Proposal Type:	Symposium
Domain:	Learning and Cognitive Science
SIG:	Instructional Design
Type	Submitted Symposium
Title	Why constructivist teaching does not work
Abstract	The last half century has seen a considerable emphasis on minimising guidance during teaching with the use of discovery learning or constructivist teaching techniques gaining prominence. The popularity of these techniques has been maintained despite a near total lack of supporting empirical evidence based on randomised, controlled experiments. Instead, the empirical evidence almost uniformly supports a heavy emphasis on instructional guidance. Furthermore, most current conceptions of human cognitive architecture and the epistemology of learning and teaching either explicitly or implicitly reject the notion of learners discovering knowledge with minimal instructional assistance. The four presentations of this symposium explore the various empirical, cognitive and epistemological issues associated with this debate.
Equipment	PC and projector
Keywords	Cognition Information processing Instructional strategies

Chair list
Name	Surname	Institution	Country	E-Mail	EARLI Number
Jeroen	van Merrienboer	Open University of the Netherlands	Netherlands	Jeroen.vanMerrienboer@ou.nl

Organiser list
Name	Surname	Institution	Country	E-Mail	EARLI Number
John	Sweller	University of New South Wales	Australia	j.sweller@unsw.edu.au
Richard	Clark	University of Southern California	United States	clark@usc.edu
Paul	Ayres	University of New South Wales	Australia	p.ayres@unsw.edu.au
Paul	Kirschner	University of Utrecht	Netherlands	P.A.Kirschner@uu.nl

Discussant list
Name	Surname	Institution	Country	E-Mail	EARLI Number
Alexander	Renkl	University of Freiburg	Germany	renkl@psychologie.uni-freiburg.de

Paper Details

Title	Human cognitive architecture and its implications for constructivist teaching
Abstract	Human cognitive architecture constitutes a natural information processing system whose evolution has been driven by another natural information processing system, evolution by natural selection. Considering human cognition from an evolutionary perspective has considerable instructional consequences. Those consequences can be used by theories such as cognitive load theory to generate instructional procedures. All such procedures place their emphasis on direct instruction rather than versions of discovery learning or constructivist teaching. Discovery learning techniques were developed prior to our current understanding of human cognitive architecture and are incompatible with that architecture. As a consequence and unsurprisingly, the field has failed to produce a large body of empirical research based on randomised controlled experiments demonstrating the effectiveness of constructivist teaching techniques.
Summary	Human cognitive architecture constitutes a natural information processing system whose evolution has been driven by another natural information processing system, evolution by natural selection. Considering human cognition from an evolutionary perspective has considerable instructional consequences. Geary (2002; 2005; in press) considered how and why particular human cognitive characteristics evolved. For example, we are able to assimilate the large amount of information required for a first language relatively easily but have much more difficulty assimilating mathematical principles. Learning a first language, recognising faces, understanding complex social relations, constitute primary knowledge that we have evolved to acquire and so can learn effortlessly and largely unconsciously. In contrast, learning the topics covered in educational contexts constitutes secondary knowledge that we have not evolved to acquire. The processes used to acquire secondary knowledge are quite different to those required to acquire primary knowledge. The acquisition of secondary knowledge is conscious and deliberate, and follows underlying principles that are identical to the principles used by biological evolution to acquire general information (Sweller, 2003; Sweller & Sweller, in press). These principles, outlined next, constitute a natural information processing system. Information store principle. Natural information processing systems rely on a huge store of information to allow them to function in natural environments. A genome provides that store in the case of biological evolution while knowledge held long-term memory provides a similar function in the case of human cognition. Borrowing and reorganizing principle. Almost all information in a natural information store is borrowed from another store and reorganised. Sexual reproduction provides a biological example. While all information is borrowed from ancestors, the reorganizing process ensures that descendants always differ from ancestors. Similarly, almost all knowledge held in a person’s long-term memory is borrowed from the memories of other individuals. We imitate others, listen to what they say and read what they write. The borrowed information is reorganized during schema construction. Randomness as genesis principle. The borrowing and reorganizing principle does not create new information, it reorganizes old information. Random mutation is the engine of creativity in biological evolution. Each mutation is tested for effectiveness with effective mutations retained and ineffective mutations jettisoned. In human cognition, novelty is similarly generated during problem solving. When deciding on a problem-solving move, we use (a) knowledge; (b) random generation followed by tests for effectiveness; or (c) most commonly, a combination of knowledge and random generate and test. As is the case for biological evolution, random generate and test is the ultimate source of novelty. Furthermore, other than random generate and test, to this point no other source of creativity has been isolated in either biological evolution or human cognition. Narrow limits of change principle. The narrow limits of change principle is concerned with information flowing from the environment to the information store. The epigenetic system acts as a mediator between the environment and the DNA-based genome. For example, the epigenetic system can determine mutation rates and locations and in this sense, the epigenetic system effects the information that is stored in DNA. Working memory has the same role in cognition as the epigenetic system has in evolution by natural selection. Information from the environment to be stored in long-term memory is processed by working memory. Only very small amounts of new, environmental information can be processed because that information is generated via the previously mentioned randomness as genesis principle and the probability of a large amount of novel information being effective is slight. Environmental organizing and linking principle. The environmental organizing and linking principle deals with information flowing from the information store rather than to the store. Because it is organized, unlike information coming to the store, information from the information store has no processing limits. Huge amounts of information in DNA is used to control the synthesis of protein. The epigenetic system again mediates the flow of information from DNA. Working memory has a similar function in human cognition. There are no known limits to the amount of organized information from long-term memory that working memory can process (Ericsson & Kintsch, 1995). Together, these principles that govern the functioning of the human cognitive system also determine effective instructional procedures. Primary knowledge is acquired without direct instruction because we have evolved to acquire that information. Much constructivist teaching theory implicitly assumes that the procedures that work when acquiring primary information should be just as effective when dealing with the secondary information relevant in educational contexts. In fact, if we ask learners to acquire secondary information in the same manner as they acquire primary information, they have no choice but to use the randomness as genesis principle instead of the far more efficient borrowing and reorganising principle that emphasises direct instruction. The result of emphasising the randomness as genesis principle is minimal learning. As a consequence and unsurprisingly, the field has failed to produce a large body of empirical research based on randomised controlled experiments demonstrating the effectiveness of constructivist teaching techniques (Kirschner, Sweller, & Clark, 2006). References Ericsson, K. A., & Kintsch, W. (1995). Long-term working memory. Psychological Review, 102, 211-245. Geary, D. (2002). Principles of evolutionary educational psychology. Learning and Individual Differences, 12, 317-345. Geary, D. (2005). The origin of mind: Evolution of brain, cognition, and general intelligence. Washington, DC: American Psychological Association. Geary, D. (in press). Educating the evolved mind: Conceptual foundations for an evolutionary educational psychology. In J. S. Carlson & J. R. Levin (Eds.), Psychological perspectives on contemporary educational issues. Greenwich, CT: Information Age Publishing. Kirschner, P., Sweller, J., & Clark, R. (2006). Why minimal guidance during instruction does not work: An analysis of the failure of constructivist, discovery, problem-based, experiential and inquiry-based teaching. Educational Psychologist, 41, 75-86. Sweller, J. (2003). Evolution of human cognitive architecture. In B. Ross (Ed.), The psychology of learning and motivation (Vol. 43, pp. 215-266). San Diego: Academic Press. Sweller, J., & Sweller, S. (in press). Natural information processing systems. Evolutionary Psychology.
Keywords	Cognition Information processing Instructional strategies
Appendices

Authors
Name	Surname	Institution	Country	e-mail	EARLI Number	Presenting
John	Sweller	University of New South Wales	Australia	j.sweller@unsw.edu.au		*

Title	Borrowing Expertise: Cognitive Task Analysis for Complex Learning
Abstract	This presentation extends the discussion of the “borrowing and reorganizing principle” introduced by Geary (2005, In Press) and elaborated by Sweller (2006) as an explanation for the failure of constructivist learning strategies (Kirschner, Sweller & Clark, 2006). During schema construction all “learned” information is “borrowed” from the experiences of experienced (expert) others through observation, modeling and direct instruction and then reorganized to achieve performance goals. Over the millennia, the borrowing process has, by necessity, focused on observable aspects of expertise. Complicating this process for learners is evidence that about 70 percent of expert processes are automated, unconscious and not observable by either the expert or the “borrower” (Clark & Elen, 2006). The consequence is that learners, instructional designers and teachers find it difficult to observe and borrow the many covert, implicit cognitive strategies and processes used by experts. This presentation offers evidence from a number of studies to support the claim that the borrowing process can be made significantly more efficient and effective if we employ Cognitive Task Analysis (CTA) to capture the implicit expert processes necessary to support successful complex task performance and provide them to learners as part of the instructional design and development process. Finally, the use of CTA in instructional design is briefly described here and elaborated in the presentation by van Merrienboer in this symposium.
Summary	This presentation extends the discussion of the “borrowing and reorganizing principle” introduced by Geary (2005, In Press) and elaborated by Sweller (2006) as an explanation for the failure of constructivist learning strategies (Kirschner, Sweller & Clark, 2006). During schema construction all “learned” information is “borrowed” from the experiences of experienced (expert) others through observation, modeling and direct instruction and then reorganized to achieve performance goals. Over the millennia, the borrowing process has, by necessity, focused on observable aspects of expertise. Complicating this process for learners is evidence that about 70 percent of expert processes are automated, unconscious and not observable by either the expert or the “borrower” (Clark & Elen, 2006). The consequence is that learners, instructional designers and teachers find it difficult to observe and borrow the many covert, implicit cognitive strategies and processes used by experts. This presentation offers evidence to support the claim that the borrowing process can be made significantly more efficient and effective if we employ Cognitive Task Analysis to capture the implicit expert processes necessary to support successful complex task performance and provide them to learners as part of the instructional design and development process. Cognitive task analysis (CTA) consists of a variety of interview and observation strategies designed to capture an accurate and complete description of the explicit and implicit knowledge experts use to perform complex tasks (Clark et al, In press). Complex tasks are defined as those where performance requires novices to learn tasks requiring a number of interacting elements that, taken together, exceed working memory processing limits as well as the integrated use of both controlled (conscious, conceptual) and automated (unconscious cognitive strategy) knowledge. The results of CTA have been used as the basis of expert systems, the development of tests to certify job or task competence and as the content of instruction when people must acquire new and complex knowledge in order to achieve a performance goal. In most of the current CTA processes, knowledge extraction is performed in the following sequence: 1) Collect preliminary knowledge 2) Identify knowledge representations 3) Apply focused knowledge elicitation methods 4) Analyze and verify data acquired 5) Format results for the intended application Research evidence indicates that the accurate identification of experts’ cognitive processes can be adapted into training materials that are substantially more effective than those developed through other means (e.g., Merrill, 2002; Schaafstal, Schraagen, & van Berlo, 2000; Velmahos et al., 2004). When content is inaccurate or incomplete, any instruction based on that knowledge will be flawed (Clark & Estes, 1996). Such flaws interfere with performance and with the efficacy of future instruction (Schwartz & Bransford, 1998). Resulting misconceptions resist correction, despite attempts at remediation (Bargh & Ferguson, 2000). Several studies provide direct evidence for the efficacy of CTA-based instruction. For example, in a study of medical school surgical instruction, an expert surgeon taught a procedure (central venous catheter placement and insertion) to first-year medical interns in a lecture/demonstration/practice sequence (Velmahos et al., 2004). The treatment group’s lecture was generated through a CTA of two experts in the procedure. The control group’s lecture consisted of the expert instructor’s explanation as a free recall, which is the traditional instructional practice in medical schools. Both conditions allotted equal time for questions, practice, and access to equipment. The students in each condition completed a written posttest and performed the procedure on multiple human patients during their internships. Students in the CTA condition showed significantly greater gains from pretest to posttest than those in the control condition. They also outperformed the control group when using the procedure on patients in every measure of performance; including an observational checklist of steps in the procedure, number of needle insertion attempts needed to insert the catheter into patients’ veins, frequency of required assistance from the attending physician, and time-to completion for the procedure. Lee (2004) conducted a meta-analysis to determine how generalizable CTA methods are for improving training outcomes across a broad spectrum of disciplines. She located 318 studies where CTA had been compared with other methods of capturing expert knowledge for use in performance improvement. Only seven studies qualified, based on the criteria of: Training based on CTA methods with an analyst, conducted between 1985 and 2003, and reported pre and post test measures of training performance. A total of 39 comparisons of mean effect size for pre- and posttest differences were computed from the seven studies. Analysis of the studies found effect sizes between .91 and 2.45, which are considered to be large (Cohen, 1992). The mean effect size was d=+1.72, and the overall percentage of post-training performance gain was 75.2%. Results of a chi-square test of independence on the outcome measures of the pre- and posttests (χ2 = 6.50, p < 0.01) indicated that CTA most likely contributed to the performance gain. For optimal application to instruction, CTA methods should be fully integrated with a training design model to facilitate the alignment between learning objectives, knowledge (declarative and procedural) necessary for attaining the objectives, and the level of instructional guidance required by the prior knowledge of learners. Currently, there are three major systems that take this approach: Integrated Task Analysis Model ( Ryder & Redding, 1993), Guided Experiential Learning (Clark, 2004, 2006), and the Four Component Instructional Design system (4C/ID; van Merrienboer, 1997; van Merrienboer et al., 2002; van Merrienboer & Kirschner, in press). Of these, the 4C/ID model is the most extensively developed and will be discussed in the presentation by van Merrienboer (2006).
Keywords	Cognition Information processing Instructional strategies
Appendices

Authors
Name	Surname	Institution	Country	e-mail	EARLI Number	Presenting
Richard	Clark	University of Southern California	United States	clark@usc.edu		*
Kenneth	Yates	University of Southern California	United States	Kenneth.yates@usc.edu
Sean	Early	University of Southern California	United States	searly@usc.edu

Title	Why constructivist mathematics teaching does not add up
Abstract	This paper argues that the constructive approach many ‘reformists’ advocate for teaching mathematics is flawed. Some of the main tenets of constructivism are examined along with how these central ideas have been linked directly to teaching mathematics. The emphasis on problem solving and shared social interactions rather than the use of expository teaching is considered. Several points are made. Firstly, the paper seeks to define constructivist teaching, and explain how different interpretations of epistemological theory have led to diverging opinions on the role of problem solving and discovery learning. Secondly, it argues that the constructivist approach fails to take account of recent findings in human cognitive architecture, preferring a one-size-fits-all model irrespective of the knowledge base of the learner. Thirdly, it is argued that constructivists have ignored the wider findings of teacher effectiveness research, which clearly identify a significant role for direct instruction. Fourthly, it is argued that constructivism has a weak research base, heavily reliant on small-scale qualitative data and lacking randomised, controlled experiments. Finally, the paper argues that the strong push for reforms based on constructivism has fuelled a public perception that mathematics teaching is in decline due to the advent of ‘fuzzy’ maths.
