Using Computer Simulations for Learning*
Ton de Jong -
Wouter R. van
Joolingen -
Faculty of Educational Science and Technology, University of Twente
* This text is based on an internal report by T. de Jong and W.R. van Joolingen entitled: Discovery learning with Computer Simulations of Conceptual Domains. University of Twente, Faculty of Educational Science and Technology, IST-MEMO-96-02. For references please order this report by sending an email to the first author: jong@edte.utwente.nl
In the field of learning and instruction we now see an impressive influence of the so-called "constructivistic" approach. In this approach a strong emphasis is placed on the learner as an active agent in the knowledge acquisition process. As in the objectivistic tradition, where developments were followed and encouraged by the computer based learning environments, such as programmed instruction, tutorials, and drill and practice programs (Alessi & Trollip, 1985), we also find within the constructivistic approach computer learning environments that help to advance developments. Examples are hypertext environments (see e.g., Gall & Hannafin, 1994), concept mapping environments (see e.g., Novak & Wandersee, 1990), simulations (De Jong, 1991; Reigeluth & Schwartz, 1989), and modeling environments (e.g., diSessa & Abelson, 1986; Riley, 1990; Smith, 1986). In this contribution we concentrate on the use of computer simulations for learning because learning with simulations is closely related to a specific form of constructivistic learning, namely scientific discovery learning.
Computer simulations are programs that contain a model of a system (natural or artificial, e.g., equipment), or a process. Computer simulations can broadly be divided into two types: simulations containing a conceptual model, and those based on an operational model. Conceptual models hold principles, concepts, and facts related to the (class of) system(s) being simulated. Operational models include sequences of cognitive and non-cognitive operations (procedures) that can be applied to the (class of) simulated system(s). Examples of conceptual models can be found in economics (Shute & Glaser, 1990), and in physics (e.g., electrical circuits, White & Frederiksen, 1989; 1990). Operational models can, for example, be found in radar control tasks (Munro, Fehling, & Towne, 1985). Operational models are generally used for experiential learning, in a discovery learning context we mainly find conceptual simulations. Conceptual models still cover a wide range of model types such as qualitative vs. quantitative models, continuous vs. discrete, and static vs. dynamic models (see Van Joolingen & De Jong, 1991a). Models may also differ considerably in complexity, and range from very simple straightforward models, e.g., simple Mendelian genetics (Brant, Hooper, & Sugrue, 1991) to very complex models, e.g., the medical simulation HUMAN (Coleman & Randall, 1986) in which 200 variables and parameters can be changed. Also, specific characteristics like the place of variables in the model, or the distance between theoretical and operational variables characterise the conceptual model (Glaser, Schauble, Raghavan, & Zeitz, 1992). In the practice of instruction we find a large number of simulations both with operational and/or conceptual models. For operational models the advantages of using simulation lay in the practical area. Learners can practice under safe, non threatening circumstances. The way of learning with operational models is quite often denoted as experiential learning. For conceptual models the associated mode of learning is discovery learning. In scientific discovery learning the main task of the learner is to infer the characteristics of the model underlying the simulation. The learners’ basic actions are changing values of input variables and observing the resulting changes in values of output variables (De Jong, 1991; Reigeluth & Schwartz, 1989). Originally, the means of giving input to and receiving output from simulation environments were rather limited, but now increasingly sophisticated interfaces using direct manipulation for input, and graphics and animations as outputs, are emerging (e.g., Härtel, 1994; Teodoro, 1992; Kozma, Russel, Jones, Marx, & Davis, 1996) with as the latest development virtual reality environments (see e.g., Thurman & Mattoon, 1994). Theories on scientific discovery learning are usually based on theories of scientific discovery. Rivers and Vockell (1987), for example, describe a plan (design experiment), execute (carry out experiment and collect data), and evaluate (analyze data and develop hypothesis) cycle. Friedler, Nachmias, and Linn (1990) say that scientific reasoning comprises the abilities to "(a) define a scientific problem; (b) state a hypothesis; (c) design an experiment; (d) observe, collect, analyze, and interpret data; (e) apply the results; and (f) make predictions on the basis of the results." (p. 173). De Jong and Njoo (1992) added the distinction between transformative processes (processes that directly yield knowledge such as the ones mentioned by Friedler et al., and Rivers & Vockell) and regulative processes (processes that are necessary to manage the discovery process such as such as planning and monitoring).
2. Effectiveness of simulation based learning
In an early overview on computer-based education, Bangert-Drowns, Kulik, and Kulik (1985) report that simulation based learning does not raise examination scores. Later studies that contrasted (sometimes as part of a larger set of comparisons) learning from "pure" simulation (containing conceptual models) with learning from some form of expository instruction (computer tutorial, classroom) cover a variety of domains, such as biology (Rivers & Vockell, 1987), economics (Grimes & Willey, 1990), Newtonian mechanics (Rieber, Boyce, & Assad, 1990; Rieber & Parmley, 1995), and electrical circuits (Carlsen & Andre, 1992; Chambers et al., 1994). Sometimes the single simulation is compared to expository instruction (Rieber & Parmley, 1995), but quite often a comparison is made between a simulation embedded in a curriculum or expository instruction and the curriculum or expository instruction as such (Carlsen & Andre, 1987; Chambers et al., 1994; Grimes & Willey, 1990; Rieber et al., 1990; Rivers & Vockell, 1987). Also, in some cases, the expository instruction to which the simulation is compared is "enhanced", e.g., by "conceptual change features" (Chambers et al., 1994) or by questions (in one condition of Rieber et al., 1990). As an overall picture, favourable results for simulation based learning are reported in the study by Grimes and Willey (1990), and no difference between simulation based learning and expository teaching is reported by Carlsen and Andre (1992), and Chambers et al. (1994). A mixture of favourable and no difference results was found between several sub-studies by Rivers and Vockell (1987). In Rieber et al. (1990) the group of students receiving a simulation in addition to a tutorial scored higher on a test measuring "application of rules" than the tutorial only group, but scored at the same level as a tutorial group that received additional questions while learning. In Rieber and Parmley (1995) subjects who received only an unstructured (pure) simulation fell short of the performance of subjects receiving a tutorial.
