Constructivism
	In the introduction to this paper, I described the view that knowledge could be divided into two forms, inert and non-inert knowledge (Bransford et al. (1990) talk about inert knowledge but they do not give a term to describe the type of knowledge that can be used for problem solving. While non-inert knowledge is used here, active knowledge might well have been a more suitable term). The basic difference between the two being that inert knowledge is not transferable and is not used for problem solving. It was argued that this 'non-inert' knowledge was stored in memory as a 'mental model' and that the manipulation of this model enabled problem solving to take place. It was also suggested that the information contained in this model could take the form of any sensory information and not just confined to visual information. It was also stated that there were mentions of mental models in the literature and that the theory of semantic networks was compatible with it. A semantic network is a knowledge construction in memory in which information is held in 'nodes' and that these nodes are extensively linked to produce a network. The more nodes that exist within a network, the more effective that network is. I would argue that the network is the mental model and that the theory explains how such a model is constructed (see Jonassen 1988, 13-16). While semantic network or schema theory attempts to explain how information is held and organised in long-term memory, it does not explain how it is acquired in the first place. The following are some of my personal ideas, based on reading, discussion and thinking, as to how this is accomplished. Learning involves the manipulation of information in some way, and so learning cannot take place without an input of information. There would appear to be three ways in which this information can be obtained. The first is from long term memory. When involved in discussion, argument or problem solving, the individual takes information which is already in memory, but which may form a part of a different semantic network and manipulated to arrive at the required solution. This solution would, presumably form part of a new semantic network. New information can be obtained from the external environment and there can be two possible sources of it; I will call these 'primary' and 'secondary' sources. Primary information sources are those which we experience directly from observation, experimentation and so on. These are very rich sources of information and it is up to the learner to decide which elements of it are going to be accepted. The learner has to manipulate and interpret this information in order to make some 'sense' out of it, a process which would result in relevant information being added to the semantic network. Secondary information is material which has been processed by another individual and is presented to the learner. There are many such sources, for example, teachers, books, television programs, conversations, and so on. Here, some other person has obtained the information from either some primary source, or more often a secondary one, and, hopefully, entered it into their semantic network (this is not necessarily so in the case of written information, as the writer can simply transfer the writings of another person on to their own piece of paper without it having entered much more than their short term memory). This person processes the information according to such things as their own level of motivation, interest, and prior knowledge before it enters their semantic network. When they then write a book, produce a television program, or talk to someone else, they extract some of this processed information and pass it on to the learner. Learning using information from a primary source is much more difficult as the learners have to provide their own processing and interpretation of it whereas secondary information is 'pre-digested' as it has already been processed at least once and the interpretations of at least one other person applied to it. Because of this, while it might be easier for the learner to absorb, it is information poor as much of the original information has been stripped from it. The problem with this is that, to be effective, a mental model must be as complete as possible, and this means that it should be compiled from information-rich sources. If this cannot be done, then it should at least be compiled from several secondary sources so that there is the possibility that some of the information from one source lost during processing is present in the others. A further way of presenting more information-rich material is to use a 'multi-media' format so that it may be processed in a number of different ways. Unfortunately, much of the current education system is based on secondary sources of information. Teachers receive much of their information from secondary sources, further process it and pass it on to their students; the authors of textbooks do the same thing, and so on. Unfortunately, there is no real alternative to this situation as each individual has not got the time to obtain all the information that they need from primary sources. Even in traditional communities where the knowledge base would be much smaller than in our society, secondary knowledge was passed from parent to children. During the course of their lives, these children would gain knowledge from primary sources and pass this, plus much of the secondary knowledge that they gained from their parents to their children. In our society, with its ever increasing rate of knowledge generation, not only is there less opportunity to learn from primary sources, but the total amount of knowledge that an individual