Full Paper

A wide array of corresponding instructional approaches has been studied and implemented across educational and training settings. Learning environments are routinely characterized using terms such as anchored instruction, problem-based, guided discovery, direct instruction, and student-centered. New perspectives and approaches have, alternately, been the target of widespread skepticism, the focus of unbridled advocacy, and the object of scorn as advocates and critics position themselves to support or refute the legitimacy of a particular approach. While the clamor for new design models to support the unique design requirements of the approaches has risen, the implications for instructional design practice are not always clear. A good deal of instructional design practice has evolved as a kind of procedural and media-production craft rather than a grounded process (Heinich, 1995; Winn, 1997). Evidence of mismatched instructional goals vs. methods are widespread in schools as well as in other training and education performing settings. It is not unusual, for example, for schools to proclaim an emphasis on critical thinking or problem solving, but focus largely on rote learning of declarative knowledge. Likewise, training efforts characterized as "lean and mean" frequently include unnecessary information while failing to provide sufficient practice in critical performance areas. Universities espouse the virtues of "learning science as a scientist" but teach and test for rote learning. There are often significant gaps dividing rhetoric and theory espousing what should happen from design practices that influence what actually happens.

How can so many alternative approaches be advocated, yet so little be linked to their respective epistemologies? We submit that there is insufficient grounding, affecting virtually all approaches, in the practices associated with learning systems design. What does it mean to describe a learning environment as being behaviorally-based, or cognition as being situated in a learning environment? From a psychological or pedagogical perspective, what is implied by describing a learning system being web-based? What beliefs, tacit or explicit, are embodied in the learning environments we design, and what assumptions are attendant to such beliefs? The purposes of this paper are to introduce the fundamentals of grounded design, to describe how underlying foundations and assumptions are aligned with the corresponding methods, and to introduce grounded case examples featuring alignments consistent with different root foundations

Rowland (1993, p. 80) defined design generically as "...a disciplined inquiry engaged in for the purpose of creating some new thing of practical utility" and designing as "..requiring a balance of reason and intuition...and an ability to reflect on actions taken." There are many kinds of design; the distinctions among them are not merely semantic. Merrill et al (1996), for example, note that much of what is described as instructional design today likely does not adhere to the explicit and tacit assumptions of rationalism, that is, a process driven and informed by known rules, established principles, and reliable procedures. If we seek systems that support other kinds of learning, then we need approaches that are sensitive to the differences as well as the similarities reflected in the approaches.

In the present context, grounded design involves establishing links between the practice of learning systems design and related theory and research. Winn (1997) suggested that while theory-based approaches do not provide discrete, "clean" instructional prescriptions, they better equip designers to address the ambiguities inherent in their craft. Bednar, Cunningham, Duffy, and Perry (1995, p. 101-102) stated that "...effective instructional design is possible only if the developer has reflexive awareness of the theoretical basis underlying the design...(it) emerges from the deliberate application of some particular theory of learning." This statement forms the basis of grounded design, but additional criteria must also be met.

For design practice to be grounded, four tests must be met:

• First, design must be based in a defensible theoretical framework. Multiple theoretical frameworks are available from which design decisions can be rooted provided they establish connections among key foundations, identify the assumptions consistent with the corresponding foundations, and lead to methods that are consonant with them.
• Next, methods must be consistent with research conducted to validate or extend the theory upon which it is based. In effect, grounded designs reflect a close link between empirically verified approaches and those employed in a given learning system.
• Third, grounded designs are generalizable. They transcend the individual instances in which success may be evident, and can be adapted or adopted by other designers. This does not imply a literal, algorithmic mapping of methods according to strictly defined conditions, but the availability of well-prescribed processes that can be clearly articulated and readily applied in comparable circumstances.
• Finally, grounded designs are validated empirically through successive implementation, that is, methods must prove effective in practice according to the theoretical framework ascribed to. The design processes and methods continuously inform, test, validate, or contradict the theoretical framework and assumptions upon which they were based, and vice-versa.

Clearly, not all design practice is grounded. Some approaches may even prove effective, but lack grounding. Design practice is frequently influenced by factors such as personal preferences, experience with "what works," and familiarity [see, for example, Bednar, et al (1995)]. These do not inherently promote poor design; indeed, many approaches appear to work, but designers cannot determine why or, more importantly, if similar methods are likely to prove effective in subsequent applications. Many designs reflect the evolved preferences of the designer with particular methods or technologies. Again, the approaches may prove successful, but cannot be readily replicated by others or reconciled with either available research and theory.

