The application of didactic methods when applied to direct teaching-learning of multimedia has been well-documented (see Dillon and Gabbard 1998). Much of this research has focused on the learning of externally defined knowledge and skills facilitated through cognitively-rooted, design principles. Indeed, many time-tested, research-based learning and cognition principles may be applicable across technologies. However, some research may prove of questionable applicability to student-centered, Web-based learning. In this paper, we examine the cognitive implications of student-centered learning in Web-based environments, compare the cognitive demands of student-centered and externally directed learning, and describe implications for research and practice.

Web-based learning, per se, is not inherently student-centered in nature. Rather, it may be externally-directed, student-directed, free-choice, or combinations of each (see Berge and Mrozowski 2001; Bernard et al. 2004). In directed online learning, for example, the boundaries are often specified explicitly by course designers, such as by identifying required resources, specifying learning and performance expectations, and assessing students over defined concepts and constructs (Sharma et al. 2007). Indeed, the student may seek resources beyond the boundaries established by the designer but the structures define and support the expected learning and performance (Hannafin and Hill 2007). Similarly, learning and performance needs may be highly specific in nature and require only specific answers rather than reasoning. In both instances, student effort is directed via external design structures.

In contrast, Hannafin et al. (1999) described student-centered learning where the locus of activity and control shifts to individual responsibility for establishing learning goals and/or determining learning means. Rather than posing direct questions that require correct or incorrect answers, for example, Blumenfeld et al. (1991) implemented “problem[s] that serves to organize and drive activities [that] result in a series of artifacts, or products [where] students can be responsible for the creation of both the question and the activities, as well as the nature of the artifacts” (p. 371). Linn et al. (2003) described a Web-based environment (WISE) that scaffolds middle schoolers’ scientific inquiry by posing questions such as, “How far does light travel?” The student learns scientific concepts, but also the reasoning associated with doing science as they evolve and test working theories.

Thus, student-centered, cognitive demands shift from primarily selecting, processing, and encoding via directed activities to anticipating, seeking, and assessing the relevance of affordances based on individually evolving needs and goals (Kuiper et al. 2005). Often, this has proven both difficult and problematic. Students often failed to develop theories or explanations (de Jong and van Joolingen 1998), to reflect or enact metacognitive processes (Moos and Azevedo 2008b), and to develop coherent, evidence-based explanations (Nicaise and Crane 1999). Land (2000, pp. 75–76) concluded that without effective support,

misperceptions, misinterpretations, or ineffective strategy use … can lead to significant misunderstandings that are difficult to detect or repair…metacognitive and prior knowledge are needed to ask good questions and to make sense.

The epistemological shifts associated with student-centered, Web-based learning raise important, but largely unanswered questions. Sweller et al. (1998), for example, detailed links between instruction strategies and cognitive architecture, and provided recommendations for managing cognitive demands. Mayer (2003) and Clark and Mayer (2003) subsequently detailed how time-tested principles of cognition and instruction should be applicable to multimedia and online learning.

According to Hill and Hannafin (2001), however, this may become substantially more complex in the digital era: “a digital resource’s ‘meaning’ is influenced more by the diversity than the singularity of the perspectives taken” (p. 40). In effect, the potential for increased and largely unregulated resources alters the predictably of cognitive demands associated with resource access and use. Designers are unable to account for individual cognitive demands in advance since the context of learning is often spontaneous and the availability and use of resources evolving continuously. In addition, the evolved cognitive constructs and design principles have been based on directed-learning models where the learning requirements are determined by external agents, not individual students. In this paper, we briefly discuss several issues that have emerged through efforts to design and validate, and identify implications for research and theory related to student-centered, Web-based learning.

Cognitive demands of externally directed and student-centered learning

Table 1 contrasts the demands associated with externally directed and student-centered learning. The American Psychological Association published learner-centered psychological principles which delineated criteria for student-centered learning design and implementation (Alexander and Murphy 1998). Although much of the supporting research pre-dates the advent of the Web, the principles have informed the theory and design of student-centered, Web-based learning. Table 2 summarizes relationships between and among student-centered learning assumptions and design strategies. Since it is beyond the scope of this paper to address all relevant constructs, we focus briefly on constructs with particular relevance for student-centered learning: Prior knowledge, cognitive load, metacognition, beliefs and dispositions, and scaffolding.

