was qualified by a significant interaction between age group and display, F(4, 90) = 4.37, p = .003, g2p = .16. This interaction is shown in Fig. 1; the 9-year-old and adult participants
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
117
remembered more items in the Typical condition than in either the Single or Novel condition, whereas this was not the case for the 6-year-old group. The simple main effect of display was significant in adult participants, F(2, 44) = 3.94, p = .027, g2p = .15. Multiple comparisons showed that digit memory in the Typical condition was significantly better than that in both the Single (p = .017) and Novel (p = .016) conditions, and the Novel and Single conditions did not differ significantly (p = .905). The simple main effect of display was also significant for the 9-year-old participants, F(2, 44) = 4.93, p = .012, g2p = .18); performance in the Typical condition was significantly better than that in both the Single (p = .024) and Novel (p = .004) conditions, and the Novel and Single conditions did not differ significantly (p = .559). The simple main effect of display in the 6-year-old participants did not reach significance, F(2, 44) = 2.41, p = .102, g2p = .10. To establish whether there were any differences in the pattern of performance between the 9-year-olds and adults, a 2 3 mixed-design ANOVA was conducted on the 9-year-old and adult groups only. There was a significant main effect of display, F(2, 62) = 10.81, p < .001, g2p = .26, but neither the main effect of age, F(1, 31) = 2.63, p = .115, g2p = .08, nor the interaction, F(2, 62) = 0.23, p = .80, g2p = .01, was significant, a pattern indicative of comparable performance with no significant differences between the 9-year-old and adult participants.
Discussion In the current study, 9-year-olds showed evidence of visuospatial bootstrapping that was effectively the same as that observed in adult participants. In contrast, at 6 years of age there was no evidence of a bootstrapping effect. Because of the individual titration of difficulty used in this study, it is unlikely that overall lower verbal span in 6-year-olds could be the reason behind the absence of any bootstrapping effect in that group—an inference that is supported by the lack of a main effect of age group in the main experimental conditions. A strong inference can be drawn that the bootstrapping effect was not a simple result of development of basic WM, either verbal or visuospatial. Capacity, as assessed in the initial digit span task, increased with age; a gradual increase was evident among 6-year-olds, 9-year-olds, and adults. This is typical of patterns seen in developmental studies of memory capacity in both visual and verbal domains where capacity increases are asymptotic later in adolescence (see Gathercole, 1999, for a summary). This contrasted with the development of bootstrapping, which was completely absent in 6year-olds but was equivalent in 9-year-olds and adults, suggesting that the cognitive resources supporting bootstrapping are maturationally discrete from those supporting digit span and visuospatial WM. The forgoing discussion raises the issue of which mechanisms may develop between 6 and 9 years of age to facilitate bootstrapping. The current data are equivocal between several reasonable possibilities. One candidate is consolidation of LTM learning; LTM representations of the typical keypad are critical in order to demonstrate a bootstrapping effect over 24 trials in adults (Darling et al., 2012), so it is possible that acquisition of a reliable representation of the typical keypad may drive the emergence of bootstrapping in older children. This could be due to acquisition of new learning regarding the typical keypad and its importance, or it may be a consequence of increased efficiency of LTM systems themselves over development; both possibilities are equally consistent with the observation that older children were better at repopulating the blank keypad grid from LTM in the keypad assessment task. Darling and colleagues (2012) observed that 24 trials was insufficient to develop a sufficiently robust representation of a novel keypad for a beneficial bootstrapping effect to emerge (in an adult sample), suggesting that such LTM representations needed to be reasonably well established and could not be learned satisfactorily over the course of a single set of trials. A second possibility is that the ability to integrate information between modality-specific cognitive subsystems and LTM into a robust representation develops between 6 and 9 years of age. This kind of process is held to be the role of the episodic buffer (Baddeley et al., 2011), and the current data also fit well with such an explanation. The assertion that the bootstrapping task invokes multimodal representations that are implicit remains speculative, and to be certain that there are no central attention resources implicated in the bootstrapping task requires further work beyond the scope of this study. A third possible
118
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
explanation is that the bootstrapping pattern is a consequence of explicit strategic use of visuospatial resources and that 6-year-olds were unable to explicitly select the (better) visuospatial strategy. The literature suggests some grounds to suspect that this might happen; children older than 7 years spontaneously convert visually presented stimuli into a verbal format, whereas younger children do not (Hitch, Halliday, Dodd, & Littler, 1989). The emergence of bootstrapping, therefore, may be reflective of the development of metacognitive skill. Relatedly, it may be that improvements in bootstrapping reflect improvements in controlled searching of (visuospatial) secondary memory (Unsworth & Engle, 2007). Settling which of these competing explanations is most satisfactory is an important goal for future research addressing the theoretical underpinnings of bootstrapping. The current observations are both novel and noteworthy in that they represent the first evidence showing visuospatial bootstrapping in children. The observed data reflect a pattern of integration of spatial information to assist STM that is clearly cross-modal. The task demands do not explicitly require multimodal representation, and neither do they explicitly require processing or manipulation of information, yet the results could have occurred only if multimodal connections had been established. Moreover, the data demonstrate that the bootstrapping effect seen in 9-year-olds was comparable to that seen in adults but was not observed in 6-year-olds; hence, the bootstrapping effect is relatively independent of the development of basic WM capacity. It is also noteworthy that the maturational trajectory of visuospatial bootstrapping is very different from that of complex span (which continues to develop through adolescence; e.g., Siegel, 1994; see also Gathercole, 1999). In fact, given that the maturation of bootstrapping appears to be complete by around 9 years of age, and that this precedes asymptotic performance on basic WM tasks at around 11 or 12 years of age and on complex span tasks even later (Gathercole, 1999), it is tempting to speculate that maturation of LTM–STM support systems may contribute to performance of basic and complex STM tasks.
