Liesbeth Kester a,*, Paul Kirschner a,b, Gemma Corbalan a
complex individual or collaborative learning. This introduction provides the context for the issueand a short overview of the contributions.
mean? To best introduce this issue, three elements of the title rst need to be claried,
* Corresponding author. Tel.: +31 45 5762 428; fax: +31 45 5762 802.E-mail address: firstname.lastname@example.org (L. Kester).
Computers in Human Behavior 23 (2007) 10471054
Computers inHuman Behavior0747-5632/$ - see front matter 2006 Elsevier Ltd. All rights reserved. 2006 Elsevier Ltd. All rights reserved.
Keywords: Electronic learning environments; Whole-task practice; Self-regulated learning; Collaborative learn-ing
This opening article in the special issue of Computers in Human Behavior is about learn-ing in powerful learning environments that are electronic in nature. But what does thata Open University of the Netherlands, Educational Technology Expertise Center,
P.O. Box 2960, 6401 DL Heerlen, The Netherlandsb Utrecht University, Research Centre Learning in Interaction, P.O. Box 80140, 3508 TC Utrecht, The Netherlands
Available online 15 November 2006
This special issue reects current developments in instructional design for powerful electroniclearning environments. It presents a compilation of contributions to a combined special interestgroup (SIG) meeting (2006) of Instructional Design and Learning and Instruction with Computers.Both SIGs are part of the European Association for Research on Learning and Instruction(EARLI). The SIG-meeting focused on the design of powerful electronic learning environmentsfor complex learning. The articles in this issue describe how to design support to help learners duringDesigning support to facilitate learning inpowerful electronic learning environmentsdoi:10.1016/j.chb.2006.10.001
1048 L. Kester et al. / Computers in Human Behavior 23 (2007) 10471054namely what is meant by powerful learning environment, what do we mean by electronic,and how to design such an environment.
A powerful learning environment is a place (a) where deep learning is stimulated (i.e., asopposed to rote, surface level learning), (b) where students or groups of students are(intrinsically) motivated and stimulated to study and learn (i.e., as opposed to makinguse of external motivational techniques such as punishment and reward), and (c) whichallows for discussion, dialogue and argumentation, eventually leading to knowledgeproduction.
Deep learning involves the critical analysis of new ideas, linking them to already knownconcepts and principles, and leads to understanding and long-term retention of conceptsso that they can be used for problem solving in unfamiliar contexts. In contrast, surfacelearning is the tacit acceptance of information and memorization as isolated andunlinked facts (Biggs, 1999; Dwivedi, 2004, p. 6; Entwistle, 1988; Ramsden, 1992).
Intrinsic motivation involves doing an activity for the inherent satisfaction of the activ-ity itself (Deci, Vallerand, Pelletier, & Ryan, 1991). The eort or motivation on which con-structivist learning environments try to rely is typically intrinsic motivation, with itsassociated features as curiosity, deep level learning, explorative behavior and self-regula-tion (Martens, Gulikers, & Bastiaens, 2004). Research has shown that intrinsically moti-vated students exhibit study behaviors that can be described as explorative, reective,self-regulated, and aimed at deep level processing (e.g., Boekaerts & Minnaert, 2003; Mar-tens et al., 2004; Ryan & Deci, 2000).
The active engagement of learners in collaborative argumentation and constructive dia-logue during problem solving stimulates cognitive conict and query as mechanisms forenriching, combining and expanding understanding of problems that have to be solved(Savery & Duy, 1995) while carrying out activities encouraging learning through exter-nalization of knowledge and opinions, self-explanation, reection on information, andreconstruction of knowledge through critical discussion (Andriessen, 2005; Kanselaar &Erkens, 1996; Kanselaar, De Jong, Andriessen, & Goodyear, 2000).
This means, among other things that it is an environment where learners nd a su-cient number of source materials (including relevant others) and learning aids (i.e., sup-port, guidance, and tools), and are given a chance to interact with source materials in ameaningful way.
A powerful electronic learning environment means that the environment is multi-medial(i.e., it makes use of written materials, sound, motion in both stored form and real-time),that it is connected to distributed sources of information (i.e., it is resource rich), and thatit is connected to others (i.e., it allows collaboration and cooperation).