Table 1 Contrasting cognitive demands of externally directed and student-centered perspectives on learning
Full size table
Table 2 Assumptions and design strategies associated with student-centered learning
Full size table

Prior knowledge

Classically, cognitive psychologists characterized prior knowledge as networked schema which represents the organization of, and relationships among, each individual’s existing knowledge and skill: The more extensive and connected, the richer the prior knowledge organization (schema), the more amenable to encoding new, related knowledge (see, for example, review by Sweller et al. 1998). Prior knowledge is often elicited by posing criterion-based questions (Pressley et al. 1992), eliciting peer questioning to identify gaps in understanding (Choi et al. 2005), and providing feedback (Smits et al. 2008) to retrieve and process prior knowledge in working memory and facilitate connections between existing and new knowledge. For example, Gagné et al.’s Events of Instruction (1988), aligned with information-processing theory, emphasizes recall of prerequisite knowledge and skill to stimulate internal cognitive events (the association of existing with new knowledge) for acquiring and retaining new knowledge.

While there is broad agreement on the role and importance of prior knowledge to learning, disagreements exist regarding the locus, nature, and meaning of knowledge. Rather than assuming reality exists externally and meaning is transmitted to the learner (accretion, accumulation), constructivists assert that learners construct meaning uniquely based on personal interactions with society, individuals, and objects (Hannafin et al. 1997). Differences in underlying assumptions have compelling implications for designing grounded learning environments that are aligned with any epistemological perspective. Constructivist-inspired learning environments often provide resources for learners to manage their own learning through exploration, hypothesis formation, and student-relevant feedback (Hannafin et al. 1997).

Among student-centered theorists, prior knowledge and experience are perceived as uniquely shaping the individual’s understanding between, as well as expectations among, existing and new knowledge (Land 2000). Understanding and meaning are not assumed to be uniform across, but rather unique to, individual learners. Differences in prior domain knowledge, for example, has been reported to influence the effectiveness of both self-regulation and strategy use during learning from hypermedia (Moos and Azevedo 2008b). Thus, initial understandings, including misconceptions, influence what and how the individual knows and understands as well as the perceived relevance of candidate activities and resources. Rather than imposing a canonical perspective to supplant initial conceptions, student-centered approaches guide the learners in challenging their initial assumptions as they test and refine initial conceptions (Smith et al. 1993).

Thus, prior knowledge and experience mediate the learner’s ability to assume responsibility for their own learning—a central assumption of student-centered learning. They shape the formative, often naïve and incomplete theories-in-action learners employ as they attempt to interpret, make sense and understand (Land and Hannafin 1996). Prior knowledge also mediates how understandings are constructed initially and reconstructed subsequently (Schuh 2003). Relevant knowledge influences the individual’s ability to assess and evaluate information and detect inconsistencies and contradictions between new and existing understanding (Land and Hannafin 2000), to perceive relevance about learning tasks (Kuiper et al. 2005), and to determine when learning goals have been achieved. In many instances, however, domain knowledge alone is not sufficient to facilitate student-centered, Web-based learning. Hill and Hannafin (1997) reported among adult learners, prior Web experience (system knowledge) proved more important than prior domain knowledge in seeking and identifying relevant resources.

Cognitive load

In recent issues of Educational Technology Research & Development (van Merrienboer and Ayers 2005) and Educational Psychologist (Paas et al. 2003), researchers and theorists examined the interplay among the content and attributes of instruction and associated intrinsic, extraneous, and germane cognitive load. Explicit guidance is believed to minimize cognitive load demands (Kirschner et al. 2006). Researchers have attempted to reduce or eliminate extraneous load by amplifying those aspects considered central (germane) to defined goals (Mayer et al. 2001), providing explicit guidance (van Merrienboer and Sweller 2005) and controlling the pace of instruction and demands on working memory (Mayer and Moreno 1998). Worked examples have been reported to reduce cognitive load by providing concrete models during problem solving (Sweller and Cooper 1985). To mitigate extraneous cognitive load during learning, some authorities advocate that tools be introduced and learned prior to rather than concurrent with new learning (Clarke et al. 2005).