References Alloway, T. P., Gathercole, S. E., & Pickering, S. J. (2006). Verbal and visuospatial short-term and working memory in children: Are they separable? Child Development, 77, 1698–1716. Alloway, T. P., Gathercole, S. E., Willis, C., & Adams, A. M. (2004). A structural analysis of working memory and related cognitive skills in young children. Journal of Experimental Child Psychology, 87, 85–106. Baddeley, A. D. (1986). Working memory. Oxford, UK: Oxford University Press. Baddeley, A. (2000). The episodic buffer: A new component of working memory? Trends in Cognitive Sciences, 4, 417–423. Baddeley, A. D., Allen, R. J., & Hitch, G. J. (2011). Binding in visual working memory: The role of the episodic buffer. Neuropsychologia, 49, 1393–1400. Baddeley, A. D., & Hitch, G. J. (1974). Working memory. Psychology of Learning and Motivation, 8, 47–89. Bayliss, D. M., Jarrold, C., Gunn, D. M., & Baddeley, A. D. (2003). The complexities of complex span: Explaining individual differences in working memory in children and adults. Journal of Experimental Psychology: General, 132, 71–92. Brown, L. A., & Wesley, R. W. (2013). Visual working memory is enhanced by mixed strategy use and semantic coding. Journal of Cognitive Psychology, 25, 328–338. Cowan, N. (2005). Working memory capacity. New York: Psychology Press. Cowan, N., Naveh-Benjamin, M., Kilb, A., & Saults, J. S. (2006a). Life-span development of visual working memory: When is feature binding important? Developmental Psychology, 42, 1089–1102. Cowan, N., Saults, J. S., & Morey, C. C. (2006b). Development of working memory for verbal–spatial associations. Journal of Memory and Language, 55, 274–289. Cowan, N., Towse, J. N., Hamilton, Z., Saults, J. S., Elliott, E. M., Lacey, J. F., et al (2003). Children’s working-memory processes: A response-timing analysis. Journal of Experimental Psychology: General, 132, 113–132. Darling, S., Allen, R. J., Havelka, J., Campbell, A., & Rattray, E. (2012). Visuospatial bootstrapping: Long-term memory representations are necessary for implicit binding of verbal and visuospatial working memory. Psychonomic Bulletin & Review, 19, 258–263. Darling, S., & Havelka, J. (2010). Visuospatial bootstrapping: Evidence for binding of verbal and spatial information in working memory. Quarterly Journal of Experimental Psychology, 63, 239–245. Engle, R. W. (2010). Role of working-memory capacity in cognitive control. Current Anthropology, 51, S17–S26. Engle, R. W., Tuholski, S. W., Laughlin, J. E., & Conway, A. R. A. (1999). Working memory, short-term memory, and general fluid intelligence: A latent-variable approach. Journal of Experimental Psychology: General, 128, 309–331. Gathercole, S. E. (1999). Cognitive approaches to the development of short-term memory. Trends in Cognitive Sciences, 3, 410–419. Gathercole, S. E., & Hitch, G. J. (1993). Developmental changes in short-term memory: A revised working memory perspective. In A. F. Collins, S. E. Gathercole, M. A. Conway, & P. E. Morris (Eds.), Theories of memory (pp. 189–209). Hove, UK: Lawrence Erlbaum. Hitch, G. J., Halliday, M. S., Dodd, A., & Littler, J. E. (1989). Development of rehearsal in short-term memory: Differences between pictorial and spoken stimuli. British Journal of Developmental Psychology, 7, 347–362.