Finally, this special issue is about designing such environments. To this end we do notchoose the classical denition of design where the basic goal is the development of a planfor the physical production of an environment, but rather a much broader and more edu-cationally inspired denition aimed at creating learning situations which achieve powerfullearning. Goodyear (2005) speaks of the set of practices involved in constructing represen-tations of how to support learning in particular cases. Goodyear sees educational design asa space in which philosophy and pedagogical tactics have to be aligned (see Fig. 1).
A problem with such a situation when designing such powerful learning environmentswas described by Kirschner, Martens, and Strijbos (2004). They state that most systematicdesign process-models center on designing eective conditions for the attainment of indi-
vidual learning outcomes (Van Merrienboer, Kirschner, & Kester, 2003) and attempt to
control instructional variables to create a learning environment that supports the acquisi-tion of a specic skill (i.e., student A will acquire skill B through learning method C). Thisis complicated by the use of groups in the case of collaboration. A multitude of individualand group level variables aect the collaborative learning process, making it practically
Fig. 1. Conceptualizing the problem space of educational design (from Goodyear, 2005).
L. Kester et al. / Computers in Human Behavior 23 (2007) 10471054 1049impossible to predene the conditions of learning or instruction for a group setting so thatinteraction and competency development are controlled.
Instead of a classical causal view, powerful learning environments require a more prob-abilistic approach to design, as shown in Fig. 2. This distinction corresponds with the onemade by Van Merrienboer and Kirschner (2001) between the world of knowledge (the
Causal design view:World of knowledge
Probabilistic design view:World of learningDesign based upon chosen
SkillSkill Partial skill Skill + Unforeseen
Design based upon chosen
Learning environmentbased upon design
Learning environment based
Fig. 2. Causal and probabilistic views of design (from Kirschner et al., 2004).
loosing the realistic aspects of the learning task or (s)he could include embedded supportto the learning task.
1050 L. Kester et al. / Computers in Human Behavior 23 (2007) 10471054First, the amount of element interactivity may be initially reduced by simplifying thetasks, after which more and more elements and interactions are added (i.e., a part-wholeapproach). So, such a task sequence begins with the simplest version of a task that is stillrepresentative of the task as a whole and ends with the most complex version of this task(Reigeluth, 1999). For example, learners start studying the anatomy and functioning of thecirculatory system on an organ level (e.g., heart, blood vessels, arteries), and end studyingthe circulatory system on a cellular level (e.g., red/white corpuscles, thrombocytes) or evenmicro-cellular level (e.g., organelles, energy transmission). Or, the task may be immedi-ately presented in its full complexity while the element interactivity is reduced by havingthe learner take continually more interacting elements into account when carrying it outoutcomes) and the world of learning (the processes). In the world of knowledge, design-ers construct methods by which given learning goals in a specic subject matter domaincan be attained by the learner. In the world of learning, designers focus on methodsenhancing learning processes rather than on the attainment of predened goals. More spe-cically designers focus on methods enhancing deep level learning, intrinsic motivationand collaborative argumentation.
Such environments make use of educational techniques such as whole-task practice,self-regulated learning and fading support and guidance, that is, scaolding, and so forth.The articles in this issue zoom in on the role of support and guidance in these environ-ments and either use individual or group settings.
1.1. Whole-task practice
Powerful learning environments provide realistic, authentic learning tasks that are char-acterized by integration (i.e., training knowledge, skills and attitudes simultaneously), andcoordination (i.e., whole-task practice of constituent subskills). Such realistic learningtasks facilitate deep level learning and help learners transfer what is learned to situationsoutside school (Van Merrienboer, 1997). An emphasis on integration and coordination ofknowledge, skills and attitudes during practice pays o in a higher transfer performance(Van Merrienboer, Kester, & Paas, 2006). However, realistic, authentic learning tasksput a higher burden on the cognitive capacity of learners than compartmentalized andfragmented learning tasks.
