Computational approaches to learning – The educational perspective
Diana LaurillardInstitute of Education
Interdisciplinary relationships
Learner and
teacher activity
Learner model
Teacher model
Educational research
Technology enhanced learning
Learning theory
Pedagogic theory
Developmental psychology
Models of learning
Models of teaching
ComputationalModelling
Neuro-scientific
models of cognitive activity
Neuroscience
Cognitive
acts
Are there differences between Learning and Development?
Timescale Reversibility Age of onset
Rate Passive or Active?
Domain general
vs. specific
Underlying mechanisms
Type of info processed
Inductive method
LearningQuick
One shotMore
reversible (forgetting)
Later in life
Episodic?
Can be increased through practice
Passive internalisat
ion
Specific to task
Strengthen synapses
within architectures?
Experiences of an
individual
Search hypothesis
space given experience
DevelopmentSlow
Several years
Less reversible
Earlier in life
Cannot be accelerated
Active exploration
Global across
domains
Morphological changes of
neural architectures?
Experiences common
to all members of
species
Alter / enrich hypothesis
space
But learning also includes ‘active exploration’, and works across domains.Certainly reversible, but it’s worse than ‘forgetting’.CM ‘characterises learning as the process of change of structures in the brain’ – but what forms does ‘mis-learning’ take?
Notes from the SEN classroom
Teacher field notesResearcher interpretation of learner model
Researcher interpretation of teacher model
Interdisciplinary relationships
Learner and
teacher activity
Learner model
Teacher model
Educational research
Technology enhanced learning
Learning theory
Pedagogic theory
Developmental psychology
Models of learning
Models of teaching
ComputationalModelling
Neuro-scientific
models of cognitive activity
Neuroscience
Cognitive
acts
Siegler, Karmiloff-Smith, Lave, Resnick,
Engestrom
I tried asking him to count from 45 up. He got stuck at 49 and was unable to continue.
I asked him to go back to 5 and count up. He was then able to get from 9 to 10.
Next he counted across the decades from the 40s to the 70s. He was then able to return to the tape measure and continue up to 160.
Has representations of the number line in terms of units and 10’s, but not integrated as a single sequence
Piaget – construction,
Vygotsky – social practice
Generation of alternative incorrect representations of the number line? – models of not learning?
Recognise nature of misconception?Generate appropriate sequence of sub-tasks?
Facilitate learner constructions.Decompose the task to a component problem.Rebuild the task from the components he has.
Activate the IPS?Activate a dysfunctional network?Build the network across brain areas?
From physical to digital manipulables
Program-defined task for building across 10- learner-specific sequence of tasks- intrinsic feedback on action in relation to goal- no focus on counting- learner model based on accuracy and RTs- teacher model is rule set based on SEN practice, theory- evidence of gradual shift from counting to recognition in response to individualised feedback and sequence
Program-defined task - no learner-specific sequence- random tasks- right/wrong feedback- confusing representations
Interdisciplinary relationships
Learner model
Teacher model
Learning theory
Pedagogic theory
Models of learning
Models of teaching
Which learner activities would be productive for collaboration?Categorisation? Representation, v, s, p? Interpretation? Generalisation?
Which studies yield behavioural data or learning models we can use for diagnosis, and hypotheses we can test?
What kinds of studies would yield data that could generate learning models, and hypotheses CM can test?