Connexionnism

The connexionnism is an approach used in cognitive Sciences, Neurosciences, Psychologie and Philosophie of the spirit. The connexionnism models the mental or behavioral phenomena like emergent processes of networks of inter-connected simple units . Generally the connexionnists model these phenomena using Réseaux of neurons.

Basic principles

The basic principle of the connexionnism is that the mental phenomena can be described using simple networks of inter-connected units. The form of connections and the units can vary according to the models. For example, the units of a network can represent Neuron S and connections can represent Synapse S. Another type of model could make that each unit of the network is a word and that each connection is an indicator of the similarity Sémantique.

The diffusion of activation

An important concept for the models connexionnists is that of activation . At every moment, a unit of the network has a certain level of activation. For example, if the units of the model are neurons, activation could represent the probability that the neuron generates a Potentiel action. If this model is a model with diffusion of activation , then once the threshold of activation reached, activation will be propagated with all the connected units.

Neural networks

See the principal article: Network of neurons

The networks of neurons are, by far, the models the most widespread connexionnists today. Much research using of the networks of neurons claims connexionnism. The networks connexionnists answer two characteristics:

Each state mental can be represented like a vector with N dimensions representing the values of activation of the neuronal units;
the network can learn by modifying the weight from connections between its units. The force of connections (or weight) is generally represented like a matrix with N dimensions.

On the other hand, the models can differ from:

the definition of activation : Activation can be defined various ways. For example, in a Machine of Boltzmann, activation is interpreted like the probability of generating a potential of action. This probability is determined by a logistic function depending on the sum of the entries of a unit;
the algorithm of training : It defines the way in which the weights of connections will change time progressively.

Biological realism

Training

History

The ideas which gave rise to the connexionnism go back to more than one century. Nevertheless, the ideas connexionnists were nothing more than of simple speculations until the middle of the 20th century. It is necessary to wait the Eighties so that the connexionnism becomes a popular prospect near the scientists.

Distributed Parallel treatment ( Parallel distributed processing )

Princeps work

Connectionism apart from PDP

See too

Network of neurons
Behaviorism
Network of biological neurons
Cybernetic
Eliminative materialism
computational Neurosciences

References

Rumelhart, D.E., J.L. McClelland and the PDP Research Group (1986). Parallel Distributed Processing: Explorations in the Microstructure off Cognition. Volume 1: Foundations , Cambridge, MY: MIT Close
McClelland, J.L., D.E. Rumelhart and the PDP Research Group (1986). Parallel Distributed Processing: Explorations in the Microstructure off Cognition. Volume 2: Psychological and Biological Models , Cambridge, MY: MIT Close
Pinker, Steven and Mehler, Jacques (1988). Connections and Symbols , Cambridge MY: MIT Near.
Jeffrey L. Elman, Elizabeth A. Bates, Mark H. Johnson, Annette Karmiloff-Smith, Domenico Parisi, Kim Plunkett (1996). Rethinking Innateness: With connectionist perspective one development , Cambridge MY: MIT Near.
Marcus, Gary F. (2000). The Algebraic Mind: Integrating Connectionism and Cognitive Science (Learning, Development, and Conceptual Change) , Cambridge, MY: MIT Close

External bonds

has off Brief History Connectionism
Dictionary off Philosophy off Mind entry one connectionism
Stanford Encyclopedia off Philosophy entry one connectionism

Random links: