Srikumar Lakshmipathi
Ramanan P. S.
Uday Sreekanth
The Problem: Limitations of traditional modes of instruction in complex domains ----------- Traditional methods of instruction use LINEAR methods for imparting knowledge. These methods are not very useful for conveying post-introductory learning in domains that are ill-structured and complex. Post-introductory learning involves a shift from attaining superficial familiarity to a mastery of the subject matter and a shift from knowledge reproduction to knowledge use. It calls for the ability to adaptively re-assemble diverse elements of knowledge to fit the particular case in hand. The author labels this "cognitive flexibility". Formally defined, cognitive flexibility is the ability to spontaneously restructure one's knowledge, in many ways, in adaptive response to changing demands of the situation. Cognitive Flexibility Theory is presented as a means to foster knowledge acquisition in a complex domain. This paper focuses on a single illustrative context- a computer program in the area of literary comprehension, with an understanding of the complex patterns of individual behavior. The meaning of a literary work is more than a mere sequencing of events (i.e., linearity), since their layout does not reduce to a single interpretation. Representing non-sequential patterns of themes is perhaps the most useful scaffolding. Cognitive Flexibility Theory ---------------------------- Wittgenstein's metaphor - "the criss crossed landscape" forms the foundation of the cognitive flexibility theory. Wittgenstein in (his book) "Philosophical Investigations", treated the different topics as different sites in a landscape. These different sites were then arranged in the form of an album. The sequences in the album represented different traversals of the landscape. By rearranging the sequence of sketches presented in the album, new dimensions of the landscape can be observed or studied. The multifaceted nature of the landscape becomes clear after a number of traversals, when the student has criss-crossed it in many directions. Extending the idea, in cognitive flexibility theory, one learns akin to "criss-crossing conceptual landscapes". Issues involved are analyzed in different contexts and from different perspectives. The instructional medium makes such criss-crossing possible and knowledge representations reflect the "criss-crossing that occurred during learning". The criss-crossing nature leads to knowledge structures that can be likened to a weaving in the form of a web. This implication of high interconnectivity accounts for flexibility in application. Also, such structures can be easily re-assembled for problem solving in new domains. Hypertext is used as a means to impart these knowledge structures. The Method: Cognitive Flexibility Hypertexts ----------- Cognitive Flexibility Hypertexts are computer based instructional programs that are born from Cognitive Flexibility Theory and are built to carry out its operations. The computer's ability to utilize the highly multi- dimensional and web-like structure in multiple ways renders the term "RANDOM ACCESS INSTRUCTION" to this kind of an educational set-up. To illustrates the use of this technique in the area of literary comprehension, the classic film "Citizen Kane" has been considered as an example. A computer program "KANE" is detailed, that helps a student develop a better understanding of some of the finer aspects of the film. The program bears particular emphasis to Kane's character (i.e, his motivations, feelings, values etc. through the progression of the film.) KANE: Knowledge Acquisition by Nonlinear Exploration ---- KANE was programmed using something termed as the HANDY authoring language. It is designed for use by advanced high school to college level students and assumes that the student has ALREADY VIEWED the film. On starting the program, students can select from one of many themes. "Each theme has been put forward by some experts as being capable of providing a complete account of the Kane character's behavior, motivations, failings, and so on". For those familiar with the movie, some examples of themes include "Hollow Man", "Wealth Corrupts", and "Outsized Ambition". On selection of a theme, the student is presented with a list of scenes that are meaningful to the theme selected. The student also has access to additional themes that a scene portrays. Every scene combines a segment of the film with a textual commentary on the theme it represents. By viewing the same scene under different themes, students develop a (comprehensive grasp ?) better understanding of the subtletilies and nuances associated with some of the finer aspects. Interestingly, restriction is absent and the students are given the freedom to define and form their own themes. This feature would be particularly useful to instructors wanting to educate students on a particular aspect of the literary piece under question. Two more points need to be made- (a) The student has several options after the presentation of a scene. If the non-sequential rendition has introduced confusion there is the capability to review what was exactly seen in time sequence. Similarly, there are other useful options such as reviewing the history of commands, organize non-sequential clusters and view these sequentially, etc. (b) The student can re-edit the enitre film as a function of theme- i.e., only scenes relevant to that theme will be shown. Knowledge transfer with KANE: ----------------------------- Complex knowledge comprehension is enabled by introducing "bite-size chunks". In the course of traditional instruction, early simplifications create impediments to the mastery of complexity introduced later. Complexity has to be therefore presented earlier and this is embodied in KANE. Students are able to deal with this complexity as each scene presented runs only for about 30 to 40 seconds and is presented in context and with a full commentary. A scene (mini-case) is the fundamental unit of instruction. The paper identifies several advantages to using mini-cases as the primary instructional organizing unit. These are - i) The process of acquiring experience is enhanced. By presenting mini-cases the system is able to present to a 45-50 second scene to the learner, with the analysis that follows dealing with a variety of aspects and theories that need to be learned. Analysis of several such cases, enable the learner to gain vast experience in the related contexts. ii) The system, does not break down the problem into "simpler" components- rather, the problem is split as "bite-size chunks", that allow the user to deal with the complexity of the context in small amounts. This is referred to as "reconceptualization of instructional incrementalism". iii) The division of a case into several mini-cases breaks down the integrity of an intact case, in that it reduces the chances of the learner to overly rely on a intact case as a prototype. Mini-cases help the learner to grasp the various aspects of an intact case. Also, it leads to enhancing the multiple mapping of the necessary aspects of each mini-case. iv) Mini-cases facilitate domain-specific knowledge retrieval, which is necessary while dealing with ill-structured domains. The cognitive processes in analysis and understanding of mini-cases is much simpler than the analysis of a much larger case, and this arguably assists the process of domain specific knowledge retrieval. v) Mini-cases facilitate an in-depth analysis. In contrast, large cases do not assist the process of in-depth analysis, which means that some important aspects that are relevant in the present context may go unnoticed. Detailed analysis is necessary for better understanding of all domain-specific principles, and this assists the learner to apply the principles at a later stage. vi) System development is made easier with the use of mini-cases, for e.g. the coding of mini-cases, assists the automatic generation of several case contrasts and traversal routes (in what is claimed to be) an exponential rate. Miscellaneous Points -------------------- - Through providing 10 equal, frequently applicable themes in KANE, the system creators wish to portray their view that no single theme permits a total view of the scenario. - A non-replicative repetition scheme helps present the same context in varying cases. This (unlike traditional methods where this is employed to strengthen learning of a particular aspect), indicates the different shades of meaning that are possible in different views. - The idea of non-structured instruction relies on knowledge dissemination through repeated demonstration rather than a single abstract instantiation which is possible in structured environments. - The work stresses the equi-importance of concepts and cases, with conceptual knowledge taught in the context of actual cases of its application. A sentence we liked was "hypertext integrates conceptual and situational learning, in which each is thought about in terms of the other." - The domain causes conceptual combination to be the rule rather than the exception. This vitiates the compartmentalization of knowledge, overcoming biases created by insularity. - Two sub-theme occurrences can be grouped together and explored using a theme combination search to validate the grouping. This is a simple approach to dealing with what is otherwise a non-trivial issue. - Special initiative options can be used to "color" a theme. What this means is that a specially programmed sequence with thematic overlays can be used to guide analytical thinking, while at the same time subjecting the session with the topic of the special initiative. - Problems concerned with getting lost in hyperspace are remedied through a case-centered scheme where the user is never one step beyond the focus of instruction, since each mini-case begins a complete and independent unit of instruction.Uday Sreekanth uday@cc.gatech.edu