It is equally clear that not all frameworks can meet each of the grounded design tests. The theories themselves may be highly formative in nature, lacking maturity or compelling validation. In some cases, the presumed tenets of an approach have not been tested, the methods attributed to the perspectives may prove highly idiosyncratic in application, or the assumptions underlying the perspective may be unclear. The approaches may be interesting, even effective and provocative, but lack completeness and validation.

Simply promoting learning does not necessarily imply grounded design, and failure to learn does not necessarily reflect a lack of grounding. If all approaches were universal and algorithmic in nature, then success could be universally assured. Such is not the case in learning systems design. Learning, and the systems that promote it, are uniquely defined, complex, and conditional. The very goals that reflect important differences and values vary according to contexts and frameworks within which they derive. Learning is not a unitary concept; learning systems design cannot be either.

A serious problem arises when the credibility of a framework is not recognized, or worse, is summarily discounted. Both radical objectivists and radical constructivists have, too frequently, attempted to advance a particular perspective by attacking alternatives. Merrill, et al (1996), for example, characterized constructivist perspectives on learning and instruction as "...wild speculation and philosophical extremism" (p. 5). Ernst von Glasserfeld (1993), a radical constructivist, stated that realists (objectivists) cause no harm "...as long as you don't tell others that the reality you have constructed is the one they ought to, or, worse, must believe in" (p. 28-29). Neither position seems to reflect much understanding of the other; rather, they tend to advance a particular perspective by belittling the other. They promote intellectual isolation in a field where it has become increasingly important to understand alternative views. As Reigeluth (1997) suggested, few have the luxury of positioning themselves squarely on one end of a learning systems design continuum; fewer still can be considered justified in their single-minded endorsements or declamations.

Learning systems designers have too often become increasingly Balkanized in their beliefs and approaches, entrenched in whichever philosophical extremism suits their own purposes, where it is more important to be right than to be wise (cf. Wilson, 1997). This seems not only unwise but counterproductive in a field where so many disciplines are supported and diverse perspectives on teaching and learning are reflected. It is not necessary to agree with or to adopt perspectives to which one does not subscribe; indeed, to understand is not necessarily to endorse. Nor is it necessarily productive to reconcile fundamentally different approaches as variations on the same pedagogical theme; to the extent genuine differences exist, we may gain more by attempting to understand than to minimize them. It has, however, become increasingly important to recognize the pervasiveness of alternative approaches and to understand the assumptions and methods attendant to them.

Learning systems design can be, and has been, grounded in any of a number of established and widely subscribed and supported theoretical frameworks. It is not necessary that we agree on which are "best"; indeed, it may prove not only counterproductive but to do so but impossible to establish. We can disagree, and still advance the perceived merits of a given perspective over others, but we gain little by engaging in acrimonious debates that divide and narrow our perspectives (Wilson, 1997). Grounded design, therefore, argues not for the inherent superiority of one theoretical position or methodology over another, but for articulation of and alignment among the underlying principles that define them. It does not marginalize differences among perspectives, where such differences exist, but advocates approaches that reflect them.

Learning systems design is rooted in several foundations. Psychological foundations represent beliefs about how individuals think and learn. Historically, learning environments were rooted psychologically in behaviorism, then cognitive approaches featuring information-processing (see, for example, Gagné & Glaser, 1987; Hannafin & Rieber, 1989). Other environments derive their foundations from areas such as constructivism (Jonassen, 1991) and situated cognition (Brown, Collins, & Duguid, 1989). Pedagogical foundations emphasize how to-be-learned domains are represented and affordances provided to support learning. Psychological and pedagogical foundations, taken together, reflect underlying beliefs about the nature of learning, the methods and strategies employed, and the ways in which the to-be-learned domain should be organized and made available to the learner. Technological foundations indicate the extent to which features are available to support learning, but learning requirements dictate how, or if, capabilities should be integrated. Cultural considerations reflect things such as beliefs about education, the role of individuals in society, and the prevailing practices of a given professional community, school system, or classroom. Whereas technological foundations influence what is possible technologically, pragmatic foundations dictate the extent to which various alternatives can be implemented. In practice, foundations are interdependent. Each foundation encompasses a wide array of potential influences but all are not relevant in every learning system; rather, the subset of influences for each foundation is considered based upon its relevance to the learning problem or need as well as consonance with one another. The coincidence among foundations represents the unique ways in which they are brought to bear in any given learning environment. For any learning system, the goal is to align the corresponding root foundations in order to maximize their coincidence and shared functions.