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
119
Kapikian, A., & Briscoe, J. (2012). Semantic binding, not attentional control, generates coherent global structure in children’s narrative memory. Journal of Cognitive Psychology, 24, 751–764. Mate, J., Allen, R. J., & Baqués, J. (2012). What you say matters: Exploring visual–verbal interactions in visual working memory. Quarterly Journal of Experimental Psychology, 65, 395–400. Morey, C. C., & Cowan, N. (2004). When visual and verbal memories compete: Evidence of cross-domain limits in working memory. Psychonomic Bulletin & Review, 11, 296–301. Morey, C. C., & Cowan, N. (2005). When do visual and verbal memories conflict? The importance of working-memory load and retrieval. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31, 703–713. Pickering, S. J., Gathercole, S. E., & Peaker, S. M. (1998). Verbal and visuospatial short-term memory in children: Evidence for common and distinct mechanisms. Memory & Cognition, 26, 1117–1130. Siegel, L. S. (1994). Working memory and reading: A life-span perspective. International Journal of Behavioral Development, 17, 109–124. Towse, J. N., Cowan, N., Hitch, G. J., & Horton, N. J. (2008). The recall of information from working memory: Insights from behavioural and chronometric perspectives. Experimental Psychology, 55, 371–383. Towse, J. N., Hitch, G. J., Horton, N., & Harvey, K. (2010). Synergies between processing and memory in children’s reading span. Developmental Science, 13, 779–789. Unsworth, N., & Engle, R. W. (2007). The nature of individual differences in working memory capacity: Active maintenance in primary memory and controlled search from secondary memory. Psychological Review, 114, 104–132.
Contents lists available at ScienceDirect
Journal of Experimental Child Psychology journal homepage: www.elsevier.com/locate/jecp
Brief Report
Visuospatial bootstrapping: Implicit binding of verbal working memory to visuospatial representations in children and adults Stephen Darling a,⇑, Mary-Jane Parker a, Karen E. Goodall a, Jelena Havelka b, Richard J. Allen b a b
Division of Psychology and Sociology, Queen Margaret University, Edinburgh EH21 6UU, UK Institute for Psychological Sciences, University of Leeds, Leeds LS2 9JT, UK
a r t i c l e
i n f o
Article history: Received 13 May 2013 Revised 8 October 2013 Available online 25 November 2013 Keywords: Working memory Verbal memory Spatial memory Visual memory Bootstrapping Episodic Buffer
a b s t r a c t When participants carry out visually presented digit serial recall, their performance is better if they are given the opportunity to encode extra visuospatial information at encoding—a phenomenon that has been termed visuospatial bootstrapping. This bootstrapping is the result of integration of information from different modalityspecific short-term memory systems and visuospatial knowledge in long term memory, and it can be understood in the context of recent models of working memory that address multimodal binding (e.g., models incorporating an episodic buffer). Here we report a cross-sectional developmental study that demonstrated visuospatial bootstrapping in adults (n = 18) and 9-year-old children (n = 15) but not in 6-year-old children (n = 18). This is the first developmental study addressing visuospatial bootstrapping, and results demonstrate that the developmental trajectory of bootstrapping is different from that of basic verbal and visuospatial working memory. This pattern suggests that bootstrapping (and hence integrative functions such as those associated with the episodic buffer) emerge independent of the development of basic working memory slave systems during childhood. Ó 2013 Elsevier Inc. All rights reserved.