Summary	Constructivism in mathematics education is based on the belief that mathematical knowledge is actively constructed by the learner, rather than through passive reception of information (von Glasersfeld, 1991). It is argued by many constructivists that it is primarily an epistemological theory and does not prescribe a particular method of teaching (Simon, 1995). Nevertheless, advocates make the point that it has clear consequences for teaching. Mathematics educators and organizations have used the basic tenets of constructivism to inform classroom teaching-practices (such as the NCTM 1991 Standards). So what is constructivist teaching? This question is quite difficult to answer because theorists/ educationists have many views of what defines teaching using a constructivist approach; however, it is very clear what it isn’t. It does not involve direct instruction and the transmission of information (Confrey, 1991). Constructivist teachers do not stand at the front of the class and impart facts and procedures to students (Wood Cobb & Yackel, 1995). What is promoted is the view that mathematics should be taught with an emphasis on problem solving and social interactions, where small group collaborations and classroom discussions are an essential feature. Despite this problem-centred approach, radical constructivists like Confrey and Cobb et al. argue that this approach should not be considered as classical problem solving, as typified by the work of Polya, because the problems are not identified by the teacher but arise out of students’ attempts to complete tasks and related social interactions (Cobb, Wood & Yackel, 1991). The constructivist approach many advocate for teaching mathematics is flawed for a number of reasons. Firstly, it fails to take account of human cognitive architecture. Although it emphasises building upon prior knowledge, an important tenet of schema theory, it often fails paradoxically to consider the individual needs of the learner. The central activity of interactive problem solving is a one-size-fits-all model irrespective of the knowledge base of the learner. Research by the cognitive load theorists (Sweller, 1999; 2004) has shown that how much the learner knows influences the effectiveness of the instructional strategy (Kalyuga et al., 2003). Learners with low prior-knowledge do not benefit from a problem solving approach, compared with more direct instructional methods such as worked examples (see Atkinson et al., 2000). In contrast, learners with high- knowledge in the domain can learn from a problem solving strategy (Kalyuga et al.). Furthermore although many constructivists use the work of Vygostsky (1978), particularly from the aspect of socially constructed classroom communities, the Vygotsky concept of the ‘zone of proximal development’ is often overlooked through the inappropriate use of problem solving in a domain, which is highly complex, requiring instructional models that scaffold learning (see van Merrienboer et al., 2003). From a constructivist viewpoint, discussion and reflection can overcome such complexity, despite the research suggesting otherwise. Secondly constructivists ignore the wider findings of teacher effectiveness research. Twenty years ago Brophy (1986) argued that constructivists needed to consider this broader literature. A review of the research by Brophy & Good (1986) in that year found that effective teachers actively teach, they give information in a number of different ways, they employ differing grouping arrangement including small groups. More recent research has found that teachers utilise a whole range of strategies to help students build understanding and learning (see Ayres, Sawyer & Dinham, 2004; Cooper & McIntyre, 1996; Hay Mcber, 2000). Furthermore, the 2000 NCTM Standards made the sensible point that “… there is no one “right way” to teach.” (p.18). The legendary paragraph 243 of the Cockcroft report (1982) identified six successful approaches to teaching, including exposition and practice of skills and routines. Present day constructivists seem to ignore these methods and many other effective strategies that do not fit their ideology, just like they did 20 years ago. Thirdly, constructivism relies heavily on qualitative data, particularly case studies to provide its evidence. For a theory which makes such strong claims about the best way to learn and teach mathematics, there is a severe lack of controlled experiments that can provide evidence for these claims. Typically, students are asked to verbalise their thoughts and solutions during a constructivist learning-episode. Whereas this may provide interesting data on how students’ think mathematically, it doesn’t provide convincing evidence that constructivist teachers are effective. Brophy also made this point twenty years ago. Finally, the strong push for reform based on constructivism has fuelled a public perception that mathematics teaching is in decline- we now have the concept fuzzy maths and the recent “maths wars” in the US. The poor western TIMSS results support this perception. Clearly some East Asian countries have a more direct approach to teaching mathematics than the recommended constructivist approach. However, as Ma (1999) points out Chinese teachers seem to be both traditional (text- book based and a leader) and progressive (teaching for understanding). To their credit NCTM have watered down, if not changed their views. Whereas the 1991 Standards pushed a constructivist approach, lock-stock-and barrel, with its emphasis on discourse and the language of the post-modern, the 2000 version is much more accepting of a wider range of teaching strategies, including constructivist ideas. Overall, the constructivists views that teaching has to focus specifically on problem-based learning and social interactions is flawed. Such an approach fails to take account of human cognitive architecture and research into teacher effectiveness. It makes claims about its effectiveness but is unable to provide evidence for them in controlled experiments. From our point of view, certain aspects have merit, but such techniques must be used judiciously at the right time with the appropriate learners, and as a complement to other tried and tested strategies.