Student-centered learning researchers emphasize the individual’s capacity to identify relevant resources and mediate cognitive load. Land and Hannafin (1996) described a method for formalizing, defining and refining working theories (theories in action), where students engaged in complex learning environments, engaged both task-relevant and non-relevant resources as they individually selected and discarded candidate resources as their working theories evolve. Recently, researchers have described self-checking procedures designed to guide learners in goal-setting and Web navigation. Cues, meta-navigation checks during which students were promoted to reassess and evaluate progress, helped students to reconsider their goals, evaluate navigation decisions, and alter subsequent goals and decisions (Puntambekar and Stylianou 2005). Azevedo and Cromley (2004) trained students in self-regulated learning to facilitate learning in a hypermedia learning environment. Collectively research suggests that domain-independent guidance and advance training to refine self-regulation skills may prove beneficial, and in some cases, necessary for learners to engage student-centered environments effectively.

Metacognition

Flavell (1979) conceptualized metacognition as the active, ongoing monitoring of one’s cognitive processes. In addition, since designers’ expectations and requirements influence the extent to which to-be-learned knowledge (and to-be-solved problems) are structured, direct instruction proponents emphasized formal structures and prescribed pathways, such as providing explicit sequences (Dillon and Gabbard 1998) and assessing en-route learning (Hannafin and Rieber 1989) to verify student understanding. Student-centered learning advocates, in contrast, emphasized learning in less-structured or ill-structured environments where students regulate their individual learning (Young 1996).

Few researchers have documented significant improvements across learning types via either directed or student-centered metacognitive support. Researchers who have contrasted the effects of externally- and student-directed Web-based learning designs on student performance have suggested that varied learning requirements require different types of metacognitive support. Eveland et al. (2004) examined the influence of Website organization on three adult learning outcomes, and concluded that while linear design and support improved factual recall, non-linear designs and supports increased the density or connectedness of student knowledge. In effect, one type of learning (factual) was facilitated at the expense of the other (individual associations). Similarly, Shin et al. (2003) posed open-ended organizing questions to assess students’ solving of well- and ill-structured problems in an open-ended, multimedia learning environment and concluded that solving each type of problem required different skills. Whereas domain knowledge and justification skills were significant predictors for solving well-structured problems, ill-structured problem-solving

requires that students possess not only the necessary knowledge but also regulation of cognition, including modifications of plans, reevaluation of goals, and monitoring of one’s own cognitive efforts. If the problem is not structurally complicated enough, the students may not use their regulation of cognitive skills even though they possess them. (p. 23).

Beliefs and dispositions

Researchers and theorists conceptualize and operationalize beliefs and dispositions quite differently. Externally directed approaches tend to provide activities that stimulate processing consistent with to-be-learned canonical beliefs. Predictably, field-dependent learners—those who tend to rely on external agents for guidance—performed most effectively in structured, linear, hierarchical, or relational hypertext architecture (Graff 2003a, b). Student-centered learning advocates typically emphasize the mediating aspects of the individual’s perceptions and beliefs on both dispositions to learn and the state of understanding (Blumenfeld et al. 1991; Kauffman 2004). Typically, they identify how to-be-learned concepts are initially understood by individuals in order to build from, rather than directly correcting, existing beliefs and dispositions. Not surprisingly, field-independent learners—who tend to self-organize and work independently—performed best in a less-structured learning environment (Graff 2006).

Context has demonstrated important influence on the meaningfulness of students’ learning. Context, is inextricably tied to knowledge and influences understanding and meaning; knowledge and understanding vary according to the contexts in which learning occurs. According to Mayer (1989), learning becomes increasingly meaningful when appropriately selected, organized, and integrated within existing cognitive structures. Mayer (2001, 2005) subsequently extended and adapted classical instructional design principles to multimedia learning environment design.

Nearly a century ago, Whitehead (1929) criticized classical British schooling for engendering inert knowledge that has little or no utility and advocated a change in both the goals as well as the methods of education to emphasize knowledge as a tool for reasoning and solving problems. These perspectives were subsequently advanced through the work of Dewey (1998) and subsequently constructivist designers who emphasized the importance of authentic experience and participation.

