40
Int. J. Learning Technology, Vol. 1, No. 1, 2004
Cognitive visualisations and the design of learning technologies Michael J. Jacobson Korea University, 1, 5Ga, Anam-Dong, SungBuk-ku, Seoul, 136-701, Korea E-mail:
[email protected] Abstract: This paper discusses cognitive visualisations as a design approach in which the affordances of currently available learning technologies are used to reify or make explicit mental representations about conceptually challenging knowledge. Conceptual visualisations are intended to help learners develop understandings of a domain through dynamic visual representations that are based on cognitive research into the nature of cognitive frameworks or mental models, in particular, models with the main characteristic of naïve or intuitive ideas and of expert understandings. A rationale for cognitive visualisations is presented, followed by a discussion of two examples of cognitive visualisations in contrasting domains that were developed using different implementing technologies. A set of issues for future research in this area is also considered. Keywords: design of learning technologies; cognition and learning; knowledge representation; mental models; visualisations; agent-based modelling; conceptual change. Reference to this paper should be made as follows: Jacobson, M.J. (2004) ‘Cognitive visualisations and the design of learning technologies’, Int. J. Learning Technology, Vol. 1, No. 1, pp.40–62. Biographical notes: Michael J. Jacobson, PhD, is an Associate Professor at Korea University in Seoul, Korea and an Affiliate of the New England Complex Systems Institute. Previously, Dr. Jacobson held faculty and research positions at the University of Illinois at Urbana-Champaign, Vanderbilt University, and The University of Georgia, and he has been involved with organisational and international consulting activities. His research interests have focused on the design of learning technologies to foster deep conceptual understanding, conceptual change, and knowledge transfer in challenging conceptual domains. Most recently, his work has explored cognitive and learning issues related to understanding new scientific perspectives emerging from the study of complex and dynamical systems.
1
Cognitive visualisations and the design of learning technologies
There are many challenges facing learners from school-age students to professional adults who must come to robust and useable understandings of challenging knowledge and skills for the 21st century. The appropriate design and use of learning technologies for use in rich and stimulating learning environments holds considerable promise to help learners meet these challenges [1]. Whereas there is a wide range of design frameworks that may be used for the principled development of learning technologies, this paper Copyright © 2004 Inderscience Enterprises Ltd.
Cognitive visualisations and the design of learning technologies
41
focuses on a specific approach to enhance student and adult learning by reifying cognitive frameworks or mental models about domain knowledge through the use of cognitive visualisations [2]. (Whereas in earlier papers, the term ‘conceptual visualisations’ was used [2–4], the phrase ‘cognitive visualisations’ is preferred here to stress the function of these representations for the visual reification of novice and expert mental models as inferred from cognitive research and knowledge representation techniques.) A rationale for cognitive visualisations (CVs) is presented, followed by a discussion of two CV examples in contrasting domains that were developed using different implementing technologies. A set of issues for future research in this area is also considered.
2
Cognitive visualisations make thinking visible
Cognitive visualisations may be regarded as one of a class of techniques intended to make aspects of thinking visible [5]. Collins and his colleagues discuss “making thinking visible” in the context of their work on cognitive apprenticeships. Cognitive apprenticeships involve modelling the use of knowledge by a competent individual, providing coaching and scaffolding to learners as they are involved in activities involving the use of the knowledge being learned, and fading of the support provided to learners as they become more accomplished. ‘Visible thinking’ may also involve making explicit or reifying the structure of knowledge or the mental models learners and experts use when thinking about conceptual domains. One technique for doing this is the use of concept maps to represent knowledge by an arrangement of labelled boxes depicting concepts and the relationships between those concepts by lines and arrows. Concept maps may represent the conceptual structures students or experts have about a domain, and also have been used to assess student understanding [6]. A recent extension of concept mapping techniques is the use of systems modelling tools to represent conceptual structures [7,8]. Another technique for reifying knowledge structures is the use of dynamic and visually rich computer visualisations. Visualisation technologies were initially pioneered for the representation of complex scientific data and models. Earlier applications of these technologies in education focused on providing access to visualisation tools in support of student scientific inquiry activities involving data analysis and modelling [9,10]. In contrast, my research group has investigated the use of visualisations to provide cognitive representations of knowledge structures. The design of cognitive visualisations (CVs) takes advantage of the representational affordances of interactive multimedia, animation, and computer modelling technologies. In contrast to concept maps that are static twodimensional representations, CVs are dynamic qualitative representations of the mental models held by experts and novice learners in specific domains and the application of these models in the context of representative problems and situations [2,3]. In addition, CVs may function as metaconceptual scaffolds that help students develop an awareness of their mental representations and inferential processes [3]. It has been suggested that helping students develop a metaconceptual awareness of their mental representations in a domain may help foster conceptual change and representational growth – important but often difficult to achieve learning goals in challenging conceptual
42
M.J. Jacobson
domains [11,12]. Consequently, CVs may prove to be a valuable learning tool in helping to foster conceptual change.
3
Cognitive visualisations: two examples
To date, my research group has conducted two main projects involving the development of cognitive visualisations. The first project was a proof-of-concept development of CVs dealing with naïve and expert mental models of evolutionary biology in the context of a larger overall study. The second project involved the articulation of CVs dealing with centralised versus decentralised decision-making and control processes that was used by an executive management consultancy. These two projects are discussed in turn.
3.1 The evolution knowledge mediator project My research group initially developed CVs as part of a research project involving a case and problem-based hypermedia system intended to help students learn the neo-Darwinian theory of evolution [3]. This research led to the articulation of the Knowledge Mediator Framework that proposed a set of design features for learning technologies intended to help students construct deep and useable conceptual understandings of difficult knowledge. The Knowledge Mediator Framework consists of four design elements and six learning activities. The design elements are: 1
representational affordances of technology
2
represent knowledge-in-context
3
reify the deep structure of knowledge
4
intra/intercase hyperlinks for conceptual and representational interconnectedness.
The six learning activities are: 1
cognitive interactivity
2
scaffolded problem solving
3
cognitive preparation (e.g., ‘thought experiments’, show gaps in student problem solutions, ‘seed’ new concepts and models)
4
preliminary learning tasks with deep structure knowledge resources
5
guided conceptual criss-crossing
6
learner centred and project-based learning.
See Jacobson and Archodidou [3] for a fuller discussion of the Knowledge Mediator Framework and research related to selected design features and learning activities. CVs were used as one technique to instantiate the third design element of the Knowledge Mediator Framework: Reify the Deep Structure of Knowledge. Four evolution CVs of an expert mental model and one of a naïve mental model were created using Macromedia Director. In each of the CVs developed for different case examples of evolutionary biology, the learner selects an environmental condition (which is, of course, the selection engine in Darwinian theory), and then steps through a set of screens
Cognitive visualisations and the design of learning technologies
43
showing a series of events related to evolutionary changes over three generations. In addition, each screen is annotated with descriptions based on either verbal protocols of naïve explanations of evolution or expert scientific textual explanations.
3.1.1 Cognitive visualisations of expert and naïve mental models of evolution Two evolution conceptual visualisations were developed that represented highly contrasting perspectives based on expert and naïve mental models of how evolution works. The set of screens in Figures 1a to 1d shows a CV of an expert mental model of evolution that dealt with the case of the Peppered Moth in the polluted forests of 19th century industrialised England. The model of evolution biologists use is based on neoDarwinism that holds the importance of trait variability in a population (Figure 1a) from which some members of the population in a particular environmental set of conditions live long enough to reproduce (i.e., the parent organisms are ‘naturally selected’, Figure 1b). Pollution in the 19th century English forests meant that dark-coloured moths were better camouflaged from predator birds than lighter coloured moths and so were more likely to survive and pass on the genetic trait of darker colouration, as shown in Figure 1c (although mutations may occur such as the differently shaded moth in the lower right corner of Figure 1c). Figure 1d shows that the second-generation parents have died, thus leaving their offspring to be the parents for the next generation. Note in Figure 1d that the trait variability in colouration persists even after two cycles of natural selection favoured darker coloured moths. Indeed, despite natural selection for darker coloured moths, there remain lighter coloured moths, albeit in a greatly reduced proportion of the overall population than in the first generation, as shown in the thumbnail view on the left side of the figure. Figure 1a Second generation of moths with colour trait variations in the industry pollution environment
44
M.J. Jacobson
Figure 1b Darker moths more likely to survive on dark, polluted trees
Figure 1c Darker moths reproduce, thus passing genes for darker colouration to offspring
Cognitive visualisations and the design of learning technologies
45
Figure 1d Second generation parents die, leaving greater proportion of moths in the third generation with darker colours than beginning of first or second generations
The CV shown in Figures 2a to 2d is of a common naïve mental model of evolution identified by Bishop and Anderson [13] that is dramatically different from the neo-Darwinian mental model depicted in Figures 1a to 1d. The scenario for the second CV is a hypothetical species – the ‘Bugs’ – in which certain Bugs have a set of traits related to their green colour that allows them to survive better in a wet climate, while brown Bugs survive more robustly in a dry climate. The naïve mental model of evolution identified by Bishop and Anderson views evolution as occurring in a population that consists of homogeneous traits rather than as a population of organisms with a variety of traits (e.g., colour in this example). The naïve CV represented population homogeneity by depicting all of the Bugs as starting with the neutral colour trait of yellow that could change to either green or brown. (Note: colours are shown in grey scale in the figures, with a very light shade of grey representing yellow Bugs and a moderately dark shade of grey representing green Bugs; brown Bugs are not shown in the figures.) The learner selected a wet environment in the run of the visualisation shown in Figures 2a to 2d, so a population of Bugs with green colouration would survive better in this environment. Another feature of this particular naïve mental model of evolution is that all the organisms in a population ‘try’ to change in order to respond to the ‘need’ of the environment. To represent these aspects of the naïve mental model, there are slightly greener (i.e., a darker shade of grey) Bugs in Figure 2b to represent the Bugs that survived long enough to reproduce. Figures 2c and 2d continue to show the offspring of the Bugs are uniformly a darker shade of grey (i.e., green) colour.
46
M.J. Jacobson
This CV of a naïve mental model represents the ‘evolution’ of the Bugs over three generations in the thumbnails on the left side of the screen, where the change in colouration from the very light shade of grey (i.e., yellow) of the first generation to a darker shade of grey (i.e., green) wet-adapted Bugs in the final generation is made visually salient to the user. Figure 2a Second generation of ‘Bugs’ with homogeneous trait for light colouration; note change from first generation
Figure 2b Bugs ‘try’ to get greener since environment ‘needs’ greener Bugs, with survivors
Cognitive visualisations and the design of learning technologies Figure 2c Surviving Bugs reproduce to pass homogeneous colour change to next generation
Figure 2d Second generation parents die, leaving the third generation
47
48
M.J. Jacobson
It is important to note that this naïve mental model of evolution does have a component of the scientifically accepted view of evolution, the heritability of traits, as is shown in the set of screens in Figures 2a to 2d. However, the naïve mental model fails to incorporate the notion of the variation of traits in different organisms in a population, variations that are essential in order for natural selection to occur. In addition, the naïve model has the organisms teleologically ‘try’ to change in order to ‘adapt’ – however, an organism cannot purposefully change its DNA associated with various traits (e.g., mutations are not caused by an organism). Overall, the different CVs shown in the Figure 1 and Figure 2 sets were intended to make salient in visually compelling ways two very different mental representations of evolutionary processes.
3.1.2 Preliminary cognitive visualisation research The preliminary usability and learning assessment of the evolution CVs occurred in the context of a larger study of conceptual change and knowledge transfer related to the use of a case- and problem-based hypermedia system intended to help students learn neoDarwinian evolutionary biology. The overall study involved eight tenth-grade students, four girls and four boys, who were taking high school biology. At the time of the study, these students had not yet studied evolution in class. The students came to our laboratory to use the evolution hypermedia system and to complete the cognitive assessment tasks for four sessions of approximately two hours each. The study primarily focused on design features related to the interconnectedness of cases and concepts in the evolution hypermedia system and to special cognitive scaffolding during online problem solving with the system, consequently only selected data was collected related to the CVs. The main use of the CVs occurred during the day one pre-test activities. The students ran a neo-Darwinian CV of the Bugs (a CV that was isomorphic to the Peppered Moth CV shown in Figures 1a to 1d) and the naïve CV of the Bugs shown in Figures 2a to 2d. Next, the students were asked which of the two CVs was closest to the way they thought about evolution. Five of eight students (63%) selected the naïve CV, which was consistent with other conceptual assessment data that indicated six of the eight students (75%) used a naïve mental model of evolution to solve various evolution problems, while two used more expert-like models with the main components of the neo-Darwinian model. Later in the study, the students ran two additional CVs that were linked to two of the evolution cases they studied: the Peppered Moth CV (see Figures 2a to 2d above) and the antibiotic resistant bacteria CV. Summarising, there were two main pre-test findings concerning CVs: 1
83% of the six students who used naïve models of evolution when problem solving also selected the naïve Bugs CV
2
100% of the students using Darwinian models of evolution when problem solving selected the expert Bugs CV.
