This site contains a presentation reviewing the learning theory referred to as Cognitive Flexibility Theory (CFT) as well as a supporting literature review. This is in partial fulfillment of the requirements of the Cognitive Issues courses at Kansas State University. The professor for the course was Dr. Diane McGrath at Kansas State University in the fall of 2004. Here are some of the other cool theory sites by others who took this course.
The following is an outline of the site.
Cognitive flexibility theory (CFT) is a theory of learning initially described in a technical report by Spiro, Coulson, Feltovich, and Anderson (1988). CFT was formulated to address the factors that contribute to the failure to learn complex and ill-structured knowledge which is encountered at the advanced instructional levels (Spiro, Feltovich, Jacobson, and Coulson, 1992). In order to address this type of learning failure we need to avoid the inappropriate over-simplification of content. By doing so, we will improve learning and the transfer of complex knowledge (Jacobson and Spiro,1995).
Simplifying concepts in order to gain a level of understanding works fine in structured knowledge, however the pervasive role of oversimplification within a complex and ill-structured domain can lead to the development of misconceptions (Spiro, et. al., 1988). This general tendency to reduce important aspects of complexity is referred to by Spiro, et. al. as reductive biases which are listed below along with their remedies:
| Reductive Bias | Remedy |
| Oversimplification of complex and irregular structure | Avoidance of oversimplification by taking special measures to demonstrate complexities and irregularities. Highlight component interactions and demonstrate patterns of conceptual combination. |
| Over reliance on a single basis for mental representation | Provide multiple representations using multiple metaphors, multiple analogies, and offset misleading or negative effects of other cases. |
| Over reliance on “top down” processing | Use cases as the central point for learning. Cases are more than necessary and not just nice to have. They are irregular by their nature. |
| Context-independent conceptual representation | Provide conceptual knowledge that is knowledge in use. Tailor the concepts to their application context. Same concept should be used and applied differently across the varied set of cases. |
| Over reliance on precompiled knowledge structures | Provide flexibility in schema assembly. Avoid rigid knowledge constructs in favor of flexible knowledge that can be compiled in many different ways. |
| Rigid compartmentalization of knowledge components | Provide multiple interconnectedness by through the noncompartmentalization of concepts and cases. Enable situation dependent, adaptive schema assembly from multiple highly connected sources. |
| Passive transmission of knowledge | Cultivate active participation. Provide opportunistic expert guidance through an automated agent. |
Nelson ( 1994) identifies three hypermedia environments that support knowledge construction. These include case-based, exploratory, and cooperative hypermedia. Case-based hypermedia support many of the features of CFT in that it provides the multiple representations of knowledge. Agents can be provided as guides to throughout the system such as a scout who might demonstrate multiple alternatives to solve a problem. Multiple perspectives can be built into the system. For example: When dealing with difficult students in the classroom, a teacher, administrator, parent, or expert will each have a different perspective how how to handle these students. These perspectives can be part of the hypermedia application.
Exploratory hypermedia are exemplified by microworlds or situated problem-solving environments such as taking a trip to Spain in order to learn Spanish. Cooperative environments are built around a collaborative engine that provides the ability to share personal knowledge, a World 3 feature. A well published example is the Computer-Supported Intentional Learning Environment (CSILE) which is described in (Scardamalia, Bereiter, & Lamon, 1994).
Yang (1996, November-December) identifies four properties of constructive hypermedia. These properties include the following:
Hypertext systems allow us to easily build indexes across the text that supports the crisscrossing mechanisms that CFT suggests that we need in order to support the learning of difficult concepts. However, simply employing multiple ways to traverse the content will not necessarily lead to improved learning. In fact it may even lead to a "confusing labyrinth of incidental or ad hoc connections" (Spiro, Feltovich, Jacobson, and Coulson, 1992). Instead we need to organize the material in way that a learner "sees a range of conceptual applications close together, so that conceptual variability can be easily examined" (Spiro,et. al., 1992). Once again, this feature can be accomplished if we employ the effective use of indexes to link related concepts together.