The general conclusion that emerges from these studies is that there is no clear and univocal outcome in favour of simulations. An explanation why simulation based learning does not improve learning results can be found in the intrinsic problems that learners may have with discovery learning. In the above mentioned studies, Chambers et al. (1984), for example, analysed the videotapes of students working with the simulation and noticed that students were not able to deal with unexpected results and that students did not utilise all the experimenting possibilities that were available. Also studies that compared learning behaviour of successful and unsuccessful learners in simulation learning environments (e.g., Schauble, Glaser, Raghavan, & Reiner, 1991) have pointed to specific shortcomings of learners. For this reason, in a number of studies, additional instructional measures are suggested to help learners overcome the problems that they may have with scientific discovery learning.
3. Problems that learners encounter in discovery Learning
Finding new hypotheses is generally recognised as a difficult process (Chinn & Brewer, 1993), that clearly distinguishes successful and unsuccessful learners (Schauble, Glaser, et al., 1991). An important problem here is that learners (even university students) simply may not know what a hypothesis should look like. A second problem is that learners may not be able to state or adapt hypotheses on the basis of data gathered. A third problem in stating hypotheses is that learners can be led by considerations that not necessarily help them to find the correct (or best) theoretical principles. Van Joolingen & De Jong (1993) describe a phenomenon that they called fear of rejection.
A crucial aspect of scientific discovery is the design of experiments that provide information for deciding upon the validity of an hypothesis. In case that a learner does not yet have a hypothesis, well designed experiments can be used to generate ideas about the model in the simulation. In literature we find a number of phenomena that point to learners who use poorly designed experiments. The first phenomenon, confirmation bias, is the tendency to seek for information that confirms the hypothesis they have, instead of trying to disconfirm the hypothesis. The second phenomenon describes learners who design inconclusive experiments. In the context of discovery learning with simulations, Glaser et al. (1992), for example, point to a frequently observed phenomenon that learners tend to vary too many variables in one experiment, resulting in that they cannot draw any conclusions from these experiments. A third phenomenon is that subjects show inefficient experimentation behaviour. For example, Kuhn et al.(1992) found that subjects did not use the whole range of potential informative experiments that were available, but only a limited set, and moreover designed the same experiment several times. A fourth phenomenon describes learners that construct experiments that are not intended to test a hypothesis. Schauble, Klopfer, and Raghavan (1991) identified what they have called the "engineering approach", which denotes the attitude to create some desirable outcome instead of trying to understand the model.
Once having performed correct experiments, data that come from these experiments needs to be interpreted before the results from the experiments can be translated into hypotheses on the domain. According to Schauble, Glaser, et al. (1991) successful learners are more proficient in finding regularities in the data than unsuccessful learners. Klahr et al. (1993) found that subjects made misencodings of experimental data ranging from a mean of 35% of at least one misencoding, to a high 63% depending on the type of actual rule involved. Also the interpretation of graphs, a frequently needed skill when interacting with simulations, is clearly a difficult process.
For regulative processes it is frequently reported that successful learners use systematic planning and monitoring, whereas unsuccessful learners work in an unsystematic way (e.g., Lavoie & Good, 1988; Simmons & Lunetta, 1993). Shute and Glaser (1990) claim that successful learners plan their experiments and manipulations to a greater extent, and pay more attention to data management issues. Glaser et al. (1992) report that successful discoverers followed a plan over experiments, whereas unsuccessful ones used a more random strategy, concentrating at local decisions, which also gave them problems to monitor what they had been doing (see also Schauble, Glaser, et al., 1991). Though Glaser et al. (1992) mention persistence to follow a goal as a characteristic of good learners, these successful subjects also were ready to leave a route when it apparently would not lead to success. Goal setting is also reported as a problem (for subjects with low prior knowledge) by Charney, Reder, and Kusbit (1990). In a more general way Veenman and Elshout (1995) found that, over a number of studies, individuals with a high intellectual ability showed a better working method than individuals with a low intellectual ability, but also that working method had its own contribution to learning outcome on top of intellectual ability. For the process of monitoring differences between successful and unsuccessful learners are reported by Lavoie and Good (1988) who found that good learners make more notes during learning, and by Schauble, Glaser, et al. (1991) who found a more systematic data recording for successful learners.
4. Technological solutions to learning problems
In the current section we summarise a number of methods to support learners in the discovery process. The first means of support we describe is to provide the learner with direct access to domain information. Subsequently, we present support measures that aim to support the learner in specific discovery processes.
Direct Access to Domain Knowledge
A frequently uttered claim about learning with simulations is that learners should already know something before discovery learning is to become fruitful. Insufficient prior knowledge might be the cause that learners do not know which hypothesis to state, can not make a good interpretation of data, and move to unsystematic experimentation behavior (Glaser et al., 1992; Schauble, Glaser, et al., 1991). Several authors have introduced access to extra information as a support measure in a simulation environment, quite often in the form of a (more or less sophisticated) hypertext/hypermedia system (Glaser, Ragahvan, & Schauble, 1988; Lajoie, 1993; Shute, 1993; Thomas & Neilson, 1995). Shute (1993) described an ITS on basic principles of electricity in which learners could ask for definitions of concepts (e.g. ammeter, ampere, charge, circuit, current ...) by selecting a term from a menu and follow hypertext links. Shute (1993) reports positive effects of use of this on-line hypertext dictionary on a composite post-test measuring declarative and conceptual knowledge, problem solving, and transfer of knowledge and skills. A number of authors point to the critical aspect of timing of the availability of information.
Information cannot only be provided by the learning environment, but must also be invoked from learners’ memory. Support measures can stimulate learners to confront their prior knowledge with the experimental outcomes. In order to achieve this, Lewis, Stern, and Linn (1993) provided learners with an electronic notation form to note down "everyday life examples" of phenomena they observed in a simulation environment (on thermodynamics).