Accordingly, student-centered learning environments often provide authentic experiences or realistic vignettes to facilitate interaction and learning. In the Jasper Woodbury Series (CTGV 1992), students watch a short video to provide context and orient learning before solving mathematics problems situated in realistic settings (e.g., determining how much gas is needed and what route to take to navigate a boat to desired locations). Thus, contexts may help students to identify learning goals, form and test hypotheses, and situate learning in authentic experiences. Knowledge is constructed while individuals engage activities, receive and provide feedback, and interact within the learning environment. When authentic, active engagement enables learners to gain access (i.e., enculturation or identify formation) to the ordinary practices of a culture from a real-world perspective.

Scaffolding

Scaffolding often assumes the form of explicit directions, explicit guidance, and activities designed to increase the probability of learning defined concepts, constructs, and procedures (Azevedo and Hadwin 2005). In effect, scaffolding guides by simultaneously amplifying lesson aspects considered most central to the defined outcomes while minimizing attention to those aspects considered less essential or extraneous.

During student-centered learning, scaffolding is designed to support the individual’s efforts to identify relevant goals, pursue and monitor efforts toward those goals, and reconcile differences between existing understanding and to-be-learned concepts and constructs. Kauffman et al. (2008) examined the effects of scaffolding in the form of problem-solving and self-reflection prompts on students’ complex problem solving in a Web-based learning environment. They reported that reflection prompts positively influenced problem solving and writing, but only when students also received the problem solving prompts, suggesting that reflection alone influences problem solving when students understand what they reflect on. Similarly, navigation site maps helped students to overcome enroute disorientation due to cognitive overload and facilitate decision-making, while supporting learning goals (Shapiro 2005).

Issues and implications

Role of system versus domain knowledge: how to learn versus what to learn?

Research suggests that familiarity with Web-based tools may play a significant role in individual success or failure. Several authors have advocated affording greater decision-making control to individuals with greater prior knowledge but providing additional support and guidance to those with less (Shapiro 2008). Thus, the presence (or lack of) relevant prior knowledge and skill poses significant hurdles for designers of student-centered learning environments. Song et al. (2004) found that college students who reported greater prior knowledge of online tools managed their time more efficiently than students preferring traditional instruction. Hill and Hannafin (1997) asked teachers to locate Internet content and grade-appropriate materials on a subject of their choosing, and reported that those with previous experience with the Internet were more successful and reported greater confidence in the task—regardless of prior teaching experiences. In both studies, prior tool expertise facilitated learning more than prior domain knowledge or experience. In some student-centered learning contexts, familiarity with available Web-based tools may better predict success than prior domain knowledge and experience.

Although student-centered learning environments purportedly foster exploration and hypothesis formation, validation has proven problematic. McCombs and Whisler (1997) described learner-centered environments where learners engage in complex and relevant activities, collaborate with peers, and employ resources to collect, analyze, and represent information. However, Remillard’s (2005) synthesis indicated that teachers’ content knowledge, pedagogical content knowledge, beliefs, and their interpretation of the curriculum influenced and often dominated how presumed learner-centered activities were actually enacted in classroom settings. Researchers have also documented instances where teachers supplied rote algorithms for students to follow and did not guide students to seek and pursue unique solutions (Doyle 1988).

Likewise, although the Web affords a range of affordances in support of student-centered learning, it has proven difficult to establish conclusive relationships between technology and student learning (e.g., Roschelle et al. 2001). Some researchers have documented positive effects using technology to facilitate problem solving, conceptual development and critical thinking (Ringstaff and Kelley 2002; Sandholtz et al. 1997). Wenglinsky (1998) reported that where teachers used technology in conjunction with learner-centered pedagogies, students scored significantly higher on the mathematics portion of assessments of educational progress than students that did not: 8th graders who used technology for mathematics drill and practice scored significantly lower than peers who used no technology.

Information overload and disorientation: lost in hyperspace?