3.2 Cognitive visualisations of natural systems (and analogies for social and organisational systems) This section discusses a second cognitive visualisation (CV) example that was developed using a different technological approach than the evolution CVs. After an overview of the computational modelling tool used to create the CV, the context for the use of the CVs by
Cognitive visualisations and the design of learning technologies
49
an executive management consulting firm is described, followed by a discussion of the features and content of the two different CVs that were developed.
3.2.1 Developing cognitive visualisations with agent-based modelling tools The CVs discussed in the previous section were created using interactive multimedia software (Macromedia Director). While this technology was well suited for creating animations with high visual representation qualities, it had a major limitation in that all branching options in the visualisations had to be pre-determined. Thus, creating more elaborate CVs using this approach would involve considerable increases in instructional design and programming time and, consequently, significant increases in overall development costs. In this second example of developing and using CVs, an alternative technique for creating the CVs was explored, one that employed an agent-based modelling (ABM) environment initially developed at the Massachusetts Institute of Technology known as STARLOGO [14]. Briefly, agent-based modelling specifies rules that govern the interactions of agents or objects in a computational environment. Specified as variables in the model, these rules, which are often quite simple, are based on one’s formal or intuitive understandings about a system and its constituent elements. A related feature of ABMs, which contrasts with more traditional computer modelling approaches, is the explicit use of a relatively restricted set of variables and their attendant rules. This distillation of factors in an ABM forces model developers to make explicit their thinking about a system, which often helps in the identification of critical factors or ‘drivers’ of the behaviour of the system being modelled, not to mention exposing assumptions and/or biases [15]. (The author first heard the characterisation of the limited number factors in agent-based modelling as a ‘distillation’ from conversations with Dr. Alfred Brandstein, who is the Chief Analyst/Director of Joint Modeling Simulation and Analysis for the US Marine Corps.) With ABMs, the complexity of systems emerges based on the agent-agent and agent-environment interactions, interactions that were specified with the underlying, often simple, rules.
3.2.2 Advantages of agent-based models for developing cognitive visualisations There are several advantages of developing CVs with agent-based modelling tools rather than using multimedia or animation tools such as Director or Flash. First, agent-based CVs may have considerably more options for interactivity and branching than animation and multimedia CVs because it is quite easy to set up a range of values for important parameters in an agent-based model that the user may change individually or collectively. This allows the user to explore a wider range of questions about the behaviour of the system under different conditions, and to better understand the appropriateness or fit of the model for specifying the system of interest. Second, there are now available powerful agent-based modelling tools specifically oriented for education, such as STARLOGO [16], NETLOGO [17], and AGENTSHEETS [18]. Third, for developers of CVs, the educational agent-based tools are relatively easy to use for developing models and there are extensive online free libraries of scientifically-based models that may be used or modified. Finally, and most important, cognitive research has shown that often many naïve mental models have idea components that may be similar to expert views, but often with the naïve component ideas being arranged in ways that are different than expert
50
M.J. Jacobson
knowledge structures [12]. Consequently, it should be possible to create agent-based models of naïve mental models by revising the scientific models, thus allowing the fit of these different models to the systems of interest to be runnable, salient, and inspectable to learners. Indeed, this is the approach discussed in the next section.
3.3 Context and challenges: knowledge representations and business decision making The context for this second cognitive visualisation example is an applied setting involving the Allison Group (previously the Allison~LoBue Group), which is an executive consulting company based in New York. One of the services provided by the Allison Group was helping business executives and managers adopt new approaches for decision making and organisational structure. According to new perspectives in the business and organisational management literatures [19–21], a centrally important characteristic of the vast majority of successful modern organisations is the use of nonauthoritarian leadership styles and the distribution of decision making throughout the organisation. Unfortunately, many business executives and managers employ cognitive frameworks or mental models that regard decision making as proceeding ‘top-down’ in an authoritarian, hierarchical manner through the organisation [20,21]. Within this context, an important component of the Allison Group’s consulting practice involved helping business leaders and managers enhance the repertoire of mental models they use when dealing with the performance challenges of their organisations, in particular, to inculcate a model of decentralised leadership and distributed decision making. One of the challenges faced by Allison Group consultants, however, was to vividly, yet concisely, convey to busy, professional executives and managers two main perspectives about organisational control and decision making: •
the limitations of an authoritarian ‘command and control’ approach, such as reliance on the central control of a decision maker, ineffective use of individual resources in organisation, lack of flexibility, and difficulties adapting to change
•
the viability and advantages of decentralised control and distributed decision making, such as more rapid decision making, maximal use of individual resources in organisation, flexibility and adaptability to change, and self-organisation.