A specific case of adding domain information is when the simulation visualises aspects that are normally not visible. For example in simulation on motion stroboscopic traces of the motion can be shown, (White, 1984; Teodoro, 1992) or velocities and accelerations of objects can be indicated by adding sizable vector to objects (Härtel, 1994; Teodoro, 1992). Rieber and Kini (1991) explain the effectiveness of these additions to reality in terms of Paivo’s dual coding theory.
Support for Hypothesis Generation
Hypothesis generation is a central process in discovery learning. Several studies have created support to overcome the problems that learners have with this process. Smithtown (Shute & Glaser, 1990) a simulation learning environment in the field of economics offers the learner support for hypothesis generation by means of a hypothesis menu. This menu consists of four windows which present parts of a hypothesis e.g., variables, verbs to indicate change, and connectors. A similar means of support is a hypothesis scratchpad (Van Joolingen & De Jong, 1991b; 1993). Here, learners are offered different windows for selecting variables, relations, and conditions. These two approaches offer learners elements of hypotheses that they have to assemble themselves. A more directive support for creating hypotheses can be found in CIRCSIM-TUTOR (Kim, Evans, Michael, & Rovick, 1989), an ITS in the domain of medicine which treats problems associated with blood pressure where students are asked to state qualitatively what will happen to seven components of the cardio-vascular system. To be able to write this down learners are offered a predefined spreadsheet. One step further is to offer learners complete hypotheses. In "Pathophysiology Tutor" (PPT) (Michael, Haque, Rovick, & Evens, 1989) learners can select from a list of predefined hypothesis, ordered in nested menus providing lists of hypotheses in the field of physiopathology.
Support for the Design of Experiments
To support a learner in designing experiments the learning environment can provide experimentation hints. Hints can be generated dynamically on the basis of the actual experimentation behavior of learners. Hints are then presented if a learner displays non-optimal learning behavior. An example of a system containing this type of hints is Smithtown (Shute & Glaser, 1990). Leutner (1993) studied the effect of providing learners with adaptive advice of this kind. He found that if the advice has a limited character it helps to increase the learner’s domain knowledge, but hinders the acquisition of functional knowledge. After giving more detail to the advice it also helped to increase the functional knowledge This is knowledge that helps learners to reach an optimal result in the simulation), though the effect is less clear since it was combined with giving extra domain information.
Support for Making Predictions
While a hypothesis is a statement on the relations between variables in a theoretical model, a prediction is a statement on the value(s) of a dependent variable under the influence of values of the independent variable(s) as they can actually be observed in the simulation. One specific way to help learners express predictions is to give them a graphing tool in which they can draw a curve that depicts the prediction. Lewis et al. (1993) provided learners with such a tool. Feedback is given to learners by drawing the correct curve in the same diagram in which the learner’s prediction was drawn. Tait (1994) describes a similar mechanism, but in his case feedback also includes explanations of the differences between the system’s and the learner’s curve. Reimann (1991) who describes an environment on the refraction of light provided learners with the opportunity to give predictions at three levels of precision: as numerical data, as a drawn graph, and as an area in which the graph would be located.
Support for Regulative Learning Processes
Regulative processes are the processes that manage the learning process. Regulative aspects such as "planfulness and systematicity" are regarded to be central characteristics of successful discovery learning (Glaser et al., 1992; Schauble et al., 1995). The two most central regulative processes are planning and monitoring (De Jong & Njoo, 1992). Planning and monitoring are both supported by introducing model progression in the simulation environment. Next to model progression, we found specific measures for supporting planning or monitoring. Finally, regulative processes can be supported by structuring the discovery process.
Model progression. The basic idea behind model progression is that presenting the learner with the full complexity of the simulation at once may be too overwhelming. In model progression the model is introduced gradually, step by step. White and Frederiksen’s (1989; 1990) work on QUEST is one of the best known examples where the idea of model progression has been applied. QUEST treats electrical systems and models of electrical circuits in QUEST differ in their order (qualitative or quantitative models), degree of elaboration (number of variables and relations between variables), and perspective. While learning with QUEST, learners are confronted with models that advance from a qualitative to a quantitative nature, that are more elaborated, and that transform from a functional to a physical perspective. In this respect the instructional sequence follows the (assumed) transition from a novice knowledge state to an expert one. As far as we know, no controlled evaluation of QUEST has been undertaken. Model progression in which the model increases in complexity for the learner was studied in Swaak, Van Joolingen, and De Jong (1996). SETCOM is a simulation on harmonic oscillation where the model develops from free oscillation, through damped oscillation to oscillation with an external force. Swaak et al. (1996) found that model progression was successful in enlarging the students’ intuitive knowledge (but not their conceptual knowledge) as compared to an environment without model progression. In a study in a different domain, but within the same type of environment, De Jong et al. (1995) could not find effects of providing learners with model progression on top of giving them assignments.
Planning support. Planning support may, as Charney et al. (1990) have postulated, be especially helpful for subjects who have low prior knowledge. Planning support takes away decisions from learners and in this way helps them in managing the learning process. Support for planning can be given in different ways. Already quite early in the use of simulations for scientific discovery learning, Showalter (1970) recommended to use questions as a way to guide the learner through the discovery process. His questions (e.g. "Do rats ever reach a point at which they don’t learn more?", p. 49) focused the learners attention to specific aspects of the simulation. Zietsman and Hewson (1986) used similar types of questions in conjunction with a simulation on "velocity", and Tabak, Smith, Sandoval, and Reiser (1996) have added such questions with the aim of setting goals in a biological simulation. White (1984) helped learners to set goals in a simulation of Newtonian mechanics by introducing games. Games, as White uses them, ask learners to reach a specific state of the simulation (e.g. to get a spaceship in the simulation around a corner without crashing into any walls (p. 78). In an experiment White found that learners who learned with a simulation that contained games, outperformed learners who worked with the pure simulation on a test of qualitative problems (asking questions of the form "What would happen if ..?" or "How could one achieve ...?" (p. 81)). Also, in the ThinkerTools environment (White, 1993) games are used in a similar context as in White (1984). De Jong et al. (1994) describe different types of assignments that can be used in combination with simulations, among others investigation assignments that prompt students to find the relation between two or more variables, specification assignments that ask students to predict a value of a certain variable, and explicitation assignments that ask the student to explain a certain phenomenon in the simulation environment. In De Jong et al. (1995) using a simulation on collisions, Swaak et al. (1996) using a simulation on harmonic oscillation, and De Jong, Härtel, Swaak, and Van Joolingen (1996) using a simulation on the physics topic of transmission lines it was found that students (who were free to choose) used assignments very frequently, and that using assignments had a positive effect on gaining what they call "intuitive" knowledge.
