Learning in vivo

Three types of intelligent virtual tutoring agents to scaffold virtual laboratory experiential learning

Yu, Brown and Billett (2005: 7) concluded more than a decade ago that “Computer-based virtual learning environment is a relatively new technology in education and training. Studies to date have shown VLEs potentially benefits both educators and students, particularly in higher education and in the delivery of complex scientific concepts and principles. It eliminates time and geographical constraints. At the moment, such applications would not replace real world laboratory exercises. Students, however, could augment their real life laboratory experience. They can explore the virtual laboratory experiments in a relaxed and flexible environment at their own speed and need. They can use it at revision time, when access to the real laboratory may be prohibited in an unsupervised way, due to the expensive or dangerous nature of the laboratory exercises. Educators can add to their battery of course materials to promote more effective understanding of curricula, and more easily accessible course resource.”

About ‘virtual tutoring agents’ Yu, Brown and Billett (2005: 7) elaborate on the “development of a set of intelligent tutoring agents” to ‘scaffold’ virtual laboratory experiential learning for students. They explain an ‘intelligent agent’ as “a section of computer programme that simulates a human relationship by doing something that another person could do” and name three types:

1. Deductive tutor provides assistance to students in the course of their deduction with scientific problem solving, which is required to accomplish their tutorial. For example, if students add a certain liquid to a tube, but instead they select a wrong tube, the deductive tutor will display a warning message and play an audio track to the students.

2. Rule-based tutor provides assistance to students by:

• Encoding a set of rules of an experiment.
• Monitoring a student's action and looking for one of these rules to be "broken".
• "Visiting" students to present expert advice.

For example, there are dilution rules: (1) the amylase in tube 1 should be diluted from the Amylase bottle (the first bottle from left in Figure 2) and (2) the amylase in tubes 2 to 6 should be diluted from the tube to the left of the current one. If the students break one of the dilution rules, the rule-based tutor will remind them. It also tells students where to find a relevant case-based virtual tutor for demonstration.

3. Case-based tutor provides assistance to students by presenting an example of closest relevant experience. For example, a video for each experimental procedure is embedded in the virtual step in the Help system.

The deductive and rule-based tutoring agents are embedded in the 3D virtual environment to monitor

student's actions and record the status of the objects in the learning process to ensure the learning tasks tackled. A virtual agent provides the students with advice once the students make a mistake, overlook an object or break a rule, which result in the experimental progress toward successful completion impasse.

The advice is designed in both text and voice message formats for greater accessibility.

The case-based virtual agents are embedded in the Help system in order to reduce the working load of the 3D virtual laboratory. The case-based virtual agents play a video stream to demonstrate a procedure of the experiment once a request is explicitly made.

Yu, J.Q.; Brown, D.J. & Billett, E. (2005). Development of a Virtual Laboratory Experiment for Biology. European Journal of Open, Distance and E-Learning, 8(2). Electronically accessible from http://www.eurodl.org/materials/contrib/2005/Jian_Quing_Yu.pdf

Role of mental models in simulations

The Latin of simulation (simulō) means to imitate/pretend, which may have a negative connotation in that simulations might be regarded as false/unreal representations. Plato’s conception of mimesis (μίμησις, Greek term for simulation) had been an imperfect copy or fictitious replica of reality. In contrast, Aristotle notion of mimesis had been a “means to know nature through representations which can be valid and acceptable” (Landriscina, 2012). This, Landriscina points out, is an important shift in emphasis “from imitation to representation” where mental models are key to simulations.

In model-based instruction the underlying model assume a significant function, namely the identification of similarities and differences between the model and the phenomenon it represents. The process results in transforming ideas; confuse and approximate; resulting in more precise and rigorous comprehension. This, Landriscina says, also “happens when a student compares his own mental model of a system with that of a simulation”. He illustrates this with Plato’s example from Politics (Statesman) of cognitive role of the models, or paradeigmata (παραδείγματι):

one of the characters of the dialogue refers to the model of the weaver, where the technique of the politician is compared to that of the weaver who weaves fibers of different nature to create a single fabric