Disorientation in hyperspace—initially described for hypertext navigation (e.g., Edwards and Hardman 1999)—has become increasingly problematic in student-centered, Web-based environments where learners need to identify, select, and evaluate available resources based on their unique tasks and goals. Web resources are largely unregulated, with quality varying widely in terms of accuracy, authority, and completeness and have been criticized for containing naïve and ill-informed information and propagating misinformation, disinformation, and propaganda (Jowett and O’Donnell 2006). Since students must assess veracity and relevance while attempting to address their individual learning needs and monitoring their understanding, research is needed to examine how students’ evaluate and adapt based on perceptions of a resource’s integrity. Web resource creators can append metadata to simplify their identification and physical locations and narrative descriptors to convey their contents, cataloguing systems typically rely on content creators to generate metadata tags for online materials and cannot be aware of the myriad ways they might ultimately be accessed and interpreted (Maule 2001).

Canonical versus individual meaning: situated learning paradox?

Constructivist learning environments emphasize personal investigation, hypothesis formation and testing. Without adequate background knowledge and support, learners may fail to detect inaccurate information or reject erroneous hypotheses in the face of contradictory evidence. In Land and Zembal-Saul’s (2003) inquiries into the nature of light, participants obtained evidence during experiments, stored it in portfolios with their findings, and generated hypotheses to orient future inquiries. While some groups benefited from computer-assisted inquiry, others relied on faulty results from prior experiments and subsequently misdirected future inquiries and retained erroneous results even when later studies contradicted them. The authors suggested that student-centered inquiry functioned as anticipated only when students had sufficient background knowledge, self-evaluated their knowledge limitations, engaged in critical questioning and clarification, and feedback to challenge faulty explanations. The situated learning paradox suggests that prior knowledge, important for orienting and helping learners to make sense of phenomena, is often based on incomplete and inaccurate conceptions (Land and Hannafin 2000). Without support, misinformation and disinformation may go undetected; fundamental misunderstandings may become reified rather than reconciled.

The shifting nature of knowledge: accretion versus tool?

Whitehead (1929) advocated an emphasis on promoting knowledge as a tool. Tool-based knowledge, valued in student-centered learning, is presumed to facilitate goal acquisition and transfer: When students grasp the underlying reasoning behind the algorithms and their application to authentic problems, knowledge becomes a tool to facilitate problem solving in related contexts. Yet, researchers suggest that tools touted to support student-centered learning are often used inappropriately and ineffectively and engender dependence, rather than independence, especially among young learners. Oliver and Hannafin (2001) examined middle-school students’ use of Web-based Knowledge Integration Environment prompts designed to scaffold teacher support and student thinking and reasoning. Although guidance was intended to become internalized and faded through sustained use, students continued to rely on them. In effect, conceptual and metacognitive scaffolds served as procedural job-aids.

Scaffolding variable knowledge and skills: to guide or to tell?

Saye and Brush (2007) distinguished between hard and soft scaffolds: Hard scaffolds are fixed in nature and designed to support common learning needs across students, freeing the instructor to provide adaptable, on-demand, contextually sensitive soft scaffolding support based on emergent, individual needs. Kim et al. (2007) proposed a scaffolding framework to optimize the interplay between and among technology, teachers and students in everyday student-centered, Web-based learning contexts. Azevedo and Hadwin (2005) implemented scaffolds tailored and presented dynamically to accommodate specific types of learning, tools available, and individual learner needs. Conceptual scaffolds helped learners to identify and use information relevant to the learning context (Brush and Saye 2000, 2001). Embedded metacognitive prompts have been applied in Web-based environments to scaffold self-regulation and self-monitoring and adapted to accommodate differences in individual priori knowledge, experience and perceptions (Kauffman et al. 2008). Peer interaction and question prompts helped learners to represent and generate solutions to ill-structured problems (Chen and Bradshaw 2007; Ge and Land 2004), and coach/tutors deployed to diagnose needs and adapt scaffolding dynamically during learning (Azevedo et al. 2004). Finally, cognitive processes have been effectively modeled during Web-based learning through the use of videos, artifacts, and think-aloud protocols (Dickey 2008).

Research also indicates that soft scaffolding technologies have the potential to address the varied needs of individual student-centered learners (Saye and Brush 2007). Unlike domain supports, soft scaffolding provided by teachers, peers and other human resources is thought to accommodate real-time, dynamic changes in learner needs and cognitive demands. Again, however, these effects have often proven equivocal. Technology-enhanced support during multimedia presentation improved the learning of basic information, but not the transfer ascribed to student-centered learning (Jamet et al. 2008).