can gain becomes an increasingly small fraction of the total available knowledge. The result of this is that the education system must attempt to compensate for the lack of primary knowledge acquisition and also concentrate on information acquisition and processing skills rather than just learning. Beginning to learn a new subject is always a slow and frustrating process, and the more dissimilar the learning task is to those previously attempted by the learner, the slower and more frustrating it is. While Gagné would claim that this results from the inability to chain various stimulous-response events, from the point of view of semantic networks it could be argued that, as one's knowledge represents one large semantic network, a new network has to be linked to the existing one, and when the information is new to the learner, it is much harder to establish the links necessary to do this. When the learning process is taking place, it could be argued that the following processes are taking place. Firstly, a chunk of information is taken into short term memory. Two interlinked processes then take place. The first is an attempt to examine the existing semantic network(s) in an attempt to establish links. It should be noted that this search is totally different to the way in which a computer searches a data base for links. The computer has to conduct a linear search which is time consuming whereas the brain seems to have the ability to conduct a non-linear, global search of long term memory in an almost instantaneous manner. The result of this search for links is that ideas and associations "pop" into the learner's mind. At the same time, the learner attempts to understand the information being presented and this process is assisted by the additional information being received by the "links search". Presumably, "understanding" of this information is only gained if links to existing knowledge can be identified. If they can, then this new information can be attached as a new node of the semantic network. This link would, however, only appear to be a temporary one as, if it is not accessed again within a comparatively short period of time, it will be forgotten (but do we forget it, or is there simply a hierarchy of nodes and a non-access of a particular node simply pushes it further down the hierarchy? Every one has probably experienced the situation where they have not been able to recall a name that has not been used for several years, for example, only to remember it some hours afterwards). Not every learner bothers to identify links between new information and existing knowledge. If this is not done, then presumably, the new knowledge will not be incorporated into a semantic network. What sort of learner would fail to do this? An unmotivated and uninterested one. Therefore, it can be argued that this process is moderated by interest and motivation, and so for learners to learn the information in an appropriate way, they must be motivated and interested. One way to ensure that learners are interested in the material is to teach it in context so that the learners can see the relevance and importance of the material being presented. Jonassen (1990, 33) states that "The most common instructional sin of educators and designers ... is that of decontextualising learning". He claims that knowledge is best gained within the "context of some real-world problem or experience" and that "The most effective learning contexts are those which are problem- or case-based and activity-oriented, that immerse the learner in the situation requiring him or her to acquire skills or knowledge in order to solve the problem or manipulate the situation". He notes, however, that much of the teaching that takes place in schools is stripped of its contextual relevance and presented as truth or reality. He continues: Our youth are daily subject to acquiring countless facts and rules that have no importance, relevance, or meaning to them because they are not related to anything the learners are interested in or need to know (1991, 36). The ability to rapidly link to information is illustrated by the effectiveness of word association, and this also demonstrates how these links can branch from one to another. Jonassen described how an attempt was made to make a cognitive map of instructional designers. On first reading, my impression was that the result was not valid as it was simply a matter of word association. On consideration, however, it became apparent that, if information is held in a semantic network, then word association would give an indication as to how this information was organised within the network (see Jonassen 1989, 7-10). How does this apply to computer based instruction? If a Behaviourist approach is taken, then the learner is expected to learn a specific response when exposed to a given stimulus. The result of this will be a change in behaviour as the response pattern of the learner will have changed. The Behaviourist, having adopted a positivist stance, will want to measure this change in behaviour so as to demonstrate the success of the learning process, and it is sound scientific principle to establish these learning outcomes before the learning process takes place. In this way, an objective outcome will be achieved. Such learning outcomes are expressed through a series of Behavioural Objectives and if they are met, the learning process has been successful. What in fact the behaviourist teacher is doing is saying to the students "Here is a body of knowledge, but your knowledge will be tested in the following areas". The students, being sensible people, will concentrate on the given objectives as, if they are to be given a scientifically-valid test, these are