Senge (1990, cited by Landriscina) describes mental models as “deeply ingrained assumptions, generalizations, or even pictures and images that influence how we understand the world and how we take action”. Seel (2003, cited by Landriscina) examined the relation between mental model and education and formulated a teaching and learning theory: Model-Centered Learning and Instruction. Seel differentiated mental-; conceptual-; and design and instructional models:

• Mental model, the internal and subjective model of a system
• Conceptual model, which is objective and shared by a scientific community
• Design and instructional model, used for building the user interface and the learning tasks

To serve as simulation model, the conceptual model must be stated in mathematical terms (rules and equations) in order to be developed as computer programmes. In order to create simulated learning; a mental model of the phenomenon must be formed, studied and use as a basis for prediction, inference, and explanation. Landriscina (2012) cautions by means of Senge’s (1991) words that “simulations with thousands of variables and complex arrays of details can actually distract us [students] from seeing patterns and major interrelationships” and result in misconceptions by students. However, with “black-box simulation models”, for example SimCity® the student interacts with the various scenarios and applications without observing the rules of the game the creators programmed.

Landriscina, F. 2012. Simulation and Learning: The Role of Mental Models. In Seel, N.M. (Ed.) Encyclopedia of the Sciences of Learning. Springer Science + Business Media.

Synchronous online laboratory learning

Active laboratory experiment-interaction and problem-solving are considered essential parts of science and engineering education, because they enable direct acquisition of applicable knowledge for use in practical situations. Synchronous experimentation offers immediate feedback, rather than delayed. “Online (Remote) laboratories” report Auer, Edwards and Zutin (2011: 222) “available via internet” are therefore gaining importance. They define Online Engineering “as an interdisciplinary field utilising the areas of engineering, computing and telematics, where specific engineering activities like programming, design, control, observation, measuring, sensing, and maintenance are provided to both remote and local users in a live interactive setting over a distributed, physically dispersed network (for example: internet or intranet)”. They add that “high bandwidth internet connections world-wide and other derivative capabilities in the areas of real-time communication, control, teleconferencing, video streaming and others have made multi-site collaborative work, utilising state-of-the art equipment in remote laboratories across the globe a current reality”.

According to Auer, Edwards and Zutin (2011: 226) there are already a number of remote laboratories in operation and that the range of disciplines being catered for is growing as universities collaborate internationally. The portability and advancement of mobile technology further serve as powerful motivator for growth of online laboratory experiences.

Auer, M.E.; Edwards, A. & Zutin, D.G. 2011. Online laboratories in interactive learning environments, pp. 219-248. In N. Pachler; C. Pimmer & J. Seipold (eds). 2011. Work-Based Mobile Learning— Concepts and Cases. Oxford: Peter Lang.

Effective user interface for simulated learning

For at least three decades the design principles of user interface has a presence in computer science literature. During the last decade computer systems became increasingly more interactive and code for interface increased accordingly. Interactive user interface developments and graphic user interfaces (GUI) in particular do provide simpler ways for humans to communicate with PCs. Although GUI requires complex programming, several software tools emerged that assist with the implementation of GUI, such as NeXT Interface Builder; WindowsMAKER; TelePICTIVE; and TelePICTIRE (Sidhu, 2010: 36).

Sidhu (2010: 36-38) suggests the following design guidelines for better user interface:

• Simple, clear and standardised screen layout with liberal use of blank space. Functional aspects such as title, help text and buttons should be consistent throughout; with distinctive design that emphasise critical information. Textual information should be easily legible, by using appropriate fonts, typestyles and typesets. The resolution of graphics should eb good quality.
• Make use of easily navigatable multi-windows and menus to enable users to bring relevant windows to the front.
• Make use of menus that indicate the selected option and visited items. Provide keyboard shortcuts (accelerators) and indicators of drastic action if executed, e.g. exiting or quiting without having saved. Use hierarchical (sub) menus with clear indicators.
• Provide selection lists (pop-ups) when hovering with mouse.
• Make use of realistics icons to give visual clues to users.
• Make use of feedback indicators, e.g. progress or time/percentage remaining; and/or graphic indicators, such as bars.
• Provide online help that is context sensitive.
• Error messages should be constructive and offer remedy suggestions.
• Use colour to enhance usefulness, e.g. similar functional background colours; contrast to focus users; and bright colours for caution.
• Mouse input (and other input devices) should be consistent with contemporary software.