As root foundations vary, underlying assumptions change accordingly (and vice-versa). Different assumptions correspond to unique intersections among foundations; each coincidence, in turn, reflects specific underlying assumptions--some explicit, some tacit--about the strategies and methods likely to be appropriate. As assumptions vary, the features and methods of the learning environment likewise change.

Vastly different, yet grounded, practices are exemplified by instructionist-instructional design approaches versus constructionist-constructional design approaches. Gagné, Briggs, and Wager's (1992) approach is consistent with instructionist definitions: reality is seen as objective and independent of the individual learner, and learning principally from an information-processing perspective. Resnick's (1996) concept of constructional design, on the other hand, is rooted in constructivist epistemology, and more specifically Papert's notions of constructionism. Constructional design focuses on the creation of environments that enable and support individual construction or building by engaging in design and invention tasks; the design task is to provide a learning environment within which individual construction is facilitated, not one in which concepts are explicitly taught. The underlying foundations and assumptions of these perspective are, nearly point-for-point, counter to those of Gagné; they are grounded differently, but they are no less grounded. It is unlikely that instructionists would embrace the foundations, assumptions, and methods of constructionists (or vice-versa), but competing approaches can be well-aligned if linked with their corresponding foundations and assumptions.

Conversely, learning systems lack grounding to the extent that their foundations, assumptions, and methods are mis-aligned. A learning environment described as reflecting cognitive-information processing views of learning, yet failing to account for limitations in short-term memory or to facilitate the transfer from short-term to long-term memory, reflects a mismatch between presumed foundations and assumptions and their associated methods. A constructivist's learning environment that decontextualizes information and tutors to mastery is equally ungrounded.

As suggested previously, objectivists view meaning as existing externally, that is, independent of the individual learner. Instructionists, then, emphasize methods that establish and convey the meaning of objects and events consistently and efficiently across learners. The learner's task is to recognize and label relevant objects and events, organize them into coherent chunks, and integrate new with existing knowledge. The learner accomplishes these tasks principally by decoding the established meaning of various objects and events, using the cueing and amplification devices provided by the learning systems designer.

Methods consistent with these assumptions tend to emphasize learning contexts that support the transition from initial, propositional knowledge to signalling when and how it can be used. Instructional analysis procedures can be used to analyze the information requirements and conditional structures of performance. Consistent with Gagné's (1996) views on the learning of intellectual skills, complex skills such as problem-solving are seen as hierarchically dependent on the learning of lower-order skills and concepts, that is, lower-order skills are prerequisite to complex skill development. Thus, declarative or verbal information required for complex conceptual knowledge is identified, isolated, and taught in an appropriate sequence. This is a widely accepted approach among objectivists and is consonant with traditional cognitive psychological foundations emphasizing learning as an incremental, mathemagenically-facilitated process.

Technological foundations link both psychological and pedagogical foundations. For instance, a computer-mediated drill might be provided to support automatic subskill execution (e.g., multiplication tables) needed prior to more complex applications (Salisbury, 1988). Tutorial programs that isolate, simplify, and sequence concepts and skills according to identified task milestones and learning hierarchies might also be consistent with objectivist-instructionist epistemology. Instructionist cultural foundations generally stress well-defined and explicit learning aims and methods, where knowledge and skill requirements can be articulated, progress evaluated, and mastery demonstrated. Such systems emphasize bottom-up, "basics first" curriculum and teaching methods and "need to know" knowledge and skill training. Pragmatically, instructionists tend to reconcile theoretically ideal solutions with those best suited to available resources and constraints. These are manifest in learning systems that, for example, provide internally consistent, modular approaches that accommodate discrete subject offerings and brief, fixed-duration class periods in traditional schools.

Selection and incorporation of these foundations are guided by an implicit assumption that efficient subskill mastery is requisite for subsequent depth of understanding (Dick & Carey, 1996; Gagné, 1985). Learning is defined in terms of attainment of well-defined enabling and terminal objectives; depth of understanding results from highly practiced, successful performance in carefully isolated, systematically sequenced, and externally engineered learning activities. The approaches explicitly assume that requisite skills and associated cognitive processes can be broken down and learned separately from holistic contexts. Information deemed non-essential to the specific instructional goal and readiness level of the learner is considered "deadwood" (Smith & Ragan, 1993, p. 65) and detracts from the learning goal. External engineering of the learning process presumably reduces cognitive load, frees working memory, and eliminates unproductive efforts during the learning process.