the only areas that they will be examined resulting in the high probability that their knowledge will be fragmented. While students may well gain full marks in the test, they will not be able to form a comprehensive semantic network or mental model from the information gained as it will exist as a series of un-linked or poorly-linked nodes. As a comprehensive semantic network will not have been formed, this knowledge will be inert and will not be used for problem-solving. A constructivist approach to learning, should, however, have the potential to allow learners to construct a well-linked semantic network or mental model. In essence, the constructivist states that knowledge is constructed by the learner and that this construction process involves the use of existing knowledge, level of understanding, and attitudes of the learner. To be more precise, Jonassen defines constructivism as being: ... the belief that knowledge is personally constructed from internal representations by individuals using their experience as foundation. Knowledge is based upon individual constructions that are not tied to any external reality, but rather to the knower's interaction with the external world. Reality is to a degree whatever the knower conceives it to be (1990, 32). It is interesting to note that the term 'belief' is used in this definition as it would be more normal to use a term such as 'theory'. 'Belief' brings with it a different feeling to 'theory', but as theory is an integral part of the positivist jargon, it might well be out of place in this context. Placing the first part of this definition in terms of semantic network theory, the learning process involves the creation of a semantic network which is linked to the existing semantic network(s) of an individual. As links are going to be created to existing knowledge, then it is reasonable to assume that the new network will be built within the constraints of the existing one and so the new knowledge representation (network) will have its foundations in the prior knowledge and experience of the learner. Learning takes place within the brain. The brain is isolated from the external world by a shield of bone, a layer of fluid and a covering of membrane. The only links that the brain has within the external environment are the senses and it is dependent upon these for its information. While there may be 'average' people, it can be argued that there is no such thing as a 'normal' person. We all exhibit variations of one sort or another and, while most of us come close to the norm, a small number of people are to be found some considerable distance from it. If this is true for such things as height and weight, presumably it is also true to the way in which the sensory organs of the body collect and interpret information. The only reality for an individual is an interpretation of the information that reaches the brain. Reality for a colour blind person must be different to that perceived by a normal sighted individual. Reality for a deaf person must be very much different to that perceived by a hearing person. Finally, a person with a serious mental illness will have a totally different perception of reality to a 'normal' person. It seems reasonable to suppose, therefore that individuals experience their own reality and that knowledge will be processed and constructed according to this reality. In the words of Jonassen (1990, 33) "Constructivists believe that if knowledge is constructed by the individual, then there is no 'objective reality' and that our own experiences determine our reality". Cunningham (1991, 26) adds another dimension, social context, to this process. He argues that: At the heart of constructivism is the notion that knowledge is constructed, which in the present instance means that our theoretical views are personal creations embedded in a social context, within a social community that accepts the assumptions underlying the perspective. Consider the statement Water is composed of two atoms of hydrogen and one atom of oxygen; this has been accepted for a long period of time and countless experiments have been performed to substantiate it. At least at our present level of knowledge, this could be supposed to be a truth and presumably would be accepted by anyone who has been exposed to modern chemistry. However, if we take the statement Russia is a dangerous country, while it could be claimed that this is only a matter of opinion, it has become so embedded in western culture that it would probably, a few years ago, have been regarded as 'knowledge'. The statement would have, presumably, been accepted by people in western countries and rejected by people in communist states. The acceptance of this statement would have been governed by the social context in which it was made. Societies change with time, and if this statement was made today, its acceptance could be the complete opposite of what it would have been a few years ago. So, not only can knowledge be constructed in a social context, but this construction may well change with time. Finally, a statement such asChrist is the son of God would only find acceptance on a cultural basis. Therefore, knowledge is constructed not only in the light of the existing knowledge and experience of the learner, but also the societal and cultural context in which it is placed, and both of these may vary with time. The introduction of constructivist paradigms into education brings with it several problems. The first of these is that most educators were educated within a behaviourist environment and basically they still see themselves as being a dispenser of knowledge rather than as a facilitator of learning. They also bring with them the concept that a