⇑ Corresponding author. E-mail address: [email protected] (S. Darling). 0022-0965/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jecp.2013.10.004
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
113
Introduction There is considerable evidence that verbal and spatial information is handled by different shortterm memory processes (see Baddeley, 2000). This sits alongside considerable evidence of substantial shared variance within measures that involve the processing, rather than the simple storage, of remembered items (Engle, Tuholski, Laughlin, & Conway, 1999). Evidence of both types underlies the development of theoretical approaches that clearly differentiate process-dependent memory systems from storage systems where information is stored but not manipulated. One such approach is the working memory model (Baddeley, 1986; Baddeley & Hitch, 1974), which proposed the interaction of a central executive (CE) processing component alongside modality-specific passive storage systems (the phonological loop and visuospatial sketchpad). An alternative model (Engle, 2010; Engle et al., 1999) proposed that storage functions labeled as short-term memory (STM) may be supported by distinct subsystems, whereas processing functions labeled as working memory (WM) require a common integrated set of processes associated with a substantial shared variance. Even theoretical approaches that deemphasize modality-specific systems (Cowan, 2005) still accommodate the idea that basic storage processes may be differentiable on the basis of task demands. Despite this, STM subsystems are not entirely discrete and information stored in separate modalities can be linked (Mate, Allen, & Baqués, 2012: Morey & Cowan, 2004, 2005; see Baddeley, 2000, for a review of earlier evidence). Recently, the integration of visuospatial, verbal, and long-term memory has been demonstrated in verbal serial recall. When to-be-remembered digits were presented in a familiar visuospatial array—the standard numeric keyboard used in mobile telephones—memory was facilitated compared with when digits were presented in a single location (Darling & Havelka, 2010). Participants were asked to attend to only a single stimulus dimension (digit serial recall), but the performance improvement associated with keypad presentation indicated that they were able to extract visuospatial information and integrate it with the verbal material. Hence, this pattern was described as visuospatial bootstrapping (Darling & Havelka, 2010) because verbal memory performance was bootstrapped by the integration of redundant visuospatial information that was present in the keypad condition and not in the single item condition. A subsequent study has replicated visuospatial bootstrapping (Darling, Allen, Havelka, Campbell, & Rattray, 2012), also showing that the availability of a compatible representation in long-term memory (LTM) is necessary for observing bootstrapping. Visuospatial bootstrapping represents an effect where nonverbal memory processes interact with verbal memory; a parallel result has recently been reported, where visuospatial memory is differentially impaired by concrete and abstract verbal load (Mate et al., 2012). Taken together, these patterns are inconsistent with the possibility that WM subsystems are fully independent of each other. They also suggest that interactions between WM systems take place on an implicit level—or at least without explicit instruction to combine information from different sources. The episodic buffer (Baddeley, 2000) was proposed as an additional WM component capable of maintaining cross-modality bindings between information in LTM and STM, with a role in the encoding of specific episodes. Unlike the CE and other theoretical mechanisms targeted at understanding how information in WM is integrated (e.g., the focus of attention described by Cowan, 2005), the episodic buffer is now thought to require neither executive nor attentional resources to function, instead operating in an efficient, automatic, and rule-governed manner (Baddeley, Allen, & Hitch, 2011). Given this, it seems plausible that the potentially implicit visuospatial bootstrapping data might be explained with reference to the episodic buffer (Darling et al., 2012). Aside from the episodic buffer model, there are alternative accounts of WM that can potentially offer a theoretical account of visuospatial bootstrapping. Unsworth and Engle (2007) proposed that effective WM performance entails maintenance in primary memory alongside effective search of secondary memory, and it is plausible that the bootstrapping effect occurs because keypad displays facilitate this search of secondary memory due to the provision of additional cues. There is also robust evidence that reconstructive processes facilitate temporary memory (Cowan et al., 2003; Towse, Cowan, Hitch, & Horton, 2008; Towse, Hitch, Horton, & Harvey, 2010). Given that WM for spatial configurations is certainly affected by aspects of long-term spatial memory (Brown & Wesley, 2013), it is plausible that richer spatial arrays at presentation lead to more effective reconstructions.