Realistic learning tasks are characterized by high element interactivity where a learner isrequired to process several learning elements simultaneously in order to achieve a sucientperformance on the task (Sweller, Van Merrienboer, & Paas, 1998; Van Merrienboer &Sweller, 2005). Compartmentalized and fragmented learning tasks are often characterizedby low element interactivity which allows learners to serially process several learning ele-ments for sucient performance. Since working memory is severely limited with regard tothe maximum number of simultaneously active elements it can hold (Cowan, 2001), it isclear that realistic learning tasks demand more cognitive capacity than tasks with a lowelement interactivity. If a realistic task is too complex, as indicated by its element interac-tivity, working memory could be overloaded and learning will be hindered (Sweller, 1988).To avoid this, the designer could either reduce the amount of element interactivity without(i.e., a whole-part approach). During a rst driving lesson, for example, learners drive a
L. Kester et al. / Computers in Human Behavior 23 (2007) 10471054 1051car on an open road with almost no trac, requiring only steering and braking, while dur-ing the last lesson they have to independently operate the car on a busy city street.
In addition to, or apart from, these measures to lower the element interactivity of alearning task, one could add support to it to avoid cognitive overload and help learnersmanage task complexity. Seufert, Janen, and Brunken (this issue) and Munneke, Andries-sen, Kanselaar, and Kirschner (this issue) added graphical support to the learning environ-ment to facilitate learning from complex tasks in powerful electronic learningenvironments. In their article The impact of intrinsic cognitive load on the eectivenessof graphical help for coherence formation, Seufert and colleagues describe three studiesthat investigated the eect of graphical support on learning material with high elementinteractivity. Inter-representational hyperlinks hyperlinks that display connectionsbetween representations when clicked on were used to help learners mentally integratemultiple representations (e.g., text, pictures, graphic organizers) that are mutually refer-ring. The eectiveness of this support in relation to the learners prior knowledge wasstudied.
The article of Munneke and colleagues titled Supporting interactive argumentation:Inuence of representational tools on discussing a wicked problem focuses on graphicalsupport to help groups of learners discuss complex problems in a computer-supported col-laborative learning environment. They assumed that graphical support in the form of anargumentative diagram puts a group discussion on a higher plane than support in the formof a text outline. They compared the breadth and depth of the discussions of the diagramgroups and the outline groups to verify this assumption.
Huk and Steinke (this issue) introduce a visualization technique to aid learners during acomplex learning task and compare it to graphical support. In their article Learning cellbiology with close-up views or connecting lines: Evidence for the structure mapping eectthey describe the eect of zooming in and out between cell and cell organelles as comparedto connecting lines between cell and respective technical term (see also Seufert et al.) onlearning in a hypermedia learning environment. Both techniques aim at directing learnersattention to relevant aspects of a picture during a narrated explanation of that picture andit is examined which one is most benecial for learning.
1.2. Self-regulated learning
Powerful electronic (learning) environments allow for learning in a non-linear fashionby giving learners more control over their own learning. Learners are enabled to selectinformation, tasks, instructional formats (e.g., video, audio, graphic, or text), interfaceproperties, and content (e.g., examples, analogies) in their preferred order and at theirown pace (Merrill, 1994).
Research shows that the intrinsic motivation to learn increases when the locus of con-trol over instructional material is transferred from an instructional agent (e.g., teachers,computers) to the learner (Kinzie, Sullivan, & Berdel, 1988; Reeve, Hamm, & Nix,2003). This results in a more satisfactory learning experience which ultimately leads toan improved academic performance. In other words, learner control is an essential aspectof eective learning (Gray, 1987; Lawless & Brown, 1997; Lou, Abrami, & dApollonia,2001). However, other research indicates that learners with the highest degree of learnercontrol learned the least (Fry, 1972). Hence, the potential advantages reported for learner
control have not been consistent. Several studies (Fry, 1972; Kinzie & Sullivan, 1989;
1052 L. Kester et al. / Computers in Human Behavior 23 (2007) 10471054Lahey, Hurlock, & McCann, 1973) show that despite the negligible or even negative eectson learning outcomes using learner control, it has a positive inuence on learners atti-tudes. So, although learner control has undeniable positive eects on motivation, its eecton learning outcomes is equivocal (Judd, 1972; Lahey, 1976).
It appears that self-regulation ability and level of expertise mediate the eects of learnercontrol on learning outcomes. Research of H...