You are on page 1of 11

Complex Neural Networks

A useful model of human learning?

Mark Hardman
Canterbury Christ Church University– mark.hardman@canterbury.ac.uk 1
Complex Neural Networks
 Cilliers (1998) builds on Hebb’s law: continually stimulated
paths between neurons become more efficient.

 Through constant input the system self-organises through


‘neural Darwinism’ (Edelman 1987)
Node

Neuron

2
Artificial Neural Networks
 Artificial networks ‘learn’ to recognise
patterns or objects, or perform tasks.
 Number plates (Draghici, 1997)
 Sales forecasts (Kuo & Xue, 1999)

Learning Mode
 The ‘training mode’ involves repetition of
input and adjustment of path strengths to identify
patterns correctly.
 The system then simply responds to new
input based on it’s internal structure, but also
continues to adapt and thus ‘learn’. Predict Mode
(Johnson & Hogg, 1995)

3
Artificial Neural Networks
Cilliers (1998)
 If the training examples are too
similar, the system can’t deal with
variety of input
 Recognises a specific tree, not all
trees

 Successful systems require


complicated network structure

4
Relevant to Human Learning?
Facts and skills not enough to explain learning:
“Distributed representation” (Cilliers, 1998)
The presented context is important.
Behaviours are also learned.
The environment is important
Social Dynamics
Social
Description?
Description?

Learners respond to stimuli based upon


previous experience: Do our curricula
Repetition is important. allow this?
allow
Variety important to allow flexibility of
How does this relate
How
application. to behaviourist or
History of the ‘system’ is important. constructivist
positions?

5
Does Neuroscience Support a Complex
Networks Model?
Experimental Limitation
 Impossible to see neural development inside live mammal.

Support for model


 Hebb’s law supported by research in Aplysia
(Antonov et al, 2003).
 Dendrites allow more neurotransmitter (Sylwester, 1995)
 Also a mechanism for reduction in efficiency of paths.

 Lamprey neural networks suggest development of


paths according to usage (Kelso, 2000)

 Fits with understanding of cognitive memory formation.


 Episodal  semantic  procedural (Sylwester, 1995)
6
Does Neuroscience Support a Complex
Networks Model?
Limitations of model
 Neuroscience focuses on brain ‘modules’, which are involved
in specific tasks: brain function also determined by genetics and
biology.
 First and second repertoire (Cilliers, 1998; Edelman, 1987; Changeaux, 1984)
 Gating signals used by brain to ‘authorise’ plasticity (von der Malsberg, 1995)

 Non-Hebbian learning is important.


 Learning from ‘life changing events’ (Geake, 2009)
 Influence of motor action on brain (Edelman, 1987)
 Our ability to copy behaviour.

7
Large-scale Brain Networks
Bressier & Menon (2010)
Regions of brain with similar properties classified as nodes.
Interaction of nodes investigated.
Core networks identified.
Still very simplistic picture.

Paradigm shift to considering “conjoint function of brain areas


working together as large-scale networks.”

8
A useful model of human learning?
 Neuroscience not advanced enough to conclusively support
or refute neural network model.
 Model is certainly simplistic and does not account for
evolutionary history of humans.

Further work is needed to support:


1. Learning must be considered beyond facts and skills.

Model of learning in a complex classroom?

2.Repetition of activities must be balanced with varied


context to create flexibility of application.
Testable with social scientific methods
9
References (1 of 2)
Antonov I., Antonova I., Kandel E.R. & Hawkins R.D. (2003) Activity-dependent
presynaptic facilitation and hebbian LTP are both required and interact during
classical conditioning in Aplysia. Neuron. 2003 Jan 9;37(1):135-47.
Bressler, S. L. & Menon, V. (2010) Large-scale brain networks in cognition:
emerging methods and principles. Trends in Cognitive Sciences, 14:277-290
Bullmore, E. & Sporns, O. (2009) Complex brain networks: graph theoretical
analysis of structural and functional systems. Nature, Vol. 10, March, pp186-195
Cilliers, P. (1998) Complexity and postmodernism: understanding complex
systems, London: Routledge.Feeman, J. (1999) How Brains Make Up Their Minds.
London: Weidenfeld & Nicolson
Changeaux, J.-P., Heidemann, T. And Patte, P. (1984) Learning by Selection in
Marler, P. & Terrace (eds), The Biology of Learning. Berlin: Springer-Verlag. [Cited in
Cilliers (1998)]
Draghici, S. (1997) A neural network based artificial vision system for licence plate
recognition. International Journal of Neural Systems, February, 8(1):113-126
Edelman, G.E. (1987) Neural Darwinism: The Theory of Neuronal Group Selection.
New York: Basic Books 10
References (2 of 2)
Geake, J. G. (2009) The Brain at School: Educational Neuroscience in the
Classroom. Open University Press
Hebb, D.O. (1949), The organization of behaviour, New York: Wiley

Johnson, N. & Hogg, D. (1995) Learning the Distribution of Object Trajectories for
Event Recognition. Proc. British Machine Vision Conference, volume 2, pages 583-592
[accessed June 2010 at www.leeds.ac.uk]
Kelso, J.A.S (1995) Dynamic Patterns: The self-Organisation of Brain and
Behaviour. Cambridge, Massachusetts: MIT Press
Kuo, R. J. & Xue, K. C. (1999) Fuzzy neural networks with application to sales
forecasting. Fuzzy Sets and Systems 108:123-143
Sylwester, R. (1995). A celebration of neurons: an educator's guide to the brain.
Tuscon, AZ: Zephyr Press.

Mark Hardman
Canterbury Christ Church University– mark.hardman@canterbury.ac.uk11

You might also like