Multimedia learning and individual differences: Mediating the effects of working memory capacity with segmentation

Research in multimedia learning lacks an emphasis on individual difference variables, such as working memory capacity (WMC). The effects of WMC and the segmentation of multimedia instruction were examined by assessing the recall and application of low (n = 66) and high (n = 67) working memory capacity students randomly assigned to either a segmented instruction (SI) or non-segmented instruction (NSI) version of a multimedia tutorial on historical inquiry. WMC was found to have a significant, positive effect on participants’ recall and application scores; however, the use of segmentation mediated the effects of WMC to allow learners with lower WMC to recall and apply equal to those with higher WMC. Introduction Recent research in multimedia learning has tended to focus on multimedia design, such as the modality and segmentation principles (eg, Mayer & Chandler, 2001; Mayer & Moreno, 1998); cognitive structures, such as cognitive load and dual coding (eg, Paas, Renkl & Sweller, 2003; Sweller, 1994); scaffolding components, such as pedagogical agents and help support (eg, Craig, Driscoll & Cholson, 2004; Kim, Baylor & PALS Group, 2006) and delivery systems, such as the Web and mobile devices (eg, Cole & Todd, 2003; Wei, Chen, Wang & Li, 2007). Less well researched is the relationship between individual difference variables, such as working memory capacity (WMC; British Journal of Educational Technology (2008) doi:10.1111/j.1467-8535.2008.00848.x © 2008 The Authors. Journal compilation © 2008 British Educational Communications and Technology Agency. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA. Unsworth & Engle, 2007), and the design, structure, scaffolding and delivery of multimedia learning. For example, WMC has been demonstrated to vary significantly from individual to individual (Rosen & Engle, 1997) and is positively related to higher-order cognitive tasks such as reading comprehension (Daneman & Carpenter, 1980), attentional control (Kane, Bleckley, Conway & Engle, 2001), general fluid intelligence (Engle, Tuholski, Laughlin & Conway, 1999) and mathematical performance (Ashcraft & Kirk, 2001). Is there a relationship between WMC and learning in multimedia environments? One area in which higher-order cognitive tasks is receiving significant attention is the domain of multimedia learning. Learning in multimedia environments has been studied extensively, resulting in the cognitive theory of multimedia learning and a series of design principles (see Mayer, 2001, 2005). However, simply constructing sound multimedia environments is insufficient for learning to occur (Kozma, 1994). Rather, learners must attend to these multiple forms of media, create conceptual knowledge representations and integrate these knowledge representations in order to learn and build effective mental models (Schnotz & Bannert, 2003). Thus, the effectiveness of these principles depends on both instructional designers creating sound multimedia learning environments and individual learners actively engaging in constructing understanding. Unfortunately, despite this interaction between the multimedia instructional environment and the individual learner, little attention has been given to the role individual differences have on Mayer’s (2005) multimedia design principles. The present study examines this relationship, specifically, the relationship between segmentation and WMC in multimedia learning. Multimedia learning and the segmentation principle The segmentation principle simply states that a multimedia tutorial that provides the user with pacing control, through use of a Start/Stop button or Continue button, will result in greater learning than a tutorial that plays from beginning to end (Mayer & Chandler, 2001). The rationale for the segmentation principle is that this pacing control provides the learner with the opportunity to stop the flow of information when necessary. In stopping the flow of information, the learner is less likely to be overloaded by information, resulting in degraded learning, and is more likely to be able to process the information more deeply, resulting in enhanced learning (see Mayer, 2005). Mayer, Dow and Mayer (2003) investigated segmentation by having students engage in a 20-minute multimedia tutorial addressing the working of an electric motor in segmented (S) and nonsegmented (NS) versions. Mayer et al found that students who experienced the S version had better transfer of information than students who experienced the NS version. Similarly, Mayer and Chandler (2001) explored the segmentation principle by creating two versions, an S version and an NS version, of a 140 second multimedia tutorial addressing the cause of lightning. Mayer and Chandler had students experience both versions, sequentially, either S-NS or NS-S (Exp 1). Mayer and 2 British Journal of Educational Technology © 2008 The Authors. Journal compilation © 2008 British Educational Communications and Technology Agency. Chandler found that the S-NS group performed better on a transfer task, but not on a recall task, than the NS-S group. Mayer and Chandler attributed the superior transfer performance of the S-NS group to participants avoiding cognitive overload and being able to build models of the component parts of the lightning cause-and-effect relationship during this first engagement (S). During the second