Context assumes different values and roles for constructivists. Objects and events have no absolute meaning; rather, the individual interprets each and constructs unique meaning based upon individual experience and evolved beliefs. The design task, therefore, is one of providing a rich context within which meaning can be negotiated and ways of understanding can emerge and evolve. Unlike instructionists, constructivists tend to eschew the breaking down of context into component parts in favor of establishing contexts wherein knowledge, skill, and complexity exist naturally.

Constructivist designers draw upon psychological foundations from theories such as situated learning (Brown, Collins, & Duguid, 1989) and social cognition (Bandura, 1982). Situated cognition theorists suggest that knowledge and the conditions under which is it used are inextricably linked. Social cognitivists indicate that learning is a goal-directed activity that is connected to the social contexts, including people, in which it occurs or is ultimately applied. Both views promote learning in realistically complex contexts that do not decontextualize knowledge and skills from the circumstances in which they are applied (Cognition and Technology Group at Vanderbilt, 1992).

Pedagogical approaches, such as anchored instruction which emphasizes embedding skills and knowledge in holistic and realistic contexts (Cognition and Technology Group at Vanderbilt, 1992), are consistent with situated cognition perspectives. Anchored contexts promote complex and ill-structured problems wherein learners generate new knowledge and subproblems as they determine how and when knowledge is used. Apprenticeship models are similarly aligned as they promote scaffolding and coaching of knowledge, heuristics, and strategies while students carry out authentic tasks (Collins, Brown, & Newman, 1989).

Technology is often used as a tool to explore resources and integrate knowledge problem solving or pursuing an individual learning goal. Vast information databases, such as the World Wide Web, can be browsed to locate information needed to solve a problem or satisfy an information need (Hill & Hannafin, 1997). Telecommunications tools could be used to communicate with others or establish a dialog in a shared context (Linn, Bell, & Hsi, in press). Technology tools might prompt for reflection or guide learners to understand or solve problems. In such a system, technology is part of a larger social context that shapes, constrains, and enhances how information is processed and used. Cultural considerations are evident where far-reaching agendas are established, such as engendered with the successful launching of Sputnik, or when scientific communities establish national standards in areas such as mathematics (National Council of Teachers of Mathematics) and science (National Science Teacher's Association). In each case, priorities are established which influence, or are influence by, beliefs about learning, pedagogy, and technology. Pragmatic foundations become aligned when reasonable accommodations are made based upon unique situational resources and constraints.

Context, in this example, is critical to influencing how information is processed, negotiated, and used, and how understanding evolves. It is assumed that lesson content and heuristics for performance are best embedded in the task itself and represented and determined by the learner, not by an external agent (Brown & Palincsar, 1989). Errors and limitations in understanding form the unique basis for establishing relevance and the need to reconcile prior beliefs with current observations (Papert, 1993); they are encouraged rather than avoided. Consequently, learners are expected to assume additional control over the learning process. It is also assumed that, with the help of teachers, students, or technology to "scaffold" performance, complex tasks are made more manageable without simplifying the task itself (Glaser, 1990; Vygotsky, 1978).

Methods consistent with constructivist foundations and assumptions typically emphasize teacher-student or student-student interactions to model or scaffold understanding and performance (Palincsar & Brown, 1984); technology affords additional alternatives to support the interactions (Linn, Bell, & Hsi, in press). Teachers and students coach strategies within contexts that are complex, authentic, and holistic (Collins, Brown, & Newman, 1989), and designed to support the learning of progressively complex concepts. Similarly, problem-based learning activities (Savery & Duffy, 1996) require learners to draw upon technological, cognitive, and social resources in order solve complex, open-ended problems which serve as orienting contexts for interpretation as understanding and skills are constructed and refined through use.

(c) Michael J. Hannafin

The author(s) assign to ASCILITE and educational and non-profit institutions a non-exclusive licence to use this document for personal use and in courses of instruction provided that the article is used in full and this copyright statement is reproduced. The author(s) also grant a non-exclusive licence to ASCILITE to publish this document in full on the World Wide Web and on CD-ROM and in printed form with the ASCILITE 97 conference papers, and for the documents to be published on mirrors on the World Wide Web. Any other usage is prohibited without the express permission of the authors.