certain amount of prescribed content must be taught. This situation has got to change. The first reason for this is that, at least in primary schools in New South Wales, the curriculum appears to be at bursting point. As soon as it is suggested to teachers that a new topic or subject should be taught, the immediate response is "How are we going to fit it in?". The second reason is that with the current extremely high rate of knowledge generation (which is going to increase even more in the future), less and less is going to be taught about more and more if we continue with this trend. The third is that methods of knowledge storage and retrieval are undergoing radical changes and the children entering school this year will not enter the work force until about 2005 and they will enter a work force that has very different needs to that of today. As Smith and Westhoff (1992, 15) state: Workers are citizens in the emerging information age must be able to identify problems, seek appropriate information and test solutions. Obviously, today's students should be prepared to take their place in such a society and Smith and Westhoff suggest that this can be done by breaking down the artificial barriers that currently divide knowledge into discrete subject areas and that students are given the ability to solve real-world problems (1992, 15). Constructivism would appear to be an instructional paradigm which would go a long way to achieving this. Later in this paper the Jasper series will be described. This is an example of Anchored Instruction and takes a constructivist stance. The series is devoted to teaching complex, real-world problem solving and, while it is mathematically based, links have been established so that consideration of other subject areas can be given. A second, and arguably the most important, problem with the introduction of a constructivist outlook into the present education system is that of assessment. One of the major goals of education is to assess students so that they can be appropriately pigeon-holed. If a behaviourist stance is taken by teachers, then this task becomes very easy. The objectives are described before the instructional sequence commences and it is then simply a matter of assessing how well students have achieved these. The constructivist approach, however, makes this task much more difficult as the knowledge construction of the student has to be assessed. One of the first considerations is as to what knowledge is constructed by learners. The case has been made, so far, that the learner adds new knowledge to a semantic network thus creating a mental model. It is further argued that this can be used in the problem-solving process by manipulating it. However, is all knowledge constructed. What about knowledge which is obtained by rote learning? If this is the case, Noddings (1990, 12-14) claims that assessment is not just a case of ascertaining if knowledge has been constructed or not but rather how well it has been constructed. In view of this, she suggests that we should refer to 'weak' and 'strong' knowledge construction. In fact, this should not present a great problem as the teacher should be aware of which parts of the content has been learned by rote and which part has been constructed, and so use appropriate assessment instruments to test the rote learning. It is suggested by Perkins (1991b, 19-21) that constructivist learning should be assessed by open test instruments. While this appears to be a valid idea, there are problems associated with it. Teachers are experienced in the design of the type of closed assessment instruments required for testing the attainment of behavioural objectives, but they would have to learn to write open assessment instruments. The marking of open instruments is likely to be much more time consuming and, unfortunately, today's computers cannot help effectively in this task. While the debate and discussion about constructivism can become very deep and beyond the scope of this paper, a point made by Glasserfield (1990, 19) is worthy of note: ... those consrtructivists who are "radical" because they take their theory of knowing seriously frequently meet the same objection [the proof that what they assert is true] ... it is difficult to show the critics that what they demand is the very thing constructivists must do without. To claim that one's theory of knowing is true, in the traditional sense of representing a state or feature of an experiencer-independent world, would be perjury for a radical constructivist. One of the central points of the theory is precisely that this kind of "truth" can never be claimed for the knowledge (or any piece of it) that human reason produces.
	Conclusion
	Constructivism would appear to satisfy the reqirements of the 'mental models' concept. The learning of material in a constructivist environment would have a much greater potential in generating a mental model that learning in a behaviourist environment. In fact, a behaviourist approach would appear to prevent the development of a complete mental model as the use of behavioural objectives would tend to fragment knowledge rather than bringing it together as a whole. In this section, there has been no attempt made to apply this theory to the practical production of CBI materials. This will be addressed in the next section when four examples of the use of computer technology to produce constructivist learning environments will be considered. Unfortunately, these concepts are the work of a small number of individuals or groups and so the descriptions rely heavily on a small number of sources.

	Server: http://scs.une.edu.au Author: Tony Brown Created: 25.6.97 © The University of New England, NSW, Australia