In general CFT can be used as an underlying learning theory in the design of case-based, exploratory, or constructive hypermedia applications. What distinguishes each of these is the ability to construct new knowledge and place it back into the learning environment. A hypermedia environment that provides a concept mapping feature grounded in CFT is an example of constructive hypermedia. Without the concept mapping feature, it may simply be an exploratory hypermedia application. Alternatively, if we wrap this around a library of cases that can be rebuilt depending upon our access criteria with additional capabilities of constructing the new concept maps that tie the cases together, then this would be an example of a constructive, case-based hypermedia application.
Currently there is development on just this type of application (i.e. a constructive, case-based application) by Cañas, Leake and Maguitman (2001). This application allows the learner to build concept maps based upon remembrances of previously built concept maps. The algorithms for search and retrieval are based upon isomorphic properties of the graph. For a detailed description of these matching algorithm see (Cañas, Leake and Maguitman, 2001).
When we tell tell stories we are using one of the most natural forms of "meaning making" (Brunner, 1990). Jonassen (2002) enumerates eleven functions of story telling that have been described in research literature. It is interesting to note that story telling is a constructivist learning method and it can be easily used to support CFT. This is so because stories can be interlinked in a way such that they have all or part of the five antidote features to the oversimplification tendencies (Jacobson and Spiro, 1995).
Jonassen (2002) lists three methods where stories can be used to support learning. First, they can be used as exemplars of concepts, principles, or theories during direct instruction. Second, they can be used as problem cases to be solved by students. And, third, they can be used as advice to help students solve the problems which directly supports case-based teaching. In each of these methods, the stories can be integrated within a hypermedia environment in such a way that they can be accessed by various means. The structuring the accessing mechanisms, indexing mechanism, search mechanism over the stories is a considerable instructional design issue. In particular the indexes to the stories can be based upon the following (Jonassen, 2002):
These indexes can be populated by either (1) hard links in a hypermedia application or (2) they can be encoded within a database with search mechanisms built into the application. The second approach uses a database to store both the actual case as well as the indexes to the case. In a simple scenario a user of the database of cases (case-base) may simply want to access one index as part of a search. All related cases can be provided as a result to the query. In more complicated scenarios a user may wish to use more than one index. In these cases an intelligent search algorithm is employed which searches the case-base looking for matches based upon similarities of the search query and the indexed cases. Probably the most often used intelligent algorithm is the nearest-neighbor algorithm which employs a Euclidian distance metric for similarity. The resulting list is typically presented in descending relevance order.
Jacobson and Spiro (1995) performed a study that explored epistemic cognition within the context of using a hypertext learning environment. Epistemic cognition refers to the student beliefs about the nature of learning and the structure of the knowledge (Jacobson and Spiro, 1995). The hypertext materials used in the study consisted of two main parts. In part one, the reading stage, both the experimental and control groups read the same instructional content in a hypertext format based upon the CFT design features of (1) multiple knowledge representations, (2) linking of abstract concepts to case examples, and (3) the introduction of domain complexity early. In part two the experimental group received a hypertext treatment, Thematic Criss-Crossing Hypertext. For the experimental group this second part implemented (4) stress knowledge interrelationships and (5) emphasize knowledge assembly of the CFT design features. Also at this stage two Minimal Hypertext/Drill control groups completed a computer-based drill on facts and thematic concepts taken from the reading stage. It was the study's intent to provide for contrasts in instructional activity between the two groups.
Jacobson and Spiro (1995) hypothesized that the Minimal Hypertext/Drill control groups, which received structured practice over prespecified material, would achieve higher scores than the experimental group on factual knowledge. They also hypothesized that the Minimal Hypertext/Drill control group would not perform as well as the experimental group on the transfer tests. Jacobson and Spiro also hypothesized that some learners would prefer instructional approaches that tend toward content simplification as well as approaches that stress the memorization of prespecified knowledge. These individuals would be less able to learn in the experimental format which stressed all five of the CFT principles.