To this end, the author worked with senior Allison Group consultants to develop a cognitively-based approach to make salient key aspects of centralised and decentralised control and decision-making perspectives using cognitive visualisations (CVs) as part of consultant presentations. The decision was made to initially develop the CVs using a non-business domain of social insect colonies, specifically ants foraging for food, which illustrated selected, but important control and decision-making concepts that are also relevant to human social and organisational systems. There is another cognitive rationale for using a non-business domain for these CVs. Research has demonstrated the value of using analogies to learn new knowledge, particularly when individuals focus not just on the surface features of the analogies, but rather focus on structural similarities [22,23]. Allison Group consultants thus used the CVs as the source analogy, and then in their presentations would discuss the structural similarities to business decision making and organisations.
Cognitive visualisations and the design of learning technologies
51
3.3.1 Two contrasting mental models of ants foraging for food Figures 3a to 3f show screen shots from a CV of a ‘command and control’ mental model that was developed using STARLOGO by the author for the Allison Group consultants. The features of this CV were based in part on cognitive research into novice and expert problem solving about complex systems, research that included one problem that dealt with how ants forage for food [24]. The start of the command and control CV is shown in Figure 3a, where the dot in the middle circle represents the Queen ant who ‘controls’ or makes the decisions for the colony. Surrounding the colony are three circles of food sources in the upper left, centre right, and lower left portions of the screen. The second figure in the series, 3b, shows the Queen ant randomly searching for food. When running the computer model, there are times when the Queen ant moves very close to one of the food sources, however, the Queen (and the other ants) cannot detect food unless actual contact is made with a region of food. Finally, after approximately 20 seconds of searching (one sixth of the overall run time of the model), the Queen ant finds the food source in the lower left region, and then leaves a chemical pheromone trail (shown in lighter shadings of colour between the food source and the ant colony) while returning to the colony to ‘tell’ other ants the location of the food (i.e., via the pheromone trail that is the main mechanism for communication between ants). The other ants emerge from the colony and follow the pheromone trail left by the Queen, as shown in Figure 3d. After the new ants find the food source, they too pick up food and return to the colony while leaving their own pheromone trail, which causes the overall trail to increase in intensity from Figures 3d to 3e. However, the food source has been nearly consumed by the ants by Figure 3e, and by the final screen shot of this model run, Figure 3f, the food source has been exhausted and ants have stopped moving. Since the Queen did not go out again and find other food sources, the ants just followed their previous ‘instructions’ to forage in the same region of the environment that was now depleted of food, with the result that they die despite the proximity of two other food sources. Figure 3a Initial setting of the command and control cognitive visualisation; Queen is a dot in middle of central circle representing the ant colony, outer circles are food sources
52
M.J. Jacobson
Figure 3b Queen ant searches for food; this is one screen of the random search pattern in the actual model
Figure 3c Queen ant has found food, left a pheromone trail (in darker colours) and returned to the ant colony to ‘tell’ the other ants where to find food
Cognitive visualisations and the design of learning technologies
53
Figure 3d Additional ants from the colony follow the increasingly intense pheromone trail to the food source
Figure 3e The food source is nearly consumed by the ants; a few ants randomly stray outside the pheromone trail, but not a sufficient distance to find other food
54
M.J. Jacobson
Figure 3f Found food source depleted; the Queen does not find other food, and the ants die despite being close to other food sources
A CV of a scientifically grounded model of ants foraging is shown in Figures 4a to 4e [24]. The initial screen in Figure 4a is essentially the same as the command and control CV, with the central circle representing the ant colony and the three outer circles depicting sources of food. Figure 4b shows multiple ants streaming out of the colony where they begin to each move randomly around the environment. The next screen, Figure 4c, shows the ants continuing to stream out of the colony, with a group of ants finding the food sources located in the lower left and right sides of the screen. These ants then follow similar rules as did the Queen ant in the previous CV; once they find food, they pick it up and deposit pheromones on a trail on their way back to the colony (shown as a white trail between the food sources and the colony). The continuing random search patterns of the ants in the upper left quadrant finally results in a few ants finding the remaining food source (Figure 4d). In an interesting portion of this model run, Figure 4e shows that while the ants have completely consumed the right-side food source, the lack of any pheromone trails in the upper left quadrant indicates that collectively the ants have ‘forgotten’ about that food source. However, by Figure 4f, the ants randomly moving in the upper left quadrant ‘rediscover’ that food source. The model runs of the decentralised foraging CV typically have much longer durations than the command and control CV because the ants find all three of the food sources rather than just one.