114
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
In the current study, we asked whether there was evidence of multimodal WM processing in samples of children and adults by assessing evidence for visuospatial bootstrapping across the developmental time course. Although separate verbal and visuospatial short-term memory systems are known to be observable in children (e.g. Pickering, Gathercole, & Peaker, 1998), subsequent research (e.g., Alloway, Gathercole, & Pickering, 2006; Bayliss, Jarrold, Gunn, & Baddeley, 2003) has further suggested that storage functions of verbal and visuospatial memory systems are easily segregable, whereas processing functions are more unified, consistent with the idea that STM is differentiated in children, whereas WM processing functions are more easily understood in a domain-general framework. Research specifically probing the episodic buffer in children (Alloway, Gathercole, Willis, & Adams, 2004) has suggested its emergence as early as 4 years of age (see also Kapikian & Briscoe, 2012). However, the task used to assess it (sentence recall) was restricted to the verbal modality and, thus, does not fully characterize the multimodal aspects of the proposed episodic buffer (Baddeley et al., 2011). Other studies of multimodal WM in children have examined binding between color and location (Cowan, Naveh-Benjamin, Kilb, & Saults, 2006a) and names and locations (Cowan, Saults, & Morey, 2006b) in 8-year-old children and found reduced accuracy relative to adults, but these studies did not consider how information from different domains is integrated with long-term knowledge. The specific benefits of assessing visuospatial bootstrapping with the goal of understanding the integration of information across WM components are threefold. First, bootstrapping is superficially an implicit phenomenon and, hence, should not recruit central attentional processes. Second, there is no requirement to manipulate any information; participants should benefit from keypad-type displays by integrating visuospatial information, verbal information, and LTM knowledge to form an STM trace that is more resilient than an unintegrated one, but there is no need to judge, reorder, or process the information to obtain this benefit. Third, any bootstrapping effect is undeniably multimodal, bridging the two best-documented modality-specific STM systems. Although there is also evidence of a gradual capacity increase in WM as age increases (Gathercole, 1999), specific critical periods of change in relationships between modality-specific STM systems might be expected to occur somewhere after 6 or 7 years of age as visuospatial STM (arguably) becomes differentiated from the CE (Alloway et al., 2006) and as verbal rehearsal becomes available as a mnemonic method (Gathercole & Hitch, 1993). We recruited groups of 6- and 9-year-old children to span these changes. We also recruited a sample of young adults to serve as a comparison sample. To minimize the complicating effects of verbal span improvement as age increases (Alloway et al., 2006; Pickering et al., 1998), participants initially had their capacity assessed in a standard digit span task; subsequently, they were tested under different display conditions at their span. Three display configurations were used. The Single display involved presenting all to-be-remembered digits in a single location. The Typical display presented a representation of a standard numeric keypad within which the to-be-remembered digits were indicated by highlighting numbers in the display. The Novel display was similar but used a grid where digits were located randomly in order to assess the need for a preexisting LTM representation (see Fig. 1 in Darling et al., 2012, p. 260, for illustrations of these displays).
Method Participants Two groups of 15 children attending an after-school club in Peebles, United Kingdom, took part: a group of 6-year-olds (mean age = 80 months, SD = 4, 6 girls and 9 boys) and a group of 9-year-olds (mean age = 111 months, SD = 5, 8 girls and 7 boys). The parents of all child participants gave informed consent in writing, and the children themselves also gave verbal consent. None of the children had any known additional learning needs or sensory impairments. Child participants took part in a quiet room in their after-school club. An adult comparison sample was recruited among students and staff of Queen Margaret University. The sample size was 18 (mean age = 25.17 years, SD = 8.87). All adult participants gave informed
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
115
1 Single Typical Novel
Proporon correct
0.8
0.6
0.4
0.2
0 Younger children (~6Y)
Older children (~9Y)
Adults
Age group
Fig. 1. Memory performance (proportion of trials remembered correctly) across display condition and age group. Error bars represent standard errors of the mean.
consent in writing. Adult participants took part in a laboratory at the university. This research was approved by the research ethics committee of the university. Design On the first day, participants carried out a quick assessment of their knowledge of the standard telephone keypad. Following this, their digit serial recall span was assessed using a single digit display. On 3 subsequent days, participants took part in sets of trials in each of three display conditions: Single, Typical, and Novel. In these different conditions, participants were tested with digit sequence lengths of the maximum length correctly recalled in the span task. The order of these conditions was counterbalanced. Participants in the adult condition took part in all conditions on the same day and omitted the keypad knowledge check. Materials and procedure Keypad knowledge assessment Children were shown a blank grid containing 10 empty squares in the equivalent positions to the keys on a standard numeric telephone keypad. They were asked to write in the digits 0 to 9 in the correct positions that they appeared on a telephone. Experimental tasks Stimuli were presented using a standard personal computer. The display was set to a resolution of 1024 768 pixels. Each trial was initiated by participants pressing a key, and then a fixation cross was displayed for 500 ms, followed by a blank screen interval of 250 ms and then by digit sequence presentation. Digits were always presented in Arial font, point size 18, and presented centrally within squares of side 60 px. Once all digits in a given trial had been shown, the screen cleared for a 1000ms retention interval and then the message ‘‘RECALL’’ was presented in the middle of the screen. At this point, participants attempted to repeat the digits aloud in the correct order. Span assessment To assess maximum span, participants were asked to remember random sequences of digits using Single displays; digits were presented in the middle of the screen, inside a square with a green