Monitoring support. Support for monitoring one’s own discovery process can be given by overviews of what has been done in the simulation environment. Reimann (1991) provided learners in Refract with a notebook facility for storing numerical and nominal data from experiments. Data in the notebook could be manipulated so that experiments could be sorted on values for a specific variable, experiments could be selected in which a specific variable has a specified value, and an equation could be calculated over experiments. Also the student could replay experiments from the notebook. Similar notebook facilities are present in Smithtown (Shute & Glaser, 1990) and Voltaville (Glaser et al., 1988). In SHERLOCK learners can receive upon request an overview of all the actions they have taken so far (Lesgold, Lajoie, Bunzo, & Eggan, 1992). Schauble, Raghavan, and Glaser (1993) presented monitoring support that not only provided an overview of students’ actions, but also offered the opportunity to group actions under goals, and to ask for an "expert view" that gives the relevance of the student’s actions in the context of a specific goal (e.g. to find the relation between two variables). This support in fact combines monitoring and planning support. In all the examples presented here, learners have to select previous experiments for comparison from the complete set of experiments themselves. Reimann and Beller (1993) propose a system (CABAT) that selects previous experiments on the basis of similarity and proposes this experiment to the learner for comparison.
Structuring the discovery process. Regulative processes can also be supported by leading the learner through different stages of the process. Several studies have compared the effects of structured environments (where structuring is quite often combined with several other measures) with "unstructured environments". Linn and Songer (1991) found providing students with a sequence of experimentation steps ("before doing the experiment", "now do the experiment", "after doing the experiment") and with more detailed directions in each of these steps was effective. They report that up to two and four times as many students were able to distinguish between central concepts from the domain (heat and temperature) compared to a version that was not structured. White (1993) in her ThinkerTools environment forced subjects to follow a four phases sequence of activities of "asking questions, doing experiments, formulating laws, and investigating generalisations" (White, 1993, p. 53), and provided more detailed indications in each phase. White found a clear advantage for a simulation based curriculum compared to a traditional curriculum on a test that measured qualitative predictions in real-world situations.
5. Examples of technology driven environments
In this section we give a selective overview of a number relevant technological environments. We make a distinction between discovery environments that also provide instructional support (integrated learning environments), modelling environments, that are (easy to use) environments (for learners) for making models of domains and testing their ideas, and authoring environments, environments especially aimed for making instructional (discovery) environments.
5.1 Examples of integrated learning environments
Integrated learning environments are learning environments that offer a discovery part (a simulation) together with tools for supporting the discovery process.
Thinkertools
Thinkertools is a learning environment in the physics domain of dynamics (force and motion) centred around simulations (or microworlds as it is called). In Thinkertools learners are led through a sequence of learning phases (make predictions, do experiments, formulate laws, and investigate the generality of laws) and they can ask for ‘games’ (a kind of assignments). Finally, the domain is offered in steps (model progression). For a full description see White (1993).
LRDC environments
At LRDC in Pittsburgh several integrated learning environments were developed. Examples are: Smithtown (on economics), Voltaville (on electricity) and Refract (on refraction of light). These learning environments were among the first that offered learners support next to a simulation, such as for example a hypothesis notebook for noting down hypothesis, or a hypertext database for looking up relevant information.
Belvedere
Belvedere is a computer environment for collaborative work on theory building and revision. The basic idea of the environment is that students work together on building a theory in a specific field. The Belvedere environment offers a number of electronic tools to support this process of scientific discovery and theory building. These support tools both cover the collaborative and science building aspects. For example, students are able to store and compare their hypotheses, to make common "concept maps" of the domain, a database of discussions and notes, and to use "coaches" that for example present alternate explanations, gives clues to relevant information, or point to problematic arguments. Part of Belvedere is also the opportunity for students to create simulations and have discussions during the construction process and while running simulations. For more information:
http://advlearn.lrdc.pitt.edu/advlearn/teachers/ABOUTUS/BELVEDER.HTM
SimQuest environments
In the EC sponsored servive project an authoring system, named SimQuest, is developed that supports the creation of discovery learning environments. In SimQuest applications learners may ask for small exercises (so-called assignments) that help them plan their actions and that can point them to specific phenomena; while experimenting learners can ask for background information in the form of definitions, relations to the real world etc. (this can be any kind of multi-media material); the simulation model can be presented to the learner in small steps that increase the model in complexity (so-called model progression); learners have tools that help them to monitor what they have been doing in a simulation session, that help them replay simulation sessions, compare outcome series, and make sound interpretations of the data; and, finally, also learners will have tools that help learners to compose and check hypotheses. SimQuest learning environments currently exist in the areas of electricity, dynamics, static mechanics, and water purification.
For more information see:
http://www.to.utwente.nl/servive/public/simquest/applic/frame.htm
5.2 Examples of ‘modelling’ environments
Modelling environments are environments that help learners to build models and interact with them. Of course, there are many many modelling tools available. Here we give a short selection of modelling tools used in an education context.
Powersim
Powersim is a powerful general modelling environment based on system dynamics. The interface is much similar to Stella. Models are created in a graphical way by selecting and connecting elements of models and changing their properties. Interfaces are created fairly easy on top of the model.
http://www.powersim.com/Products/
20Sim
20Sim is a modeling and simulation environment for simulating the behavior of dynamic systems, such as electrical, mechanical and hydraulic systems or any combination of these systems. Models can be entered by using text (equations) graphics and block diagrams. 20Sim is based on Bond graphs and contains advanced integration algorithms.