Sidhu, M.S. 2010. Technology-assisted problem solving for engineering education: interactive multimedia applications. Hershey: Engineering Science Reference.

User-computer interface for simulations

Interface design concerns the practical information a multimedia-author needs in order to produce usable and well-designed interfaces, states Sidhu (2010: 30). In addition to the outcomes a learning package aims to achieve through, for example, tasks; the developing team need to consider several interacting concerns. Although several user interface standards exists—for example ISO, CEC/CENELEC and BSI—unfortunately some developers do not make use of a common standard. Human Computer Interaction (HCI) is a discipline “concerned with the design, evaluation and implementation of interactive computing systems for human” usage and “user interface is an interface developed to enable HCI”—or differently stated “a communication channel between the user and the functional elements of the computer” (Sidhu, 2010: 31) illustrated by Plass (1998, in Sidhu) as follow:

The illustration comprises two functions. At the bottom right, the user input is allocated (centre) to the application (left). At the top left, the functionality flows (centre) to the user (right). In the case of text-based HCIs only written verbal communication is used, whereas with visual HCIs manipulation of visual objects, verbal and auditory interaction is involved.

Sidhu, M.S. 2010. Technology-assisted problem solving for engineering education: interactive multimedia applications. Hershey: Engineering Science Reference.

Sensory information type and processing preference

In order to optimise learning through simulated experiences, the Felder-Silverman Learning Style Model (FSLSM) is recommended. People exhibit different ways they prefer to learn and by incorporating different learning styles in teaching plans learning is made easier. Graf, Viola, Leo and Kinshuk (2007) observe that FSLSM is often used in the design of technology-enhanced learning. The learning styles models of Kolb (1984) and Honey and Mumford (1982)—in contrast to Felder and Silverman (1988)—classify learners into a few groups. FSLSM describes learning styles in more detail, distinguishing four dimensions depicted below (Sidhu, 2010: 23):

Graf, Viola, Leo and Kinshuk (2007) further points out that FSLSM is based on tendencies. They point out that although learners with a strong preference for certain learning style may experience difficulty if the teaching style does not match their preferred learning style; indeed sometimes behave differently. They further convey that educational researchers in recent years investigated how incorporating learning style models impact technology-enhanced learning and found that FSLSM is highly appropriate for hypermedia courseware. It has been found that when students do an online course that matched their preferred learning style they achieved significantly better results compared to students who took a course that did not match.

The first dimension—sensory/intuitive—Students that prefer learning facts and concrete learning material have a sensing learning style. Such students prefer standard approaches in solving problems; to relate the learned material to the real world; and tend to be more patient with details. They are generally realistic, sensible and practical. Intuitive learners, by contrast, prefer learning abstract learning material, such as the underlying theories and their meanings. Such students tend to be innovative and creative as well as more able to discover possibilities and relationships.

The second dimension—visual/verbal—Students with a visual preference learn and retain best what they have seen, for example pictures, diagrams, and flow-charts. Verbal learners derive more from textual representations, both written and spoken.

The third dimension—active/reflective—addresses different ways of processing information. Some students learn best by actively engaging; trying out and by applying the learning material. They further tend to prefer communicating and by working in groups where they can converse. Other students, in contrast, prefer to reflect and think about learning material. They prefer working alone or possibly in a small group or with a good buddy.

The fourth dimension—sequential/global—is about characterising students according to their understanding. Those students that learn in small incremental steps in linear progression are sequential learners. They prefer following logical step-by-step paths towards solutions and are interested in details. Global learners, however, use holistic thinking processes learning in large leaps. Such students tend to absorb learning material almost randomly, not necessary seeing connections initially but suddenly grasp the whole picture. They are capable to solve complex problems; to find connections between dissimilar areas; and to put things together in novel ways. However, they can seldom explain how they did it. Global learners prefer overviews and broad knowledge.

Sidhu (2010: 23) poses four questions:

1. What type of sensory information is most perceived?
2. What type of information does the student preferentially perceive?
3. How does the student prefer to process information?
4. How does the student characteristically progress towards understanding?