Keywords	Cognition Information processing Instructional strategies
Appendices

Authors
Name	Surname	Institution	Country	e-mail	EARLI Number	Presenting
Paul	Ayres	University of New South Wales	Australia	p.ayres@unsw.edu.au		*

Title	Epistemology is not equal to pedagogy
Abstract	Just as a child is not a little adult, a learner is not a little scientist. Children’s concepts are different in structure and meaning from adults. As such, how one best learns and should be taught in a domain is quite different from how one performs or “does” in a domain. The problem is that because experimentation and discovery is central to how knowledge is acquired by scientists (i.e., experts), many then feel that experimentation and discovery should also be used as the pedagogy for acquiring knowledge, organising the curriculum, and designing the learning environment. But this is not the case, and it is naive to assume that a theory of education, teaching and/or learning can be extracted directly from a philosophy of science. They are separate entities requiring vastly different activities. Discovery presupposes a prior conceptual framework. By means of discovery one can investigate relationships between concepts, but there is no guarantee that it will lead one to new concepts. This depends upon the structure and content of existing knowledge. Constructivist teaching approaches fail to distinguish between the learner/novice and the doer/expert and thus to distinguish between teaching/learning a science and doing a science. The mistake lies in overlooking that students are not experts and do not practice a science, but they are novices and are learning about a science and/or learning to practise a science. It is the teacher's job to teach science, teach about science and teach how to do science. It is not the teacher’s job to practice science as part of the teaching exercise. This presentation discusses this educational anomaly.
Summary	Just as a child is not a little adult, a learner is not a little scientist. Children’s concepts are different in structure and meaning from adults. Both Vygotsky (children have different kinds of concepts from adults; they don’t have true concepts until puberty) and Piaget (there are shifts between major periods which can be interpreted as changes in representational formats and processes that operate on them) would argue that there are fundamental domain general changes in mental machinery from childhood to adulthood. According to Carey (1999, 2000; & Spelke, 1994), for example, adults but not children can think about their own mental processes, children lack concepts that apply widely across other domains (e.g., causality), and they have different theories about the world (i.e., they are universal novices). Vosniadou (2002), in turn, notes changes in concepts and explanatory frameworks as hallmarks of conceptual change, adding that misconceptions may arise as learners attempt to bring new ideas into prior conceptual structures. With respect to designing instruction, Vamvakoussi and Vosniadou (2004) warn that “presuppositions that constrain learning are not under the conscious control of the learner. It is important to create learning environments that allow students to express and elaborate their opinions, so that they become aware of their beliefs” (p. 466). Van Merrienboer and Kirschner (2007, In press) note that there are considerable differences between domain models which describe the effective mental models used by competent task performers and the intuitive or naive mental models of novice learners in that domain. Such intuitive or naive mental models are often fragmented, inexact, and incomplete; reflecting misunderstandings or misconceptions where learners are unaware of the underlying relationships between the elements. As such, how one best learns and should be taught in a domain is quite different from how one performs or “does” in a domain (i.e., this is the difference between learning science vs. doing science). The epistemology of most sciences, for example, is often based upon experimentation and discovery and, since this is so, experimentation and discovery could and probably should be an important part of any curriculum aimed at teaching and training future scientists. But this does not mean that experimentation and discovery should also be used as a basis for organizing a curriculum and designing a learning environment (Bradley, 2005; Kirschner, 1992; Summers, 1982). Many