116
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
background. Each digit was shown for 1000 ms, and between each digit in the sequence there was a 250-ms interval during which the screen was blank. Trials were presented in pairs. In the first 2 trials, participants were shown and then recalled a single digit. If they correctly recalled at least 1 trial, the sequence was increased in size to two items and a further pair of trials was shown. If participants recalled at least one sequence correctly, then a further pair was presented with a sequence of three. This incrementing of sequence length by one item was repeated until participants failed to recall any sequences correctly at a given sequence length. Span was defined as the maximum sequence length at which participants could recall at least one sequence correctly. Performance at span: Single digit display Participants undertook a set of 24 trials. In each trial, they needed to remember a random sequence of digits. Sequence size was set at individual span. Digits were presented in a single green square in the middle of the screen. Timing parameters were the same as in the span task. Performance at span: Typical keypad display In the Typical condition, the digits 0 to 9 were presented in the same array used in a traditional telephone keypad, aligned centrally on the screen. Detailed dimensions of the array were the same as those used by Darling and colleagues (2012). The digit sequences were presented by highlighting individual digit backgrounds in green. Timing of the highlighting of the digits followed the timing used in the Single condition, with each digit being highlighted for 1000 ms and a blank screen interval of 250 ms being imposed between sequence items. Performance at span: Novel static keypad display The Novel condition was identical to the Typical condition except that digits were presented in nontypical locations so that on beginning the study the relationship between digits and locations was unfamiliar. Digit locations were consistent throughout all trials. Results Keypad knowledge Of the younger children, 7 correctly completed the grid, whereas 8 did not. In the older group, 13 responded correctly and 2 did not. This association between age and successful completion of the keypad task was significant, v2(1) = 5.40, p = .020. Span The mean span was 4.47 items (SD = 0.64, min = 3, max = 5) for the 6-year-old group, 5.13 items (SD = 0.74, min = 3, max = 6) for the 9-year-old group, and 6.72 items (SD = 1.32, min = 5, max = 9) for the young adult participants. The main effect of digit span was significant, F(2, 45) = 22.45, p < .001, g2p = .51. Individual t tests indicated that the adults displayed significantly higher digit span than the 9-year-olds, t(31) = 4.14, p < .001, d = 1.49, whereas span was higher in the 9-year-olds compared with the 6-year-olds, t(31) = 2.63, p = .01, d = 0.95. Effects of age and display configurations on memory The proportions of the 24 trials in the three display conditions that were correctly recalled are summarized in Fig. 1. Data were assessed using a mixed-design 3 (Age Group) 3 (Display Condition) analysis of variance (ANOVA), indicating no significant main effect of age group, F(2, 45) = 1.53, p = .228, g2p = .06. There was a nonsignificant trend toward a main effect of display, F(2, 90) = 2.57, p = .082, g2p = .05. This was qualified by a significant interaction between age group and display, F(4, 90) = 4.37, p = .003, g2p = .16. This interaction is shown in Fig. 1; the 9-year-old and adult participants
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
117
remembered more items in the Typical condition than in either the Single or Novel condition, whereas this was not the case for the 6-year-old group. The simple main effect of display was significant in adult participants, F(2, 44) = 3.94, p = .027, g2p = .15. Multiple comparisons showed that digit memory in the Typical condition was significantly better than that in both the Single (p = .017) and Novel (p = .016) conditions, and the Novel and Single conditions did not differ significantly (p = .905). The simple main effect of display was also significant for the 9-year-old participants, F(2, 44) = 4.93, p = .012, g2p = .18); performance in the Typical condition was significantly better than that in both the Single (p = .024) and Novel (p = .004) conditions, and the Novel and Single conditions did not differ significantly (p = .559). The simple main effect of display in the 6-year-old participants did not reach significance, F(2, 44) = 2.41, p = .102, g2p = .10. To establish whether there were any differences in the pattern of performance between the 9-year-olds and adults, a 2 3 mixed-design ANOVA was conducted on the 9-year-old and adult groups only. There was a significant main effect of display, F(2, 62) = 10.81, p < .001, g2p = .26, but neither the main effect of age, F(1, 31) = 2.63, p = .115, g2p = .08, nor the interaction, F(2, 62) = 0.23, p = .80, g2p = .01, was significant, a pattern indicative of comparable performance with no significant differences between the 9-year-old and adult participants.