For more information:
http://www.rt.el.utwente.nl/20sim/product/product.htm
Modellus
Modellus is an interactive mathematical modelling tool that can be used for modelling in many domains (e.g., physics, chemistry, economics etc.). Modelling can be done by typing in equations, interfaces of different kind (animations, bar charts, graphs etc.) are easily created. By changing characteristics of the model (while a simulation runs) and observe the results learners can start to understand the domain and the mathematics of the model.
For more information:
http://www.krev.com/Pages/Products/Mod/modellus.html
Interactive physics
A modeling tool specifically dedicated to physics models. Interactive physics has a dedicated graphical interface (containing elements such as damper, springs etc.) for creating systems in physics. The models created can be ran and interacted with dynamically.
http://www.krev.com/Pages/Products/IP/ip.html
Authoring environments are general tools for making learning applications. There are many of them around. Here we mention three of them, the first one being a general authoring tool that can also be sued for making simulations, the other two take simulations as their starting point.
Authorware
Authorware is a powerful authoring environment that basically is made for creating tutorial software. Using the built in script language or by including external sources simulations can be embedded. For more information see:
http://www.macromedia.com/software/authorware/
Lloyd Rieber wrote a book on making simulations in Authorware
http://itech1.coe.uga.edu/Faculty/lprieber/edu605.html
IMTS/RAPIDS/RIDES
Rides is an authoring environment in the family of IMTS/RAPID/RIDES developed at the University of Southern California. It is an integrated software environment for developing and delivering computer based tutorial instruction and practice in the context of graphical simulations. Using RIDES, authors are able to build interactive graphical models of complex systems and then swiftly build interactive lessons in the context of those graphical models. Students use RIDES to interact with authored lessons, and, if the author chooses, they are able to explore the graphical simulations freely.
For more information:
http://btl.usc.edu/rides/index.html
SimQuest
SimQuest is an authoring environment that enables authors (teachers) to create computer-based simulation learning environments. Characteristic of these learning environments is that they include instructional support for learners to cope with discovery learning. The authoring environment developed has the name SimQuest. The authoring approach taken in SimQuest is an object-oriented one. SimQuest offers access to libraries of (elements of) simulation models, interface and animations objects, instructional support objects, and testing objects. The basic aspects of the authoring task are selecting, instantiating, and specialising objects from the library. The SimQuest environment has been set-up in close co-operation with authors (teachers from vocational training) who were creating applications and from whom we gathered requirements. Simquest is an open environment, that allows the authoring process to be carried out in any order. The modelling process is being supported by ready made simulation models, the possibility to enter equations directly, and the possibility to include externally created simulations. The SimQuest interface enables the author to have different views on the application and to monitor the relations between components of the application. The SimQuest authoring environment provides authors with methodological support, operational support, and pedagogical support. Much of this support is ‘hidden’ from the author. For example, instructional building blocks (objects) in the libraries are in the form of templates that prescribe how instructional support measures (e.g., an assignment for the learner) should be designed. The simquest environment also contains explicit support for the author in the form of on-line pedagogical advice, methodological support, and on-line help and a user manual. For more information:
http://www.to.utwente.nl/servive/public/home.htm
In the field of discovery learning with computer simulations a number of areas need further investigation:
Effectiveness of individual instructional support measures
A further and deeper analysis of problems that learners encounter in discovery learning and the evaluation of specific ways to support learners is, in our view, the principal item on the research agenda in this area. Studies should aim to find out when and how to provide learners with means to overcome their deficiencies in discovery learning, in other words how to provide ‘scaffolding’ for the discovery learning process.
Effects of learning with complex learning environments
Introducing additional support tools is not only meant to enable the learner to perform certain actions, but also to prevent cognitive overload (Glaser et al., 1988, p. 63). However, some instructional measures may also raise cognitive load, by introducing more complexity into the environment. Gruber et al. (1995), for example, found that adding multiple perspectives to a simulation on an economics system was detrimental for students’ performance. Gruber et al. (1995) attributed this effect to an increasing cognitive load. In other studies, where cognitive load was measured directly during the learning process (De Jong et al., 1995; Swaak et al., 1996), no increase in cognitive load following the introduction of support measures (assignments, and model progression) could be found. Further research on support measures should take into consideration the effects of additional support measures on cognitive load.
Use of multiple representations
Simulation learning environments quite often use multiple representations (e.g., graphs, animations, tables) for displaying the results of running the simulation. The understanding and coordination of these multiple representations by learners is not very well understood phenomenon.
New assessment procedures
An important issue is the design of adequate measures of knowledge as it results from discovery learning. Knowledge can be categorised in a number of types and qualities (De Jong & Ferguson-Hessler, 1996). We have seen that in studies that report a higher effectiveness of discovery learning compared to expository teaching, knowledge was measured with tests that aimed at assessing knowledge with a more ‘qualitative’ or ‘intuitive’ character. In a recent study Swaak et al. (1996) found for a simulation on harmonic oscillations that discovery learning led to only a little increase in ‘definitional’ knowledge, but to a large increase in ‘intuitive’ knowledge as measured by a speed test in which students had to make qualitative predictions of simulation states.
Use of cognitive tools for (unobtrusively) monitoring learning processes
A secondary aspect of support tools is that in learning environments these tools can also be used for unobtrusive measures, as was already recognised by Glaser et al. (1988) in the design of Voltaville. For example, in SHERLOCK (Lesgold et al., 1992) the student goes through the diagnostic problem solving process by choosing from menu’s of actions. On the one hand this helps the student in the planning process, on the other hand this helps the researcher (the system) to assess the student’s intentions. In the SHERLOCK environments information from this ‘planning tool’ for the learner is utilised for generating adequate hints. Van Joolingen (1995) describes some principles of how information gathered through a hypothesis scratchpad can be used for assessing the learner’s actual state of knowledge.
