Improving Teaching and Learning Outcomes – A Novel Cognitive Science Approach

Three principle approaches to education are behaviorism, constructivism and cognitive science. However beyond that there are a wide and diverse range of methods such as outcomes based education, problem based learning etc. All these approaches and methods have their associated advantages and disadvantages. The effectiveness of the educational process is measured by a learning taxonomy such as Bloom or Structure of the Observed Learning Outcome (SOLO). The paper links the cognitive science approach with the SOLO taxonomy as the basis of a quantitative approach to measuring cognitive load which is a measure of mental effort. This can then be used as the basis of curriculum design in order to optimize learning outcomes.


Teaching and Learning theories
In order to effectively write course materials and teach it is essential to have an understanding of how students learn.Learning theories can be organized into three main groups: behaviorism, constructivism and cognitive science.However other classification groupings are possible.The basis of behaviorism is operant conditioning.According to Skinner the desired behavior is encouraged and reinforced by operant conditioning (Skinner, 1948).Skinner identified four types of operant conditioning: negative and positive reinforcement, punishment and extinction.Hence Behaviorism emphasizes the importance of practice and is the basis of programmed instruction in which new material is provided in a set of discrete segments called frames each with an associated question or set of questions (Dick, 1990).The basis of Constructivism is that each student has a different perception and 'constructs' new knowledge based on their existing knowledge.There are two main approaches to Constructivism, namely psychological (aka individual) and social.Psychological constructivism is primarily concerned with individual learners (Piaget, 1952).Piaget defined the four processes of: assimilation, accommodation, disequilibrium and equilibration as the basis of learning.By contrast the social constructivist approach places learning in the broader context of social interactions (Vygotsky, 1978).The cognitive science approach is based on how people process information based on the interaction between short and long term memory and the development of mental 'maps' called schemata (Rumelhard, 1980).Each theory has advantages and disadvantages but the strength of the cognitive science approach is that it is possible to employ a quantitative approach (S.P. Maj, Veal, D., 2010 ).Maj demonstrated that it was possible to quantitatively evaluate course materials and also optimize the learning sequence.However this work did not employ a learning taxonomy.

Teaching and Learning taxonomies
The efficacy of teaching and learning is measured by a learning taxonomy such as Bloom (Bloom, Engelhart, Furst, Hill, & Krathwohl, 1956).Bloom identified taxonomies for the cognitive, affective and psychomotor domains.However this paper is concerned only with the cognitive domain.The cognitive taxonomy defines six major groups ranging from low order learning to high order learning namely: knowledge, comprehension, application, analysis, synthesis and evaluation.To facilitate use of this taxonomy action verbs are used.For example knowledge sample verbs could be: list, name and label.By contrast the evaluation action verbs could be: critique, evaluate etc. However according to the Assessment for Teaching and Learning Team at the University of Auckland, 'Despite the popularity of Bloom cognitive taxonomy, there is little support for the use of it in mas.ccsenet organizing by Bloom taxonomy type of lea defines lev relational as surface relational a the whole hierarchy 2004).For learner lin the hierarc with high been appli 1986).

Linking
The cognit In the proc  duration.The schema, relevant to the subject being studied, is resident in long term memory and during the process of learning it is retrieved and the new knowledge is assimilated thereby modifying the schema (Figure 1).In effect learning is the iterative process of retrieve and assimilate.The advantage of this approach to learning is that it explicitly recognizes the roles of the two types of memory -long term and short term.This is important because this model clearly identifies the intrinsic limitations of short term memory which must inform teaching and learning principles and practices.Structural knowledge (knowing why) links declarative and procedural knowledge (Diekhoff, 1983).Hence structural knowledge, also called cognitive or internal knowledge, is represented by the patterns in schemata.The underlying principles are based on data similarities also referred to as semantic proximities (Nagy, 1984).As such the objective is a quantifiable measure of closeness (Keppens, 2008).There exist a wide range of theories and associated methods associated with structural knowledge but they share the common goal of controlling the complexity of the relationships that exist in structural knowledge (Jonassen, 1993).Methods include: hierarchical structured relationships (Meyer, 1985); slots and frames (Minsky, 1975); elaboration theory (Reigeluth, 1979); special semantic maps (Larkin, 1987) to name but a few.Methods such as these are designed to identify and improve poorly designed content.The quality of the content may be measured using the SOLO taxonomy.Poorly designed material may be represented by data items with few or poorly structured relational links (figure 2).In this figure each node represents an item of knowledge such a hardware address is OSI layer 2 address; an IP address is an OSI layer 3 address.However these isolated nodes represent rote or low order learning.They have no cognitive links and represent isolated items of information.
Only when these nodes are linked to other nodes is there multi-structural leaning.For example the Address Resolution Protocol (ARP) maps hardware and IP addresses.This would be represented diagrammatically by a link between these nodes.Higher order learning therefore can be represented by an extensive pattern of inter-related nodes linked in a meaning full manner that represents understanding.In effect it is possible to traverse the nodes to answer questions like 'why do hardware and IP addresses have to be mapped?'Links in conjunction with nodes represent reasons or explanations.However it possible for the cognitive map to be poorly structured with missing cognitive links and poorly structured.In this case this represents a less than optimal learning structure with a high cognitive load: in effect students are left to create their own cognitive links which could well be incomplete, inconsistent or incorrect.However as an isolated node this would be considered rote or low order learning.There are no cognitive links; it is an isolated piece of information.Only when this node is linked to other nodes is there multi-structural learning.In the case of ARP this is a protocol that maps IP addresses to hardware addresses.Clearly this represents high order relational knowledge because of the presence of relational links, but it is poorly structured and hence represents a less than optimal learning structure as such this represents a high cognitive load, furthermore the student is left to create their own cognitive links which are likely to be incomplete, inconsistent and incorrect.Ideally content should have an optimal relational learning structure with minimal cognitive load (figure2).

Minimizing cognitive load -a quantitative method
There are a wide range of methods for organizing learning structure such as: pattern and spider maps (Hanf, 1971); semantic maps (Fisher, 1990); cross classification tables (Frase, 1969).However according to Maj 'All the techniques and methods could be used to represent and hence define explicit structured knowledge for routing protocols.Most were easy to learn and use supporting for some methods.However none of the methods provided rigorous methods for defining concepts and characters.Furthermore none of the methods provided a rigorous, quantifiable mechanism for determining and calculating structured relationships' (S.P. Maj, Veal, D., 2010 ).Maj proposed a new quantitative method, based on a parsimony algorithm that could define a logical learning sequence with the minimum cognitive load.

Cladistics
Cladistics is a method of classification in biological systems that group taxa (named group) hierarchically into discrete sets and subsets.The aim is to hypothesize relationships that may be expressed as a branching diagram (cladogram).Taxa are defined by common characteristics that may be: plesiomorphic (primitive and more inclusive) or apomorphic (advanced and less inclusive).The problem is to obtain the optimal cladogram given there can be a large number of characters that can be organized in many different ways which is computationally intensive.The Phylogenetic Analysis Using Parsimony (PAUP) was designed for biological systems in order to arrange organisms into a hierarch in the simplest and most parsimonious, quantifiable manner and runs on a standard PC.The Cisco Network Academy Program (CNAP) is recognized as a global exemplar of best practices in IT education.Maj applied PAUP to a CNAP learning topic (routing) and found that it was a less than optimal learning sequence.Using PAUP the optimal learning sequence for this topic as map and proven to be optimal (S.P. Maj,

Optima
If course m and hence order learn course ma PAUP defi learn the to

Linking
The Figure 3

Figure
Figure 4. L