Graf, S.; Viola, S.R.; Leo, T & Kinshuk. 2007. In-Depth Analysis of the Felder-Silverman Learning Style Dimensions. Journal of Research on Technology in Education, Fall 2007, 40(1). Accessed on Internet 31 May 2012 from: http://wit.at/people/graf/publications/graf_viola_kinshuk_leo_JRTEjournal.pdf

Sidhu, M.S. 2010. Technology-assisted problem solving for engineering education: interactive multimedia applications. Hershey: Engineering Science Reference.

Attributes and learning-styles of students affect their learning

Sidhu (2010: 16) identifies the following attributes that may hinder or improve learning:

• Basic background knowledge of the subject matter / discipline / field of study
• Prior academic performance impacts on a student’s performance
• Degree of exposure to contemporary educational technologies — the use of computer-based learning is enabled if students already acquired PC usage skills
• The impact of the learning styles of students have been extensively researched

Depicted below are Kolb’s as well as Honey and Mumfords learning style models, some of the most well known models:

http://www.businessballs.com/kolblearningstyles.htm

http://www.nonformality.org/2006/04/excuse-misuse-or-abuse/

Kolb’s experiential learning styles are in black, Honey and Mumford’s in red.

http://pmj.bmj.com/content/83/982/518.full

http://rphrm.curtin.edu.au/2000/issue2/outcomes.html

Sidhu, M.S. 2010. Technology-assisted problem solving for engineering education: interactive multimedia applications. Hershey: Engineering Science Reference.

Replacing natural experiments by computer simulations

Simulations as viable alternative to doing experiments on natural systems, offer anchor points to integrate scientific discovery learning or inquiry learning with instructional support in educational settings (for example distance) where for a number of reasons actual experiments may not be feasible, because (Van Joolingen & De Jong, 2003: 1):

• the natural system may not be accessible due to expense of the measuring equipment required or non-availability
• of safety hazards, such as radioactive samples, toxical chemical substances or populations of certain bacteria
• performing the experiments may be unethical, for example in the case of experiments with livestock
• time scale of experiments on natural phenomena may not fit with the educational timeframe— some experiments take days, months or even years to complete or may happen so fast that it is hard to visualize the processes the learning is about
• experiments may consume rare resources

“Simulations are computer programs containing an executable model of a natural or other system, capable of computing the behavior of the modeled system by means of dedicated algorithms and presenting the results of these computations to the user” (Van Joolingen & De Jong, 2003:2).

However, caution Van Joolingen and De Jong (2003: 3-4) the creation of computer simulations embedded in instruction is out of reach of most lecturers; and off-the-shelf simulations often do not match the specific teaching requirements. Creating simulations require specific skills for the various steps in the process:

1. Creating the model that would drive the simulation—Capability to do this requires expertise of either a programming language or a special simulation language; expertise in the domain simulated and knowledge of the target language; and in the case of a general purpose programming language one must also create the algorithm to simulate the model.
2. Creating the learner interface to the simulation—Including visual representation, using graphics representing domain elements, graphs, numbers, etc. Capability to create interfaces requires expertise of the available tools as well as to create the domain specific graphics within the interface.
3. Creating the instructional design of the environment from the perspective of the learner— Instructional design expertise, such as which activities to be performed; the goals of such activities; what information/explanations would be needed at specific moments (discipline knowledge); knowing/awareness of frequently occurring misconceptions; and which instructional interventions should form part of the simulation?
4. Integrate the parts of the simulation environment to a complete system in order to create a coherent integrated simulation— This means that all parts need to co-operate and that the interactions between simulation model, learner interface and all instructional interventions should be smooth and stable. This step requires programming skills in order to create an efficient and transparent environment.

The remainder of the referenced article discusses SimQuest—a simulation authoring system developed as free and open software by University of Twente for the designing and creating of simulation-based learning environments for discovery learning—downloadable for free.

Van Joolingen, W.R., & De Jong, T. (2003). SimQuest, authoring educational simulations. In: T. Murray, S. Blessing, S. Ainsworth. Authoring Tools for Advanced Technology Learning Environments: Toward cost-effective adaptive, interactive, and intelligent educational software. pp. 1-31. Dordrecht: Kluwer. Accessed from Internet 18 May 2012 from: http://halshs.archives-ouvertes.fr/docs/00/19/06/76/PDF/SimQuest-VanJoolingenDeJong.pdf