Cognitive visualisations and the design of learning technologies
55
Figure 4a Initial setting of the scientific ants foraging CV; central circle represents the ant colony, outer circles the food sources
Figure 4b Many ants begin to stream out of the ant-hill, moving randomly in the environment
56
M.J. Jacobson
Figure 4c While additional ants leave the ant-hill, other ants have found the food sources in the right and lower left regions; returning home they leave a pheromone trail
Figure 4d Ants in random food searches find upper left quadrant food; other ants continue to move in initial pheromone trails and to find earlier food sources
Cognitive visualisations and the design of learning technologies
57
Figure 4e Food source on right exhausted, while lower left corner food continues to be consumed; note upper left quadrant food ‘forgotten’ at this point
Figure 4f The strongest pheromone trail is in the upper left region and consequently the new foraging is concentrated in that area; lower left food is nearly gone
3.3.2 The ant cognitive visualisations and decision-making factors Important factors in the ant CVs of central versus decentralised control and decision making may also be used as analogies to understand factors associated with decision making in organisations and social systems. While the surface features of these two domains (ants and organisations) are highly contrasting, there are distinct structural
58
M.J. Jacobson
similarities related to decision-making factors such as locus of control, adaptability, efficiency, effectiveness, and so on. Given space considerations, only two decision-making factors are discussed (see Table 1). The column labels in Table 1 on organisation decision-making factors generalisations are drawn from the recent business and organisational literatures [19–21]. For example, Collins and Porras [19] contrast and compare business cases that are fascinating exemplars of centralised versus decentralised approaches for dealing with the control of large-scale organisations. According to their research, organisations that used more decentralised decision-making approaches, amongst other business practices, have been more successful than companies that employ authoritarian and bureaucratic approaches in their decision-making and business operations.
Table 1
Selected decision-making factors associated with central versus decentralised mental models Central control model
Decentralised control model
Decision-making factors
Ants
Organisations
Ants
Organisations
Locus of control
Single Queen ant initially searches for food; communicates centrally with pheromones.
CEO/project manager makes decisions, communicates expectations through a hierarchical chain of command.
Multiple ants independently search for food, communicate locally with pheromones.
Multiple individuals and/or teams in an organisation communicate directly with each other via a decentralised system.
Adaptability
Less adaptable since individual ants must wait new instructions if conditions change (e.g., found food source moves).
Less adaptable since individuals in the organisation must wait new instructions from top-level management if conditions change (e.g., found food source moves).
More adaptable since individual ant rules are always functioning even if conditions change (e.g., random searching will find food source if it moves).
More adaptable since individuals in the organisation follow shared values and beliefs (i.e., rules) that are robust even under changing situations.
From the perspective of the central control model (the first factor listed in Table 1, Locus of control), the food searching and communication decisions are made by a single ant, the Queen, for the entire ant colony, while a CEO or project manager makes and communicates decisions in an organisation. In contrast, the Locus of control for the decentralised model is distributed, with multiple ants independently finding food and communicating directly with each other via pheromones, while in decentralised organisations, multiple individuals and/or teams communicate directly with each other without going through a central hierarchy of command.
Cognitive visualisations and the design of learning technologies
59
For the second factor in Table 1, Adaptability, the command and control ant model would be less adaptable to changing conditions since individual ants must await new instructions from the Queen if/when conditions change (e.g., food source is consumed, the food source originally found by the Queen moves). Likewise, in an organisation, individuals using a command and control model would have difficulty adapting to changing conditions since they would have to wait for new instructions from top-level management in order to know how to proceed. In contrast, a decentralised model is more adaptable and flexible since the rules or behavioural norms governing how agents in a system behave are always functioning even if conditions change. Thus, even in changing environmental conditions (e.g., food sources depleted or rain washes food to a new location), ants will move randomly in the environment until food is found, then pick it up, deposit pheromones on the return to the colony, and so on. In a similar manner, a decentralised model is more adaptable for organisations since individuals are empowered to make individual decisions based on shared core values, principles, and procedures that are robust even under changing situations. (For discussions of the importance of shared values, beliefs, and principles to decision-making and the operation of organisations, see Collins and Porras [19] and Lissack and Roos [25]).