Place of discovery learning in the curriculum
A important condition for simulation based discovery learning to be successful in practice is to give it the right place in the curriculum. Currently the literature is not conclusive. Lavoie and Good (1988) suggest that a ‘Piagetian’ approach should be used, which implies that simulations are introduced in a first phase of learning where exploration is allowed, that concepts are formally introduced later, finally followed by concept application. In fact, this approach has been implemented in the Thinkertools project (White, 1993). Brant et al. (1991) compared a group of students who received a simulation before formal instruction with a group of students receiving the reversed order. Their data showed that the subjects receiving the simulation before the formal instruction outperformed the subjects that received the simulation after the formal instruction on a test which required to apply principles from the simulation. Brant et al.’s conclusion that simulations should be offered before expository instruction was partially supported by data from Andre and Haselhuhn (1995).
Design processes of teachers and the provision of authoring tools to support authors in designing discovery environments
The aspects mentioned in this paper, plus additional ones, such as safety and motivation provide simulation with a high potential for gaining a prominent place in our educational programs. However, for a full scale introduction of simulation based discovery learning we think that, in addition to more information on how to adequately support learners, it is necessary that authoring tools will be introduced that not only support creation of the simulation itself, but also supply the author with means and support to easily create the support for learners. Professional designers and teachers are not aware of the principles of constructivistic, discovery, learning. Authoring tools should therefore not only give them technical support, but also conceptual support in creating pedagogically sound learning environments. There are now developments which are promising, for example in the IMTS/RAPIDS system (Munro, Johnson, Surmon, & Wogulis, 1993; Munro & Towne, 1992; Towne, 1995), and the SimQuest system (De Jong et al., 1994; De Jong & Van Joolingen, 1995; Van Joolingen & De Jong, 1997). Further development of these authoring tools should be based on a thorough understanding of the knowledge of design methodology and specific pedagogical knowledge of designers (teachers).
Only after sufficient research results along the lines sketched in this section will be available, an appropriate design theory for instructional simulations may arise. Current attempts, though interesting, are necessarily fragmentary and incomplete (see e.g., Thurman, 1993). Based on such a theory, discovery learning with simulations can take its place in learning and instruction as a new line of learning environments based on technology where more emphasis is being given to the learner’s own responsibility.
Alessi, S.M., & Trollip, S.R. (1985). Computer based instruction, methods and development. Englewood Cliffs, NY: Prentice-Hall.
Andre, T., & Haselhuhn, C. (1995, April). Mission Newton! Using a computer game that simulates motion in Newtonian space before or after formal instruction in mechanics. Paper presented at the Annual Meeting of the American Educational Research Association, San Francisco, CA.
Bangert-Drowns, R., Kulik, J., & Kulik, C. (1985). Effectiveness of computer-based education in secondary schools. Journal of Computer Based Instruction, 12, 59-68.
Brant, G., Hooper, E., & Sugrue, B. (1991). Which comes first the simulation or the lecture? Journal of Educational Computing Research, 7, 469-481.
Carlsen, D.D., & Andre, T. (1992). Use of a microcomputer simulation and conceptual change text to overcome students preconceptions about electric circuits. Journal of Computer-Based Instruction, 19, 105-109.
Chambers, S.K., Haselhuhn, C., Andre, T., Mayberry, C., Wellington, S., Krafka, A., Volmer, J., & Berger, J. (1994, April). The acquisition of a scientific understanding of electricity: Hands-on versus computer simulation experience; conceptual change versus didactic text. Paper presented at the Annual Meeting of the American Educational Research Association, New Orleans, LA.
Charney, D., Reder, L., & Kusbit, G.W. (1990). Goal setting and procedure selection in acquiring computer skills: A comparison of tutorials, problem solving, and learner exploration. Cognition and Instruction, 7, 323-342.
Chinn, C.A., & Brewer, W.F. (1993). The role of anomalous data in knowledge acquisition: A theoretical framework and implications for science instruction. Review of Educational Research, 63, 1-51.
Coleman, T.G., & Randall, J.E. (1986). HUMAN-PC: A comprehensive physiological model [Computer software]. Jackson: University of Mississippi Medical Center.
diSessa, A., & Abelson, H. (1986). Boxer: a reconstructible computational medium. Communications of the ACM, 29, 859-868.
Friedler, Y., Nachmias, R., & Linn, M.C. (1990). Learning scientific reasoning skills in microcomputer-based laboratories. Journal of Research in Science Teaching, 27, 173-191.
Gall, J.E., & Hannafin, M.J. (1994). A framework for the study of hypertext. Instructional Science, 22, 207-232.
Glaser, R., Raghavan, K., & Schauble, L. (1988). Voltaville, a discovery environment to explore the laws of DC circuits. Proceedings of the ITS-88 (pp. 61-66). Montreal, Canada.
Glaser, R., Schauble, L., Raghavan, K., & Zeitz, C. (1992). Scientific reasoning across different domains. In E. de Corte, M. Linn, H. Mandl & L. Verschaffel (Eds.), Computer-based learning environments and problem solving (pp. 345-373). Berlin, Germany: Springer-Verlag.
Grimes, P.W., & Willey, T.E. (1990). The effectiveness of microcomputer simulations in the principles of economics course. Computers & Education, 14, 81-86.
Gruber, H., Graf, M., Mandl, H., Renkl, & Stark, R. (1995, August). Fostering applicable knowledge by multiple perspectives and guided problem solving. Paper presented at the conference of the European Association for Research on Learning and Instruction, Nijmegen, The Netherlands.
Härtel, H. (1994). COLOS: Conceptual Learning Of Science. In T. de Jong & L. Sarti (Eds.), Design and production of multimedia and simulation based learning material (pp. 189-219). Dordrecht, The Netherlands: Kluwer Academic Publishers.
de Jong, T. (1991). Learning and instruction with computer simulations. Education & Computing, 6, 217-229.
de Jong, T., & Ferguson-Hessler, M.G.M. (1996). Types and qualities of knowledge. Educational Psychologist, 31, 105-113.
de Jong, T., Härtel, H., Swaak. J., & van Joolingen, W. (1996). Support for simulation-based learning; the effects of assignments in learning about transmission lines. In A. Díaz de Ilarazza Sánchez & I. Fernández de Castro (Eds.), Computer aided learning and instruction in science and engineering (pp. 9-27). Berlin, Germany: Springer-Verlag.