Reconciling epistemological differences: the influence of attitudes, beliefs and practices

Song et al. (2007) reported conflicts when learners engage resources that are inconsistent or incompatible with individual goals and beliefs—especially when students are unable to identify and reconcile the differences. Thus, while designers and instructors of Web-based multimedia may assume that extending the array of resources will enhance learning, the individual’s familiarity, beliefs, motivations, and practices may influence the extent to which available resources complement or confound student-centered learning.

Some researchers suggest that while Web-based approaches have the potential to promote deeper learning when strategies are followed, many strategies are unutilized or underutilized. In an effort to deepen understanding of mathematics through investigation, Orrill (2006) created an extensive Website including open-ended investigations, a mathematics dictionary, discussion board, and electronic portfolios. Teachers explored available resources, selected problems, and identified their own instructional paths (combined with attendance in face-to-face workshops). Improvements in mathematics skills and depth of knowledge were expected, but teachers typically focused on technology skills and did not refine their understanding or skills. Research is needed to examine how affordances are utilized and negotiated individually, meaning is assembled differentially based on unique needs and goals, and the extent to which individual needs are addressed.

Addressing limited cognitive resources: the cognitive load conundrum

Eveland et al. (2004) reported that students learned factual content best from linear Websites, but understood relationships better from nonlinear Websites. Eveland and Dunwoody (2001) compared the performance of students assigned to browse a Website with different hyperlinking and navigation structures with a paper-only format. The paper-based control group outperformed two of the online groups, indicating that hyperlinking may increase extraneous cognitive load. Thus, nonlinear Websites may increase germane load for some types of learning but increase extraneous load for others (Eveland et al. 2004). Given the demands associated with student-centered, Web-based learning, the ability to meter or manage cognitive load will prove essential for effective online students (Hill et al. 2005).

Monitoring student-centered learning: prerequisite versus emergent

Students who have, or develop, metacognitive strategies tend to perform more successfully than those who do not. Smidt and Hegelheimer (2004) interviewed high, middle, or low-performing adult learners regarding their Web learning strategies; only advanced learners used strategies (as well as cognitive ones). Intermediate and lower-level students relied on cognitive strategies only, suggesting that advanced metacognitive abilities may be either associated with or requisite to effective online learning. Thus, we need to clarify the extent to which learners must possess metacognitive strategies, require advance training, or can develop the requisite skills needed to monitor their progress.

Closing comments

The purpose of this paper was neither to advocate for, nor refute or minimize the value of, either student-centered or direct-instruction approaches. Rather, we clarified where differences are alleged to exist, and described how researchers have studied cognition during directed and student-centered, Web-based learning. Many cognitive constructs are relevant to student-centered and externally directed learning, but the manner in which they are manifest and studied may vary significantly. While Web-based and student-centered learning have gained considerable momentum recently, it is important to distinguish between the promise (and perhaps potential) and the evidence supporting these approaches. Clearly, many issues remain unaddressed, and those issues will not be readily resolved.

Much of the published literature has emphasized research on design, where specific strategies are examined for differential impact on learning. Indeed, while cognition and instruction design practices have been proposed (Clark and Feldon 2005; Clark and Mayer 2003; Jacobson 2008; Mayer 2005; Moreno and Mayer 1999, 2000), their application in student-centered, Web-based, multimedia, and online learning environments has been questioned (DeSchryver and Spiro 2009; Hannafin et al. 2007, 2009; Knowlton 2000). Comparatively little design-based research has documented the theory and practice related to optimizing student-centered learning. Both methods provide potentially important findings, but they do not provide identical information, pose or address the same questions, or generate equivalent evidence. The design issues raised in this paper require not simply to reexamine our assumptions underlying student-versus-externally centered approaches, but to also reconceptualize how we frame research questions, the discipline employed to address the questions, and our goals for the inquiry. As interest in optimizing the impact of student-centered, Web-based learning environments continues to emerge, it becomes critical that we address design and performance questions using methods that extend and refine research, theory, and practice.