Cognitive theories of learning have evolved from experimental psychology and find their roots in the study of behaviour. The early emphasis was on measuring responses to external stimuli and the so called behaviourist approach was the dominant paradigm up until the latter part of the last century. Much pioneering work in experimental learning psychology was based on observing results from animal and human experiments typically involving the quest to understand relationships between stimuli and observed responses (Hilgard & Bower, 1966; Mackintosh, 1997; Travers, 1967). These insights into learning focused only on external observable behaviour.
Early exponents of behaviourism such as John B Watson, Edward L Thorndike and B. F. Skinner have contributed much to our understanding of some of the basic characteristics of learning. They achieved this by focusing almost exclusively on the observable interaction between the learner and the external environment. Behaviourists limited their enquiry to what could be measured or recorded by experiment. They considered the internal workings of the mind to be beyond the bounds of observation and therefore disregarded it. This was later seen as a severe limitation. However, viewed in the context of the knowledge and technology available at the time it might also be seen as a very practical approach. In contrast, today’s cognitive theories are based on models of the internal processes of the brain and are usually referenced to advanced experimental and computer modelling techniques. Both behaviourist and cognitive approaches largely ignore the emotional and social dimensions of learning.
Some useful instructional principles derive from behaviourist approaches to learning. Let’s consider what these theories can contribute to the specific case of Peter. Suppose the challenge for Peter is one of learning a new task for example using a dictionary to find the meaning of a new word. Probably the major contribution of the behaviourist approach is the concept of task analysis (John R. Anderson, 2000: 383) -the instructional strategy of decomposing the task into constituent sub-tasks and categorising these. In this case the task of using the dictionary to find the meaning of a word would be decomposed into sub-tasks such as understanding alphabetical order and scanning the word lists to find the target word.
The notion of specifically identifying and naming learning objectives also originates in behaviourist theory (Leigh, 2004; Reiser, 2001). Tenant (2006) includes a useful discussion on the value and limitations of behavioural objectives in instructional approaches. He notes that objectives are frequently cited in terms of observable and measurable outcomes. This is a common practice in e-learning whereby specific objectives, including pre-requisite or enabling objectives, for each unit of instruction are stated at the outset. Tenant questions the validity of using such a structured approach to instruction. He provides an example of mastering the complex skill of playing the piano –in this case, for different people, learning can proceed in a multitude of different directions such as posture, finger position, notation, scale drill and so on. The key to mastery is how these skills come together to form an integrated whole and following sequentially arranged specific objectives will not ensure mastery as this happens at different times and in different ways for different learners. Instructional analysis such as tabled by Gagne (1977) is based on the assumption that learning outcomes can be deconstructed into specific types of objectives and that instructional strategies should be matched to the characteristics of the objectives. Later theoretical approaches became more concerned with recognising individual learner characteristics.
Problem Solving Approaches
Cognitive science has been supported by new experimental techniques the relevant area of inquiry has moved beyond observable behaviour to deal with the internal processes of thinking and learning. A key foundation was the work of Newell and Simon (cited in John R. Anderson, 2000; , 1972) who proposed a model of learning based on problem solving and drawing on the emerging fields of computer simulation and artificial intelligence. Newell and Simon’s learning paradigm was encapsulated in what they called the General Problem Solver or GPS. The process involved applying a means-end analysis to a given learning task in a sequence of logical steps. For example applied to the dictionary task above if Peter wants to find the meaning of a new word –this is the goal. Peter selects some action that reduces the difference between his present state and the goal –this action is described by Newell and Simon as the ‘operator’. In this case the operator is ‘look at word description in dictionary’ and if this can be applied then Peter applies it. If the operator cannot be applied, then the next step is make a new goal to enable the operator –in this case find the word in the dictionary -thus dealing with alphabetical order becomes the means to the primary goal. Newell and Simon’s approach is the foundation for what is often referred to as an information processing model of learning. From this several new theories of cognitive architecture emerged and the scientific study of mental processes has evolved.
An example of one such cognitive approach is Anderson’s adaptive control of thought –rational or ACT-R theory (John R. Anderson et al., 2004; John R. Anderson & Lebiere, 1998). ACT-R and its derivatives are examples of cognitive models they are attempts to describe the internal architectures of the mind and in this way contribute to an understanding of the learning process. ACT-R theory is based on a complex model of human thought and is often represented as a computer simulation. It is in fact a theory of cognition rather than a learning theory.
