|
|
We are particularly interested in enhancing theoretical understanding about how the capacity for learning changes from early childhood through to maturity, and related issues of 'readiness' for learning. We use computer-based models to implement our theories of learning and test these against evidence derived from our empirical investigations of learning and instruction. We also use such models to implement and evaluate different theories about how the mind develops with age; theories which promise to explain age-related changes in learning and understanding.
A cognitive model aims to explain aspects of human behaviour, like learning and performance, in terms of a set of specified mechanisms. Computers are a useful tool for this purpose because they force every mechanism of the model to be specified clearly and completely. Also, because the computer is a dynamic tool, it is particularly suitable for testing models of the continuous changes that occur during processes such as human learning.
Our work is guided by the following questions:
- How is an individual's perception of a learning task influenced by their prior knowledge i.e. How are perception and memory linked such that what people perceive depends upon what they already know?
- How does perception relate to task performance?
- How does learning impact on memory, perception and performance?
- What cognitive changes take place as an individual moves along the path from novice to expert?
- Can age-related differences in readiness for learning be explained entirely by changes in knowledge and skill (i.e. the transition from novice and expert) or do some underlying cognitive processes (such as memory capacity) also change with age?
Specific project lines designed to address these questions include the extension and application of well established computational models of learning. These extensions are designed to:
(a) see how well-established models developed to explain adult cognition can also explain knowledge and learning in children, and
(b) provide models of learning which can be exploited in the design and use of computer-based teaching and learning environments.
Work in progress includes:
- language acquisition
- children's learning and understanding of aspects of the physical world, and
- explorations of how such models can be extended to learn from diagrammatic representations, and with multiple representations
We are also examining ways in which this knowledge can be harnessed to design improved learning environments and technologies.
An effective theory of learning explains how, why and under what conditions experience leads to changes in knowledge, skill and understanding. It helps us to understand why purposeful learning in formal settings is sometimes hard to achieve, and in turn, improves our ability to design environments to support the learner in such difficult circumstances.
Research into the processes of learning and tutoring forms a major theme within CREDIT. Two, inter-linking theories lie at the core of this work: scaffolding and contingent tutoring.
Scaffolding refers to the process whereby a tutor provides support for learning as the learner moves towards mastery and autonomy. This work has grown out of a long tradition of work initiated by David Wood and Jerome Bruner.
Contingency theory, developed by David and Heather Wood, started with work on face-to-face tutoring. Contingent tutoring focuses on the nature of tutorial support and the principles which govern the provision of help in response to learner performance over time.
CREDIT research is extending, developing and evaluating work on scaffolding and contingency in connection with the design and evaluation of computer-based tutoring systems. Work on tutoring is also being integrated with CREDIT research on computer-based models of learning and expertise (see, Modelling the Learning Process) and investigations into the development of collaboration, helping and peer tutoring skills in children (see, Collaborative learning and computer Supported Communication).
Implementing computer-based models of contingent tutoring illustrates some of the complexities involved in skilled teaching. Its achievement involves three main sub-tasks:
(a) maintaining a shared focus on problems or tasks which are adapted to the learner's current level of performance (domain contingency)
(b) adjusting the level and content of help to meet learner needs (instructional contingency)
(c) timing tutorial interventions to fit the flow of learner activity (temporal contingency).
Contingent tutoring systems achieve:
- learning opportunities and challenges tailored to individual understanding
- guided practice in the extension of concepts and skills
- graded, on-line help adjusted to individual need
- consistently high levels of collaborative success
These systems assess:
- how much help (if any) a student needs to succeed in achieving new learning goals
- how well individuals exploit help to support their own learning
- how and when autonomy and mastery are achieved
- how students regulate their own learning
Critically, these systems also offer:
- guidance and help to promote improvements in self-regulated learning
- information to teachers identifying learning difficulties being faced by individual learners that may need to be supported by off-line teaching and learning activities.
There are many ways of representing information and communicating about it - in words, images, numbers, diagrams and graphs, for instance. Information, which is hard to understand and to learn in one representational form, may be relatively easy to grasp and learn in another.