4
General discussion
The two CVs described in this paper were essentially proof-of-concept prototypes, the first in a study of learning evolutionary biology with a cognitively based technology system and the second in an applied setting involving consulting services. Regarding the research involving learning with the evolution CVs, these findings are interesting, but clearly provisional given the small sample size, lack of control for sequencing effects, and so on. However, despite these caveats, this data suggests learners identified with the CVs that had characteristics similar to their personal mental models about evolution. Further research could more systematically explore ways CVs might be used either as cognitive assessment tools (as in the pre-test reported here), or as part of learning activities in which students might interactively run CVs of naïve and expert mental models, contrast and compare differences in these models, and reflect on the relationship between the students’ ideas (i.e., their personal mental models) and the mental models represented by the CVs. In addition, research could investigate the hypothesis that the use of CVs might help students develop a metaconceptual awareness of their mental representations of knowledge and inferential processes, and that in turn might enhance the process of conceptual change in challenging domains of knowledge. Note there might be concerns that developing cognitive visualisations for helping students learn particularly challenging concepts at the pre-college and college levels in the natural and social sciences might not be ‘scalable.’ While there are a number of possible mental models learners might have, cognitive research investigating the nature of mental representations in a number of domains suggests that there are a limited set of mental models that students have at various stages of their learning trajectories in different subject areas [12,13,26,27]. Consequently, developing CVs for the more common mental models in different domains should be tractable both from a research perspective to identify common mental models students use when solving problems in difficult subject areas at different grade and learning levels (e.g., evolutionary biology,
60
M.J. Jacobson
thermodynamics, economics, history) and from a software development perspective given currently available visualisation, modelling, and animation tools that are available. Thus there is a need for systematic studies to explore the creation of CVs that are based on cognitive research into the characteristics and trajectories of learners’ (both students’ and adults’) mental models about challenging conceptual domains and about diverse educational and professional settings in which such representations might be used. By making mental representations of thinking visible and inspectable, it is hoped learners will develop richer conceptual understandings as well as metaconceptual skills that could foster conceptual change and knowledge transfer in challenging domains, whether as students or as adults. It will also be important to examine: 1
the technical integration of CVs with other types of learning technologies
2
the curricular and pedagogical integration of learning technologies with rich representational affordances and cognitive scaffolding features at various grade levels and professional development contexts.
5
Conclusion
This paper discussed cognitive visualisations as an approach in which the affordances of currently available learning technologies are used to reify or make explicit mental representations about conceptually challenging knowledge. CVs are intended to help learners, whether in school or out, to develop understandings of a domain through dynamic visual representations that are based on cognitive research into the nature of cognitive frameworks or mental models, in particular, models with the main characteristics of naïve or intuitive ideas and of expert understandings. An overview of the features and implementing technologies of two different prototype CVs was also provided. Finally, it is hoped that the proposed perspective of cognitive visualisations is generative as a design approach for learning technologies, one that is suggestive of ways that the sciences of learning and cognition and the sciences of technology and design might work together in the service of learning.
Acknowledgements This research was supported in part by grants from the National Science Foundation Applications of Advanced Technologies program (RED-9253157 and RED 9616389). In addition, the support of the Allison Group (formerly the Allison~LoBue Group) for portions of the work is gratefully acknowledged. Punya Mishra contributed both graphic design and content expertise to the development of the evolution cognitive visualisations. I am also grateful for the constructive comments and suggestions of two anonymous reviewers.