engagement (NS), the participants were then able to connect and organise the component parts. The results of Mayer and Chandler (2001) and Mayer et al, (2003) provide partial support for segmentation, benefiting transfer but not recall. This partial support of segmentation is in agreement with a large body of research addressing the broader concept of learner control. Learner control includes not only pacing control (segmentation), but also control over the inclusion of content, the depth of content experiences, the order of content presentation, the amount of practice and the type of feedback (Pollock & Sullivan, 1990). When examining the learner control literature addressing specifically learner control of pacing, the results tend to be conflicting (see Aly, Elen & Willems, 2005; Dalton, 1990). WMC WMC represents an individual’s ability to simultaneously (1) process a primary task in working memory, (2) maintain relevant information regarding the primary task in working memory and (3) access and retrieve relevant information regarding the primary task from long-term memory—especially in the presence of distraction (Unsworth & Engle, 2007). This concept of WMC moves beyond more traditional measures of working memory storage capacity (see Miller, 1956) to include both storage and processing capacity (see Daneman & Carpenter, 1980). This measure of storage and processing capacity has been interpreted as an assessment of attentional control, the ability to control the processing and maintenance of information in working memory, especially in the presence of internal (eg, thoughts, drives and feelings) or external (eg, talking, music and motion) distractions taxing the attentional system (Unsworth & Engle, 2007). The literature on WMC provides evidence that high WMC benefits performance on complex mental tasks including general fluid intelligence (Conway, Cowan, Bunting, Therriault & Minkoff, 2002; Kane et al, 2001), long-term memory activation (Cantor & Engle, 1993), attentional control (Kane et al, 2001), resistance to proactive interference (Kane & Engle, 2000), primary memory maintenance and secondary memory search (Unsworth & Engle, 2007) and resistance to goal neglect (Kane & Engle, 2003). Beyond these cognitive construct effects, individual differences have been indicated in a variety of cognitive performance measures; that is, individuals with high WMC have been demonstrated to perform better than individuals with low WMC in reading comprehension (Daneman & Carpenter, 1980), language comprehension (Just & Carpenter, 1992), vocabulary learning (Daneman & Green, 1986), reasoning (cf. Buehner, Krumm & Pick, 2005; Conway et al, 2002) computer language learning (Shute, 1991), lecture note taking (Kiewra & Benton, 1988), Scholastic Aptitude Test performance (Turner & Engle, 1989), mnemonic strategy effectiveness (Gaultney, Kipp & Kirk, 2005) Multimedia learning and individual differences 3 © 2008 The Authors. Journal compilation © 2008 British Educational Communications and Technology Agency. and story telling (Pratt, Boyes, Robins & Manchester, 1989). This research has demonstrated a strong, positive relationship between variations in WMC and variations in complex cognitive task performance. One domain of complex cognitive tasks that has seen little research related to WMC is multimedia learning. The examination of individual differences in WMC on multimedia learning is of interest as both WMC and multimedia learning are influenced by attentional control (see Mayer, 2001, 2005). Specifically, Mayer (2001) describes multimedia learning as based on three essential processes requiring attentional control: selecting relevant information, organising relevant information and integrating relevant information. Each of these processes requires attentional control in much that same way as WMC—the learner must (1) attend to and maintain the goal of the learning episode; (2) attend to the available information; (3) select the information relevant to the learning goal from the available information; (4) organise the selected information based on the goal of the learning episode; (5) maintain the learning goal and organised information in working memory while retrieving necessary information from long-term memory; and (6) integrate the working memory and long-term memory information to achieve the learning goal. Given this potential overlap between the structures and processes of WMC and multimedia learning, might learners with low WMC find it difficult to engage in the selecting, organising and integrating processes necessary for learning? Also, if learners with low WMC are indeed having difficulty selecting, organising and integrating the flow of information, might it be beneficial to provide these learners with control of the flow of the information? The purpose of this research is to examine the effects of WMC and segmentation on learning in a multimedia instructional environment. Methodology Previous studies of segmenting multimedia instruction have indicated that dividing multimedia instruction into short, user-controlled segments leads to increased recall and transfer of the multimedia content (Mayer & Chandler, 2001; Mayer & Moreno, 2003). These studies, however, did not take into account individual difference variables that may mediate learner performance. Thus, the purpose of this study is to assess the effects of WMC on content recall and application resulting from S and NS multimediabased instruction.