The results of Jacobson and Sprio's (1995) experiment showed that the Thematic Criss-Crossing Hypertext group (the experimental group) performed significantly better on essay tests that required knowledge transfer. Jacobson and Spiro suggested that an instructional hypertext system that demonstrates critical conceptual interrelationships and assembly of different case-specific and abstract knowledge components does a better job in preparing students to use their knowledge in new ways and new situations. The study also found that epistemic beliefs may enhance or constrain the educational effectiveness of a hypertext learning environment. Jacobson and Spiro recommended further research into this.
A followup study on epistemic beliefs was performed by Jacobson, Maouri, Mishra, and Kolar (1995). They found that providing students with flexible access and a high degree of user control to hypertextually interconnected materials is not enough. Some specific hypertext design features were found to significantly influence learning outcomes on different knowledge acquisition and transfer tasks. Additionally their study did not find a significant difference on the essay tests that required knowledge transfer between the two groups except for those students who preferred complex knowledge presentation (referred to in this study as high EBP students) and those who preferred content simplification (low EBP students). For those two groups, the difference in their performance was significant.
The followup study (Jacobson, et. al., 1995) identified some of the reasons for outcomes of their study. They suspected that three interacting factors contributed to their findings: (1) the structure of knowledge represented in the hypertext, (2) the nature of the learning activity, and (3) the cognitive support provided.
In life people make goals, they have an intentional learning perspective, and they make plans on how to achieve these goals (Schank and Cleary, 1995). Solving the problems which allow us to meet these goals is facilitated through past experiences. In our mind we remember these experiences. CFT provides us with a theory of organization for these experiences that can be used by a learner to support knowledge construction and transfer (Spiro, et. al., 1992). Peer review provides the learner with feedback and helps he or she learn from these evaluations. These are the theories upon which Kolodner's Learning by Design (LBD) project is built upon (Kolodner, 2003).
LBD is an instructional design endeavor that has been applied to the learning of science in middle school students (grades 6 to 8; ages 12 to 14) that has been demonstrated to be effective. LBD takes suggestions from cognitive apprenticeship, problem-based learning, and fostering communities of learners approaches. It situates science learning in the context of designing devices. The construction and trial of real devices provides students opportunity to experience uses of science and to test their conceptions. Through this testing the learner is able to revise and repair their models based upon these experiences and the peer review process provided within the LBD design.
Two iterative learning cycles are used to provide feedback to each other. These cycles are (1) design/redesign and (2) investigate/explore. Both cycles are grounded in case-based reasoning. Wrapping these cycles in a problem-based learning context provides the LBD framework. In a nutshell students gain understanding of the problem by messing around with the devices that are used to solve problem. This understanding allows them to more clearly identify what needs to be investigated in order to meet the challenge. Groups then present their findings to the teacher as well as the class. The teacher helps the class identify heuristics or rules of thumb about investigative procedures as well as results. Students then investigate and explore a question based upon the research rules, hypothesis, and questions generated in the design/redesign cycle.
Kolodner (2003) reported that students spontaneously make reference to previous experiences over the course of several months of LBD activities. In particular they reference both skills and practices. LBD students are more capable of engaging as scientists and collaborators than non LBD students. Finally, D'Avolio (2003), in his review LBD and its related research reported that mixed-achievement LBD students performed as well or better than non-LBD honors students.
D'Avolio (2003) points out some of the unique challenges that are presented in the application of LBD. These challenges are enumerated as follows:
Kolodner (2003) also mentions similar challenges. She divides them into three categories including teacher preparation, assessment, and time. In order to support teachers new to LBD they need to first experience LBD. An understanding of the cycles and rituals of LBD including iteration, doing, reflecting, etc. need to be acquired by the teacher. Teachers need the support of learning the content of what they are teaching, the experience in constructing and testing everything the students construct, and the experience of trying out their skills with children. These are not simple issues.
Assessment also changes in an LBD environment. Grade assignment needs to take into account both results and the skills used to get those results. Additionally consideration needs to be given to how the student performs as an individual as well as in groups.
Time or time-on-task is the final challenge listed by Kolodner (2003). Deep learning is hard to do in science because of the standards requiring breath of knowledge versus depth. Using PBL helps as part of LBD does help in this area.