Declarative and Procedural Knowledge
An important underlying concept is the distinction between of two types of knowledge -declarative knowledge and procedural knowledge (John R. Anderson, 1982). Declarative knowledge is knowledge of factual information, for example Dublin is the capital city of Ireland. Declarative knowledge is also explicit in that a person is aware of what they know. Procedural knowledge is knowledge that may be displayed in behaviour but one is not conscious of it -it is implicit knowledge often connected with how to perform tasks. Procedural knowledge often specifies how to bring declarative knowledge to bear in problem solving. In using a dictionary Peter, like many people, may use declarative knowledge of alphabetical order –simply reciting the alphabet while flicking the pages until arriving at the relevant letter. Others, who regularly use sources that are arranged in alphabetical order, may have procedural knowledge –quite simply they know that P comes later than L and before R and they have no need to recite the alphabet. ACT-R describes declarative knowledge in terms of small primitive units called chunks and procedural knowledge in terms of rule-like units called productions. The cognitive model also includes goal structures similar to Newell and Simon’s goals and sub-goals.
Anderson (1982) relates the transition from declarative to procedural knowledge with the development of a cognitive skill. He proposes three characteristic stages in the development of skills (John R. Anderson, 2000: 310). The first stage is the cognitive stage this is followed by the associative stage and the third stage is the autonomous stage. During the cognitive stage the learner often works from instructions and commonly represents the knowledge verbally –as above when Peter recites the letters of the alphabet. The associative stage indicates a transition from slow and deliberate use of knowledge to a more direct representation of what to do. The example that Anderson gives is learning to use a conventional gearbox for driving –at the associative stage verbalisation of the task is reduced or dropped completely and the actions remain deliberate. The third stage is the autonomous stage –at this stage the skill is increasingly automated and often a person even looses the ability to verbally describe the skill.
Relating ACT Theory to Learning
Anderson and Schunn (2000) discuss the implications of ACT-R learning theory for education. The model allows for the acquisition of declarative knowledge in two ways either in a passive or receptive mode. These are respectively encoding from the environment or storing as a result of mental computations. To go back to Peter’s task of finding the meaning of a new word; he could be told the meaning of the word directly by a person or through a dictionary, this is passive or receptive mode, or he could work it out through deduction and this would be active or constructive mode. Interestingly according to ACT-R theory there is no inherent difference in the memorability of the knowledge generated from the two types of acquisition.
Procedural knowledge is gained through the generation of production rules and enhanced through practice. As practice continues toward a particular skill there is a gradual and systematic improvement in performance that corresponds to a power law. This assertion is an outcome of the ACT-R model and experimental results.
What then are the implications of cognitive theories for Peter’s learning? Overall they describe the processes through which a person can acquire new skills. There are useful insights into the stages of skill development and the power law of improvement through practice. It would seem desirable that learners such as Peter are aware of, indeed well informed of, these insights should they choose to embark on a process to acquire a new set of skills.
Anderson, J. R. (1982). Acquisition of Cognitive Skill. Psychological Review, 89 (4), 369-406.
Anderson, J. R. (2000). Learning and memory : an integrated approach (2nd ed.). New York: Wiley.
Anderson, J. R., Bothell, D., Byrne, M. D., Douglass, S., Lebiere, C., & Qin, Y. (2004). An Integrated Theory of the Mind. Psychological Review 111 (4 ), 1036-1060
Anderson, J. R., & Lebiere, C. (1998). The atomic components of thought. Mahwah, N.J.: Lawrence Erlbaum Associates.
Anderson, J. R., & Schunn, C. D. (2000). Implications of the ACT-R learning theory: No magic bullets. Advances in instructional psychology: Educational design and cognitive science, 5, 1-34.
Gagné, R. M. (1977). The conditions of learning (3d ed.). New York: Holt, Rinehart and Winston.
Hilgard, E. R., & Bower, G. H. (1966). Theories of learning (3d ed.). New York,: Appleton-Century-Crofts.
Leigh, D. (2004). A Brief History of Instructional Design, Performance Improvement Global Network-International Society for Performance Improvement ND (Vol. 20).
Mackintosh, N. J. (1997). Has the Wheel Turned Full Circle? Fifty Years of Learning Theory, 1946-1996. The Quarterly Journal of Experimental Psychology A, 50, 879-898.
Newell, A., & Simon, H. A. (1972). Human problem solving. Englewood Cliffs, N.J.,: Prentice-Hall.
Reiser, R. A. (2001). A History of Instructional Design and Technology: Part II: A History of Instructional Design. Educational Technology Research and Development, 49(2), 57-67.
Tennant, M. (2006). Psychology and adult learning (3rd. ed.). London ; New York: Routledge.
Travers, R. M. W. (1967). Essentials of learning; an overview for students of education (2d ed.). New York,: Macmillan.