Discussion In the current study, 9-year-olds showed evidence of visuospatial bootstrapping that was effectively the same as that observed in adult participants. In contrast, at 6 years of age there was no evidence of a bootstrapping effect. Because of the individual titration of difficulty used in this study, it is unlikely that overall lower verbal span in 6-year-olds could be the reason behind the absence of any bootstrapping effect in that group—an inference that is supported by the lack of a main effect of age group in the main experimental conditions. A strong inference can be drawn that the bootstrapping effect was not a simple result of development of basic WM, either verbal or visuospatial. Capacity, as assessed in the initial digit span task, increased with age; a gradual increase was evident among 6-year-olds, 9-year-olds, and adults. This is typical of patterns seen in developmental studies of memory capacity in both visual and verbal domains where capacity increases are asymptotic later in adolescence (see Gathercole, 1999, for a summary). This contrasted with the development of bootstrapping, which was completely absent in 6year-olds but was equivalent in 9-year-olds and adults, suggesting that the cognitive resources supporting bootstrapping are maturationally discrete from those supporting digit span and visuospatial WM. The forgoing discussion raises the issue of which mechanisms may develop between 6 and 9 years of age to facilitate bootstrapping. The current data are equivocal between several reasonable possibilities. One candidate is consolidation of LTM learning; LTM representations of the typical keypad are critical in order to demonstrate a bootstrapping effect over 24 trials in adults (Darling et al., 2012), so it is possible that acquisition of a reliable representation of the typical keypad may drive the emergence of bootstrapping in older children. This could be due to acquisition of new learning regarding the typical keypad and its importance, or it may be a consequence of increased efficiency of LTM systems themselves over development; both possibilities are equally consistent with the observation that older children were better at repopulating the blank keypad grid from LTM in the keypad assessment task. Darling and colleagues (2012) observed that 24 trials was insufficient to develop a sufficiently robust representation of a novel keypad for a beneficial bootstrapping effect to emerge (in an adult sample), suggesting that such LTM representations needed to be reasonably well established and could not be learned satisfactorily over the course of a single set of trials. A second possibility is that the ability to integrate information between modality-specific cognitive subsystems and LTM into a robust representation develops between 6 and 9 years of age. This kind of process is held to be the role of the episodic buffer (Baddeley et al., 2011), and the current data also fit well with such an explanation. The assertion that the bootstrapping task invokes multimodal representations that are implicit remains speculative, and to be certain that there are no central attention resources implicated in the bootstrapping task requires further work beyond the scope of this study. A third possible
118
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
explanation is that the bootstrapping pattern is a consequence of explicit strategic use of visuospatial resources and that 6-year-olds were unable to explicitly select the (better) visuospatial strategy. The literature suggests some grounds to suspect that this might happen; children older than 7 years spontaneously convert visually presented stimuli into a verbal format, whereas younger children do not (Hitch, Halliday, Dodd, & Littler, 1989). The emergence of bootstrapping, therefore, may be reflective of the development of metacognitive skill. Relatedly, it may be that improvements in bootstrapping reflect improvements in controlled searching of (visuospatial) secondary memory (Unsworth & Engle, 2007). Settling which of these competing explanations is most satisfactory is an important goal for future research addressing the theoretical underpinnings of bootstrapping. The current observations are both novel and noteworthy in that they represent the first evidence showing visuospatial bootstrapping in children. The observed data reflect a pattern of integration of spatial information to assist STM that is clearly cross-modal. The task demands do not explicitly require multimodal representation, and neither do they explicitly require processing or manipulation of information, yet the results could have occurred only if multimodal connections had been established. Moreover, the data demonstrate that the bootstrapping effect seen in 9-year-olds was comparable to that seen in adults but was not observed in 6-year-olds; hence, the bootstrapping effect is relatively independent of the development of basic WM capacity. It is also noteworthy that the maturational trajectory of visuospatial bootstrapping is very different from that of complex span (which continues to develop through adolescence; e.g., Siegel, 1994; see also Gathercole, 1999). In fact, given that the maturation of bootstrapping appears to be complete by around 9 years of age, and that this precedes asymptotic performance on basic WM tasks at around 11 or 12 years of age and on complex span tasks even later (Gathercole, 1999), it is tempting to speculate that maturation of LTM–STM support systems may contribute to performance of basic and complex STM tasks.