modern curriculum developers and reformers confuse the epistemological with the psychological and pedagogic bases for teaching. Epistemology refers to the way knowledge is acquired and the accepted validation procedures of that knowledge. In the natural and social sciences, for example, this is essentially based upon experimentation, discovery and testing. Because experimentation and discovery is central to how knowledge is acquired by scientists (i.e., experts), many feel that experimentation and discovery should also used as the pedagogy for acquiring knowledge, organising the curriculum, and designing the learning environment. Novak (1988), in noting that the major effort to improve secondary school science education in the 1950s and 1960s fell short of expectations, goes so far as saying that the major obstacle which stood in the way of “revolutionary improvement of science education…” was the obsolete epistemology that was behind the emphasis on ‘enquiry’ oriented science” (79-80). Hodson (1986, 1992) characterises this as the widespread adoption of the “epistemologically weak and pedagogically inappropriate" discovery learning and process approaches that deliberately avoid giving the learner a prior theoretical understanding of the experiment and its context. Curriculum designers using a discovery learning/constructivist teaching approach operate on the belief that the way science is practised is also the best way to teach and learn science. It is, however, “naive to assume that a theory of education can be extracted directly from a philosophy of science. These two phenomena belong to different domains; albeit overlapping domains in some aspect” (Swift, 1982, p. 39). Critics of inquiry-based (activity-driven) instruction such as Sewall (2000), caution that such approaches to teaching and learning are over-emphasized at the expense of “carefully prepared lesson… focused and guided…with frequent interchange between student and teacher, student and student; interspersed with small group work when appropriate; and with a clear sense of direction at the beginning and summary at the end, leaving all participants with a feeling of completion and satisfaction” (p. 6). This lack of clarity about the difference between learning science and doing science coupled with the priority afforded to knowledge construction, thus, has led many educators to advocate the discovery method as the way to teach science. Not only does this seem to mesh well with ideas in the philosophy of science, but this also meshes well with constructivist, learner centred, supply-centred views emphasising direct experience and individual enquiry. As far back as 1978, Cawthron and Rowell stated that it all seemed to fit, the coalescing of the logic of knowledge and the psychology of knowledge under the “mesmeric umbrella term ‘discovery’”. Why should educators look further than the “traditional inductivist-empiricist explanation of the process”? However, because discovery learning relies heavily on inductivism, it also presents a distorted and inadequate view of the methodology of science. Although skills such as observing, recording, accuracy, generalising and so forth are important in themselves, they only partially represent what a scientist does. The idea that unbiased observation leads "infallibly to conceptual explanations is philosophically and psychologically absurd…to `discover' anything at all they [the learners] need a prior conceptual framework" (Hodson, 1988, p. 40-41). Discovery, returning to Vosniadou (2002), presupposes a prior conceptual framework. By means of discovery one can investigate relationships between concepts, but there is no guarantee that it will lead one to new concepts. This depends upon the structure and content of existing knowledge. "If we confront the world with an empty head, then our experience will be deservedly meaningless. Experience does not give concepts meaning; if anything concepts give experience meaning" (Theobald, 1968). The origin of this constructivist teaching approach, thus, lies in the failure of educators and instructional designers to distinguish between learner/novice and doer/expert and thus on the difference between teaching/learning science and doing science.
Keywords	Cognition Information processing Instructional strategies
Appendices

Authors
Name	Surname	Institution	Country	e-mail	EARLI Number	Presenting
Paul	Kirschner	University of Utrecht	Netherlands	P.A.Kirschner@uu.nl		*