Bruner, J. (1990). Acts of Meaning. Cambridge: Harvard University Press
D'Avolio, L. (2003) A review of the hybrid CBR/PBL approach to technology in the classroom. [Online] Available: http://polaris.gseis.ucla.edu/ldavolio/CBR-PBLInTheClassroom-2003.pdf
Jacobson, M. J., & Archodidou, A. (2000). The design of hypermedia tools for learning: Fostering coceptual chage and transfer of complex scientific knowledge. The Journal of the Learning Sciences, 9 (2), 145-199.
Jacobson, M. J., Maouri, C., Mishra, P., & Kolar, C. (1996). Learning with hpertext learning environments: Theory, design, and research. Journal of Educational Multimedia and Hypermedia, 5 (3/4), 239-281.
Jacobson, M. J., & Spiro, R. J. (1995). Hypertext learning environments, cognitive flexibility, and the transfer of complex knowledge: An empirical investigation. Journal of Educational Computing Research, 12 (4), 301-333.
Jonassen, D. H. (1989). Hypertext/Hypermedia . Englewood Cliffs, NJ: Educational Technology Publications. {pp. 40-50: Problems in Using and Authoring Hypertext, and pp. 50-62: Designing hypertext}.
Jonassen, D.H. (2002). Case-based reasoning and instructional design: Using stories to support problem solving. Educational Technology Research and Development, 50(2), 66-77
Joyce, M. (1988, November). Siren shapes: Exploratory and constructive hypertexts. Academic Computing, , 10-14; 37-42.
Kolodner, J. (2003). Facilitating the Learning of Design Practices: Lessons Learned from an inquiry into science education. Journal of Industrial Teacher Education, 39(3). Avaliable: http://scholar.lib.vt.edu/ejournals/JITE/v39n3/kolodner.html
Alberto Cañas, David B. Leake and Ana Maguitman (2001). Combining concept mapping
with CBR: Towards experience-based support for knowledge modeling. Proceedings of the Fourteenth International Florida Artificial Intelligence Research Society Conference . AAAI Press, Menlo Park, pp. 286-290. Available as a pdf from: http://www.cs.indiana.edu/~leake/papers/p-01-04.pdf
Nelson, W. A. (1994). Efforts to improve computer-based instruction: The role of knowledge representation and knowledge construction in hypermedia systems. Computers in the Schools, 10 (3/4), 371-400.
Scardamalia, M., Bereiter, C. & Lamon, M. (1994). The CSILE Project: Trying to bring the classroom into World 3. In K. McGilly (Ed.). Classroom lessons: Integrating cognitive theory and classroom practices, p. 201-228. Cambridge , MA : Bradford Books, MIT Press. Available as a pdf file from: http://fcis.oise.utoronto.ca/~ewoodruff/papers/csile.pdf.
Schank, R. C., & Cleary, C. (1995). Engines for education. Retrieved June 10, 2004 from http://www.engines4ed.org/hyperbook/
Spiro, R. J., Coulson, R. L., Feltovich, P. J., & Anderson, D. K. (1988). Cognitive flexibility theory: Advanced knowledge acquisition in ill-structured domains. (Technical Report No. 441). Champaign , Illinois : University of Illinois at Urbana-Champaign.
Spiro, R. J., Feltovich, R. P., Jacobson, M. J., & Coulson, R. L. (1992). Cognitive flexibility, constructivism, and hypertext: Random access instruction for advanced knowledge acquisition in ill-structured domains. In T. M. Duffy, & D. H. Jonassen (Eds.), Constructivism and the technology of instruction: A conversation (pp. 57-76). Hillsdale , NJ : Erlbaum.
von Glasersfeld, E. (1990). An exposition of constructivism: Why some like it radical. In R.B.Davis, C.A.Maher, & N.Noddings (Ed.), Monographs of the J. for Research in Mathematics Education, #4 . Reston, VA: National Council of Teachers of Mathematics, 19-29.
Yang, S. C. (1996, November-December). Designing instructional applications using constructive hypermedia. Educational Technology, , 45-50.