de Jong, T., & van Joolingen, W.R. (1995). The SMISLE environment: Learning with and design of integrated simulation learning environments. In P. Held & W.F. Kugemann (Eds.) Telematics for education and training (pp. 173-187). Amsterdam: IOS Press.
de Jong, T., van Joolingen, W., Scott, D., de Hoog, R. , Lapied, L., Valent, R. (1994). SMISLE: System for Multimedia Integrated Simulation Learning Environments. In T. de Jong & L. Sarti (Eds.), Design and production of multimedia and simulation based learning material (pp. 133-167). Dordrecht, The Netherlands: Kluwer Academic Publishers.
de Jong, T., van Joolingen, W.R., Swaak, J., Veermans, K., Limbach, R., King, S., & Gureghian, D. (1998). Combining human and machine expertise to foster self-directed learning in simulation-based discovery environments. Journal of Computer Assisted Learning.
de Jong, T., Martin, E., Zamarro J-M., Esquembre, F., Swaak, J., & van Joolingen, W.R. (1995, April). Support for simulation-based learning; the effects of assignments and model progression in learning about collisions. Paper presented at the Annual Meeting of the American Educational Research Association, San Francisco, CA.
de Jong, T., & Njoo, M. (1992). Learning and Instruction with computer simulations: learning processes involved. In E. de Corte, M. Linn, H. Mandl & L. Verschaffel (Eds.), Computer-based learning environments and problem solving (pp. 411-429). Berlin, Germany: Springer-Verlag.
van Joolingen, W.R. (1995). QMaPS: Qualitative reasoning for intelligent simulation learning environments. Journal of Artificial Intelligence in Education, 6, 67-89.
van Joolingen, W.R., & de Jong, T. (1991a). Characteristics of simulations for instructional settings. Education & Computing, 6, 241-262.
van Joolingen, W.R., & de Jong, T. (1991b). Supporting hypothesis generation by learners exploring an interactive computer simulation. Instructional Science, 20, 389-404.
van Joolingen, W.R., & de Jong, T. (1993). Exploring a domain through a computer simulation: traversing variable and relation space with the help of a hypothesis scratchpad. In D. Towne, T. de Jong & H. Spada (Eds.), Simulation-based experiential learning (pp. 191-206). Berlin, Germany: Springer-Verlag.
van Joolingen, W.R., & de Jong, T. (1996). Design and implementation of simulation-based discovery environments: the SMISLE solution. Journal of Artificial Intelligence and Education, 7, 253-277.
van Joolingen, W.R., & de Jong, T. (1997). An extended dual search space model of learning with computer simulations. Instructional Science, 25, 307-346.
Kim, N., Evens, M., Michael, J.A., & Rovick, A.A. (1989). CIRCSIM-TUTOR: An intelligent tutoring system for circulatory physiology. In H. Maurer (Ed.), Computer Assisted Learning. Proceedings of the 2nd International Conference ICCAL (pp. 254-267). Berlin, Germany: Springer-Verlag.
Klahr, D., Fay, A.L., & Dunbar, K. (1993). Heuristics for scientific experimentation: A developmental study. Cognitive Psychology, 25, 111-146.
Kozma, R.B., Russell, J., Jones, T., Marx, N., & Davis, J. (1996). The use of multiple, linked representations to facilitate science understanding. In S. Vosniadou, E. De Corte, R. Glaser & H. Mandl (Eds.), International perspectives on the design of technology supported learning environments (pp. 41-61). Hillsdale, NJ: Erlbaum.
Kuhn, D., Schauble, L., & Garcia-Mila, M. (1992). Cross-domain development of scientific reasoning. Cognition and Instruction, 9, 285-327.
Lajoie, S.P. (1993). Cognitive tools for enhancing learning. In S. P. Lajoie & S.J. Derry (Eds.), Computers as cognitive tools (pp. 261-289). Hillsdale, NJ: Erlbaum.
Lavoie, D.R., & Good, R. (1988). The nature and use of predictions skills in a biological computer simulation. Journal of Research in Science Teaching, 25, 335-360.
Lesgold, A., Lajoie, S., Bunzo, M., & Eggan, G. (1992). SHERLOCK: A coached practice environment for an electronics troubleshooting job. In J.H. Larkin & R.W. Chabay (Eds.), Computer-assisted instruction and intelligent tutoring systems: Shared goals and complementary approaches (pp. 201-239). Hillsdale, NJ: Erlbaum.
Leutner, D. (1993). Guided discovery learning with computer-based simulation games: effects of adaptive and non-adaptive instructional support. Learning and Instruction, 3, 113-132.
Lewis, E.L., Stern, J.L., & Linn, M.C. (1993). The effect of computer simulations on introductory thermodynamics understanding. Educational Technology, 33, 45-58.
Linn, M.C., & Songer, N.B. (1991). Teaching thermodynamics to middle school students: What are appropriate cognitive demands? Journal of Research in Science Teaching, 28, 885-918.
Michael, J.A., Haque, M.M., Rovick, A.A., & Evens, M. (1989). The pathophysiology tutor: a first step towards a smart tutor. In H. Maurer (Ed.), Computer Assisted Learning. Proceedings of the 2nd International Conference ICCAL (pp. 390-400). Berlin, Germany: Springer-Verlag.
Munro, A., Fehling, M.R., & Towne, D.M. (1985). Instruction intrusiveness in dynamic simulation training. Journal of Computer-Based Instruction, 2, 50-53.
Munro, A., Johnson, M.C., Surmon, D.S., & Wogulis, J.L. (1993). Attribute-centred simulation authoring for instruction. In P. Brna, S. Ohlsson & H. Pain (Eds.), Proceedings of AI-ED 93 (pp. 82-89). Edinburgh, United Kingdom: Association for the Advancement of Computing in Education.
Munro, A., & Towne, D.M. (1992). Productivity tools for simulation-centered training development. ETR&D, 40, 65-80.
Novak, J.D., & Wandersee, J.H. (1990). Perspectives on concept mapping (special issue). Journal of Research in Science Teaching, 27, 921-1079.
Reigeluth, C.M., & Schwartz, E. (1989). An instructional theory for the design of computer-based simulations. Journal of Computer-Based Instruction, 16, 1-10.
