IDENTIFYING COGNITIVE TRAITS AND LEARNING STYLES

To incorporate cognitive traits and/or learning styles in educational systems, information about cognitive traits and learning styles need to be first collected. One approach is to let students perform comprehensive tests or questionnaires to find out the cognitive traits or learning styles. Such an approach; however, has potential to suffer from the biases and indecisiveness of the learners. A more meaningful approach is to track the students’ behaviour and infer the required information from their behaviour. Cognitive Trait Model (Kinshuk & Lin, 2004; Lin & Kinshuk, 2005) uses this approach to profile learners according to their cognitive traits. For the identification of learning styles, approaches for detecting the dimensions of Felder-Silverman learning style model are introduced.

Identification of Cognitive Traits

Cognitive Trait Model (CTM) is a student model that profiles learners according to their cognitive traits. Four cognitive traits, working memory capacity, inductive reasoning ability, processing speed, and associative learning skills are included in CTM so far. The CTM offers the role of ‘learning companion’, which can be consulted by and interacted with different learning environments about a particular learner. The CTM can still be valid after a long period of time due to the more or less persistent nature of cognitive traits of human beings (Deary, 2004). When a student encounters a new learning environment, the learning environment can directly use the CTM of the particular student, and does not need to “re-learn the student”.The identification of the cognitive traits is based on the behaviour of learners in the system. Various patterns, called Manifests of Traits (MOT), are defined for each cognitive trait. Each MOT is a piece of an interaction pattern that manifests a learner’s characteristics. A neural network (Lin & Kinshuk, 2004) is responsible for calculating the cognitive traits of the learners based on the information of the MOTs.

Identification of Learning Styles

There are a number of adaptive systems available in the literature incorporating learning styles. For example, CS383 (Carver, Howard, & Lane, 1999) was the first adaptive hypermedia system that was based on Felder-Silverman learning style model (FSLSM). The course conducted in the system included comprehensive collection of media objects. The system offered students the option to order these objects in accordance with how well they fit to the learning style of the student. Also MAS-PLANG (Peña, Marzo, & de la Rosa, 2002), a multi-agent system which has been developed to enrich the intelligent tutoring system USD (Fabregat, Marzo, & Peña, 2000) with adaptivity with respect to learning styles is based on FSLSM. Another example is INSPIRE (Papanikolaou & Grigoriadou, 2003) that is based on Honey and Mumford’s learning style theory. In all these systems and in most other systems which incorporate learning styles, the learning style is identified based on a questionnaire that needs to be filled out by learners before using the system. These questionnaires are based on the assumption that learners are aware of how they learn. Jonassen and Grabowski (1993, p. 234) pointed out that “because learning styles are based on self-reported

measures, rather than ability tests, validity is one of their most significant problems”. García, Amandi, Schiaffino, and Campo (2006) studied the use of Bayesian networks (Jensen, 1996) to detect students’ learning styles based on their behaviour in the educational system SAVER. Based on the Felder-Silverman

learning style model, they determined patterns of behavior, which are representative for the respective dimensions, as well as the different states these variables/patterns can take. Because SAVER does not incorporate the visual/verbal dimension, this dimension is left out from investigations. 

While the above approaches are developed for specific systems, Graf and Kinshuk (2006) proposed an approach to detect learning styles in learning management systems (LMS) in general. Equal to the approach by Garcia et al. Felder-Silverman learning style model is used as the basis but in this case all four dimensions are incorporated. The patterns of behaviour are derived from commonly used features in LMS such as forums and exercises. Regarding the calculation of the learning styles, the approach used in the Index of Learning Styles (Felder & Soloman, 1997), a questionnaire for identifying the learning style according to Felder-Silverman learning style model, is applied.