References Alloway, T. P., Gathercole, S. E., & Pickering, S. J. (2006). Verbal and visuospatial short-term and working memory in children: Are they separable? Child Development, 77, 1698–1716. Alloway, T. P., Gathercole, S. E., Willis, C., & Adams, A. M. (2004). A structural analysis of working memory and related cognitive skills in young children. Journal of Experimental Child Psychology, 87, 85–106. Baddeley, A. D. (1986). Working memory. Oxford, UK: Oxford University Press. Baddeley, A. (2000). The episodic buffer: A new component of working memory? Trends in Cognitive Sciences, 4, 417–423. Baddeley, A. D., Allen, R. J., & Hitch, G. J. (2011). Binding in visual working memory: The role of the episodic buffer. Neuropsychologia, 49, 1393–1400. Baddeley, A. D., & Hitch, G. J. (1974). Working memory. Psychology of Learning and Motivation, 8, 47–89. Bayliss, D. M., Jarrold, C., Gunn, D. M., & Baddeley, A. D. (2003). The complexities of complex span: Explaining individual differences in working memory in children and adults. Journal of Experimental Psychology: General, 132, 71–92. Brown, L. A., & Wesley, R. W. (2013). Visual working memory is enhanced by mixed strategy use and semantic coding. Journal of Cognitive Psychology, 25, 328–338. Cowan, N. (2005). Working memory capacity. New York: Psychology Press. Cowan, N., Naveh-Benjamin, M., Kilb, A., & Saults, J. S. (2006a). Life-span development of visual working memory: When is feature binding important? Developmental Psychology, 42, 1089–1102. Cowan, N., Saults, J. S., & Morey, C. C. (2006b). Development of working memory for verbal–spatial associations. Journal of Memory and Language, 55, 274–289. Cowan, N., Towse, J. N., Hamilton, Z., Saults, J. S., Elliott, E. M., Lacey, J. F., et al (2003). Children’s working-memory processes: A response-timing analysis. Journal of Experimental Psychology: General, 132, 113–132. Darling, S., Allen, R. J., Havelka, J., Campbell, A., & Rattray, E. (2012). Visuospatial bootstrapping: Long-term memory representations are necessary for implicit binding of verbal and visuospatial working memory. Psychonomic Bulletin & Review, 19, 258–263. Darling, S., & Havelka, J. (2010). Visuospatial bootstrapping: Evidence for binding of verbal and spatial information in working memory. Quarterly Journal of Experimental Psychology, 63, 239–245. Engle, R. W. (2010). Role of working-memory capacity in cognitive control. Current Anthropology, 51, S17–S26. Engle, R. W., Tuholski, S. W., Laughlin, J. E., & Conway, A. R. A. (1999). Working memory, short-term memory, and general fluid intelligence: A latent-variable approach. Journal of Experimental Psychology: General, 128, 309–331. Gathercole, S. E. (1999). Cognitive approaches to the development of short-term memory. Trends in Cognitive Sciences, 3, 410–419. Gathercole, S. E., & Hitch, G. J. (1993). Developmental changes in short-term memory: A revised working memory perspective. In A. F. Collins, S. E. Gathercole, M. A. Conway, & P. E. Morris (Eds.), Theories of memory (pp. 189–209). Hove, UK: Lawrence Erlbaum. Hitch, G. J., Halliday, M. S., Dodd, A., & Littler, J. E. (1989). Development of rehearsal in short-term memory: Differences between pictorial and spoken stimuli. British Journal of Developmental Psychology, 7, 347–362.
S. Darling et al. / Journal of Experimental Child Psychology 119 (2014) 112–119
119
Kapikian, A., & Briscoe, J. (2012). Semantic binding, not attentional control, generates coherent global structure in children’s narrative memory. Journal of Cognitive Psychology, 24, 751–764. Mate, J., Allen, R. J., & Baqués, J. (2012). What you say matters: Exploring visual–verbal interactions in visual working memory. Quarterly Journal of Experimental Psychology, 65, 395–400. Morey, C. C., & Cowan, N. (2004). When visual and verbal memories compete: Evidence of cross-domain limits in working memory. Psychonomic Bulletin & Review, 11, 296–301. Morey, C. C., & Cowan, N. (2005). When do visual and verbal memories conflict? The importance of working-memory load and retrieval. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31, 703–713. Pickering, S. J., Gathercole, S. E., & Peaker, S. M. (1998). Verbal and visuospatial short-term memory in children: Evidence for common and distinct mechanisms. Memory & Cognition, 26, 1117–1130. Siegel, L. S. (1994). Working memory and reading: A life-span perspective. International Journal of Behavioral Development, 17, 109–124. Towse, J. N., Cowan, N., Hitch, G. J., & Horton, N. J. (2008). The recall of information from working memory: Insights from behavioural and chronometric perspectives. Experimental Psychology, 55, 371–383. Towse, J. N., Hitch, G. J., Horton, N., & Harvey, K. (2010). Synergies between processing and memory in children’s reading span. Developmental Science, 13, 779–789. Unsworth, N., & Engle, R. W. (2007). The nature of individual differences in working memory capacity: Active maintenance in primary memory and controlled search from secondary memory. Psychological Review, 114, 104–132.