Cognitive visualisations and the design of learning technologies
61
References 1
2
3
4
5 6
7
8 9 10
11
12 13 14 15
16 17 18 19
Jacobson, M.J., Angulo, A.J. and Kozma, R.B. (2000) ‘Introduction: New perspectives on the designing the technologies of learning’, in M.J. Jacobson and R.B. Kozma (Eds.) Innovations in Science and Mathematics Education: Advanced Designs for Technologies of Learning, Mahwah, NJ, Lawrence Erlbaum Associates, pp.1–10. Jacobson, M.J. (1995) ‘Evolution, conceptual models, and hypermedia learning environments: designs for making thinking visible’, Paper presented at the Annual Meeting of the American Educational Research Association, San Francisco, CA. Jacobson, M.J. and Archodidou, A. (2000) ‘The design of hypermedia tools for learning: fostering conceptual change and transfer of complex scientific knowledge’, The Journal of the Learning Sciences, Vol. 9, No. 2, pp.149–199. Jacobson, M.J. and Archodidou, A. (2000) ‘The knowledge mediator framework: toward the design of hypermedia tools for learning’, in M.J. Jacobson and R.B. Kozma (Eds.) Innovations in Science and Mathematics Education: Advanced Designs for Technologies of Learning, Mahwah, NJ, Lawrence Erlbaum Associates, pp.117–161. Collins, A., Brown, J.S. and Holum, A. (1991) ‘Cognitive apprenticeship: making thinking visible’, American Educator, Vol. 15, No. 3, pp.6-11; 38–46. Jonassen, D.H., Reeves, T.C., Hong, N., Harvey, D. and Peters, K. (1997) ‘Concept mapping as cognitive learning and assessment tools’, Journal of Interactive Learning Research, Vol. 8, Nos. 3/4, pp.289–308. Eseryel, D. (2003) ‘Using causal influence diagrams to assess progress of learning in complex domains: an exporatory study in instructional design’, Paper presented at the Annual Meeting of the American Educational Research Association, Chicago. Jonassen, D. (2003) ‘Cognitive simulations with systems modeling tools’, Paper presented at the Annual Meeting of the American Educational Research Association, Chicago. Gordin, D.N. and Pea, R.D. (1995) ‘Prospects for scientific visualization as an educational technology’, The Journal of the Learning Sciences, Vol. 4, No. 3, pp.249–279. Pea, R.D. (1994) ‘Seeing what we build together: distributed multimedia learning environments for transformative communications’, Journal of the Learning Sciences, Vol. 3, No. 3, pp.285–299. Vosniadou, S. (1996) ‘Learning environments for representational growth and cognitive flexibility’, in S. Vosniadou, E. DeCorte, R. Glaser and H. Mandl (Eds.) International Perspectives on the Design of Technology-Supported Learning Environments, Mahwah, NJ:,Lawrence Erlbaum Associates, pp.13–24. Vosniadou, S. and Brewer, W.F. (1994) ‘Mental models of the day/night cycle’, Cognitive Science, Vol. 18, No. 1, pp.123–183. Bishop, B.A. and Anderson, C.W. (1990) ‘Student conceptions of natural selection and its role in evolution’, Journal of Research in Science Teaching, Vol. 27, No. 5, pp.415–427. Resnick, M. (1994) Turtles, Termites, and Traffic Jams: Explorations in Massively Parallel Microworlds, Cambridge, MA, MIT Press. Jacobson, M.J., Allison, M.A. and Ropella, G.E.P. (2000) ‘Agent-based modeling of disrupted market ecologies: a strategic tool to think with’, Paper presented at the Fourth International Conference on Complex Systems, Nashua, NH. STARLOGO available from: http://www.media.mit.edu/starlogo/ NETLOGO available from: http://ccl.northwestern.edu/netlogo/ AGENTSHEETS is available from: http://agentsheets.com Collins, J.C. and Porras, J.I. (1997) Built to Last: Successful Habits of Visionary Companies, New York: HarperCollins.
62 20
21 22 23 24
25 26
27
M.J. Jacobson Kelly, S. and Allison, M.A. (1998) The Complexity Advantage: How the Science of Complexity Can Help Your Business Achieve Peak Performance, New York, NY, McGraw-Hill. Senge, P.M. (1990) The Fifth Discipline: the Art and Practice of the Learning Organization, New York, Doubleday. Gentner, D. (1983) ‘Structure mapping: a theoretical framework for analogy’, Cognitive Science, Vol. 7, pp.155–170. Vosniadou, S. and Ortony, A. (Eds.) (1989) Similarity and Analogical Reasoning, New York, Cambridge University Press. Jacobson, M.J. and Angulo, A.J. (2000) ‘Complex systems, cognition, and problem solving: A preliminary investigation of differences between novices and experts’, in Y. Bar-Yam (Ed.) Proceedings of the Second International Conference on Complex Systems, Nashua, NH, New England Complex Systems Institute. Lissack, M. and Roos, J. (1999) The Next Common Sense: Mastering Corporate Complexity Through Coherence, London, Nicholas Brealey Publishing. Wiser, M. and Carey, S. (1983) ‘When heat and temperature were one’, in D. Gentner and A.L. Stevens (Eds.) Mental Models, Hillsdale, NJ, Lawrence Erlbaum Associates, pp.267–297. Vosniadou, S., & Brewer, W. F. (1992) ‘Mental models of the earth: a study of conceptual change in childhood’, Cognitive Psychology, Vol. 24, pp.535–585.