Reimann, P. (1991). Detecting functional relations in a computerized discovery environment. Learning and Instruction, 1, 45-65.
Reimann, P., & Beller, S. (1993). Computer-based support for analogical problem solving and learning. In D.M. Towne, T. de Jong & H. Spada (Eds.), Simulation-based experiential learning (pp. 91-105). Berlin, Germany: Springer-Verlag.
Rieber, L.P., Boyce, M., & Assad, C. (1990). The effects of computer animation on adult learning and retrieval tasks. Journal of Computer-Based Instruction, 17, 46-52.
Rieber, L.P., & Parmley, M.W. (1995). To teach or not to teach? Comparing the use of computer-based simulations in deductive versus inductive approaches to learning with adults in science. Journal of Educational Computing Research, 14, 359-374.
Riley, D. (1990). Learning about systems by making models. Computers & Education, 15, 255-263.
Rivers, R.H., & Vockell, E. (1987). Computer simulations to stimulate scientific problem solving. Journal of Research in Science Teaching, 24, 403-415.
Schauble, L., Glaser, R., Raghavan, K., & Reiner, M. (1991). Causal models and experimentation strategies in scientific reasoning. The Journal of the Learning Sciences, 1, 201-239.
Schauble, L., Klopfer, L., & Raghavan, K. (1991). Students’ transitions from an engineering to a science model of experimentation. Journal of Research in Science Teaching, 28, 859-882.
Schauble, L., Raghavan, K., & Glaser, R. (1993). The discovery and reflection notation: A graphical trace for supporting self regulation in computer-based laboratories. In S. P. Lajoie & S.J. Derry (Eds.), Computers as cognitive tools (pp. 319-341). Hillsdale, NJ: Erlbaum.
Showalter, V.M. (1970). Conducting science investigations using computer simulated experiments. The Science Teacher, 37, 46-50.
Shute, V.J. (1993). A comparison of learning environments: All that glitters .... In S.P. Lajoie & S.J. Derry (Eds.), Computers as cognitive tools (pp. 47-75). Hillsdale, NJ: Erlbaum.
Shute, V.J., & Glaser, R. (1990). A large-scale evaluation of an intelligent discovery world: Smithtown. Interactive Learning Environments, 1, 51-77.
Simmons, P.E., & Lunetta, V.N. (1993). Problem-solving behaviors during a genetics computer simulation: beyond the expert/novice dichotomy. Journal of Research in Science Teaching, 30, 153-173.
Smith, R.B. (1986). The Alternate Reality Kit: An animated environment for creating interactive simulations. Proceedings of IEEE Computer Society Workshop on Visual Programming (pp. 99-106). Dallas, TX.
Suthers, D., Weiner, A., Connelly, J., & Paolucci, M. (1995). Belvedere: Engaging students in critical discussion of science and public policy issues. AI-Ed 95, the 7th World Conference on Artificial Intelligence in Education, August 16-19, 1995, Washington DC, pages 266-273.
Swaak, J., & de Jong, T. de (1996). Measuring intuitive knowledge in science: the what-if test. Studies in Educational Evaluation, 22, 341-362.
Swaak, J., van Joolingen, W.R., & de Jong, T. (1996). Support for simulation based learning; The effects of model progression and assignments on learning about oscillatory motion. Enschede, The Netherlands: University of Twente, Centre for Applied Research on Education.
Tabak, I., Smith, B.K., Sandoval, W.A., & Reiser, B.J. (1996). Combining general and domain-specific strategic support for biological inquiry. In C. Frasson, G. Gauthier & A. Lesgold (Eds.), Intelligent Tutoring Systems (pp. 288-297). Berlin, Germany: Springer-Verlag.
Tait, K. (1994). DISCOURSE: The design and production of simulation-based learning environments. In T. de Jong & L. Sarti (Eds.), Design and production of multimedia and simulation-based learning material (pp. 111-133). Dordrecht, The Netherlands: Kluwer Academic Publishers.
Teodoro, V. D. (1992). Direct manipulation of physical concepts in a computerized exploratory laboratory. In E. de Corte, M. Linn, H. Mandl & L. Verschaffel (Eds.), Computer-based learning environments and problem solving (NATO ASI series F: Computer and Systems Series) (pp. 445-465). Berlin, Germany: Springer-Verlag.
Thomas, R., & Neilson, I. (1995). Harnessing simulations in the service of education: the Interact simulation environment. Computers & Education, 25, 21-29.
Thurman, R.A. (1993). Instructional simulation from a cognitive psychology viewpoint. Educational Technology Research & Development, 41, 75-89.
Thurman, R.A., & Mattoon, J.S. (1994). Virtual reality: Towards fundamental improvements in simulation-based training. Educational Technology, 34, 56-64.
Towne, D.M. (1995). Learning and instruction in simulation environments. Englewood Cliffs, NJ: Educational Technology Publications.
Towne, D.M., & Munro, A. (1989). Artificial intelligence in training diagnostic skills. In D. Bierman, J. Breuker & J. Sandberg (Eds.), Proceedings of the 4th international conference on AI & Education (pp. 291-298) Amsterdam: IOS.
Veenman, M.V.J., & Elshout, J.J. (1995). Differential effects of instructional support on learning in simulation environments. Instructional Science, 22, 363-383.
White, B.Y. (1984). Designing computer games to help physics students understand Newton’s laws of motion. Cognition and Instruction, 1, 69-108.
White, B.Y. (1993). ThinkerTools: causal models, conceptual change, and science education. Cognition and Instruction, 10, 1-100.
White, B.Y., & Frederiksen, J.R. (1989). Causal models as intelligent learning environments for science and engineering education. Applied Artificial Intelligence, 3(2-3), 83-106.
White, B.Y., & Frederiksen, J.R. (1990). Causal model progressions as a foundation for intelligent learning environments. Artificial Intelligence, 42, 99-157.
Zietsman, A.I., & Hewson, P.W. (1986). Effect of instruction using microcomputers simulations and conceptual change strategies on science learning. Journal of Research in Science Teaching, 23, 27-39.