One concrete example is the way in which decimal notations, spoken and written terms for fractions and percentages, together with diagrammatic configurations such as pie charts, can each be used to represent an equivalent quantity.
A principled understanding about how, when and why representations exert such differential effects on knowledge acquisition can make an invaluable contribution to learning theory and to the technology for designing learning environments.
For instance, it may be easier to learn how to solve some classes of problems when they are represented diagrammatically but harder to do so when equivalent information is presented as a set of written propositions or expressed as rules. In other circumstances, written notation may prove easier and more reliable to handle than a diagrammatic representation. A central thrust of CREDIT work is to test principles for designing and using diagrammatic representations that can support visualisation, discovery and learning, helping to overcome such barriers to learning and understanding.
Consider, for example, the use of algebra to support conceptual learning in maths and science. Research has shown that, for many learners, the use of such complex notation tends to:
- hide the important meaning of concepts and law
- disconnect abstract ideas from an understanding of concrete examples
- make learners' focus on procedures to manipulate the representations rather than exploring the concepts they serve to represent
- inhibit exploration of more than one interpretation or conjecture
- raise barriers to modelling and problem solving, because such notations are too hard to grasp and use productively.
The use of multiple representations (e.g. pictorial representations used in combination with numerical ones) may offer a way of helping learners to achieve a better grasp of unfamiliar or abstract notations and the concepts they serve to represent. The potential role of such uses of multiple representations is another major focus for CREDIT research.
Here too, the work is motivated by a search for principles to help in the effective design and use of learning environments.
Although such 'multi-media' representations may have the potential to support learning and understanding, they can also create additional difficulties for the learner. Clearly, their use increases the amount that has to be learned - the conventions involved in using each representation have to be learned. Further, learning the connections or mappings between different representations may be just as hard as learning to interpret abstract notations used in isolation. It is therefore vital to explore the various claims about the presumed value of 'multi-media' technologies in support of learning.
To advance our understanding of the interplay between knowledge representation and learning in order to improve the design of educational environments, a number of key research aims are being pursued:
- identifying principles for designing effective and representations to support problem solving and conceptual learning
- investigating how best to support learners as they construct connections across different ways of representing and knowing
- supporting collaborative learning with distributed, multiple representations (see Collaborative learning and computer supported communication)
- extending our computational theories of learning and expertise to model the underlying processes of learning with different but related knowledge representations.
There is ample evidence that collaborative problem solving and peer tutoring can support conceptual learning and the development of communication skills. CREDIT's research is exploring the claim that the acquisition and practice of communication skills through collaboration and tutoring also provides opportunities for learners to develop and practice skills in regulating learning - the learning of peers, and the regulation of one's own learning.
A series of investigations under this theme is designed to extend and evaluate the utility of our model of contingent tutoring as a framework for analysing and understanding the development of helping and peer
tutoring abilities in children aged between 3 and 8 years. We have articulated and illustrated theoretical connections between the acquisition of tutoring abilities and the development of children's 'theory of mind' (see below) and social understanding. We are also exploring the cognitive underpinnings of the ability to help another and competence in theory of mind tasks to look at their relation with the development of skills involved in activities such as planning, explaining and narrating.
One field of application for this work is the design, use and evaluation of ICT to support collaborative learning in both children and adults. ICT makes demands for the development and extension of existing, non computer-mediated communication skills and offers opportunities for the acquisition of new knowledge and practices. CREDIT's work into the impact of ICT on communication and learning in adults is designed to identify the nature of these relations between technology and communication. By integrating such work with our theoretical and empirical studies of the development of skills in peer collaboration, and the impact of communication on learning, we are also developing principles for the design and evaluation of ICT to support learning and educational practice.
Studies of theory of mind assess when the child comes to appreciate the fact that other people may have beliefs about a situation that differs from their own belief, and come to realise that these beliefs influence how and why people act in the way that they do. When this ability develops to the point at which the child can reason about others peoples' beliefs e.g. to entertain such thoughts as "Oh - they think that I think that....", then theory of mind understanding is deemed to move from first- to second-order thinking.
