Social Research Glossary A B C D E F G H I J K L M N O P Q R S T U V W X Y Z Home
Citation reference: Harvey, L., 2012-24, Social Research Glossary, Quality Research International, http://www.qualityresearchinternational.com/socialresearch/
This is a dynamic glossary and the author would welcome any e-mail suggestions for additions or amendments.
|
|
_________________________________________________________________
Connectionism
Connectionisn is a neural network approach to artificial intelligence, natural-language processing, and vision analysis and simulation.
Connectionist approaches are supposedly based on a modelling of the human brain and the more advanced research in this area focused on parallel distributed-processing networks.
Connectionism is underpinned by a view that argues the brain functions by neurons in the brain inhibiting or activating the firing of other neurons. Thoughts and memories are a product (although exactly how is not known) of the activations, their intercommunication and their interaction with the environment.
This view has been around in one form or another since the 1950s although it is far from a definitive theory. Its relevance is also challenged by traditional artificial intelligence approaches that argue that if a function is computable it is computable with a conventional computer (or Turing machine).
The fundamental premise of the connectionist approach to computing is that individual units do not transmit large amounts of data but compute simply by being connected to a large number of similar units.
A typical neural-network model consists of a set of nodes containing a number (its activation) and connections (which are weighted) between nodes. The system is dynamic and mathematical rules take the system from one state to the next. These rules usually operate so that the influence on a given node is constrained by the weights and activations of the nodes directly connected to it.
The difference between this set up and conventional computing is that neural-networks are not programmed, rather they obey laws (as physical systems supposedly do) and simply behave in accordance with these laws or rules. The rules may be simply linear, (i.e., a node has activations based on aggregation or product of associated nodes) or more complex associationist rules that operates to strengthen the connection between two nodes that are highly activated at the same time. This responding to rules is seen as providing a mechanism where intelligence can arise from physical laws.
The connectionist view is that conventional computers have a vast amount of memory lying idle at any one time. Connectionist machines operate in parallel with a large number of processors each with small amounts of memory. In practice connectionist computer hardware involves a machine that has pre-wired hardware connections between elements. It is essential that these are mediated by software connections.
In the Hillis Connectionst Machine, for example, many small processors each containing a small amount of memory operated in parallel. They had a fixed architecture of hardware links although these are augmented by software programmable connections (called routers). A wide variety of network models from meshes (used in image processing) to semantic networks (for knowledge representation) are realisable.
Berkeley (1997) wrote:
Connectionism is a style of modeling based upon networks of interconnected simple processing devices. This style of modeling goes by a number of other names too. Connectionist models are also sometimes referred to as 'Parallel Distributed Processing' (or PDP for short) models or networks. Connectionist systems are also sometimes referred to as 'neural networks' (abbreviated to NNs) or 'artificial neural networks' (abbreviated to ANNs). Although there may be some rhetorical appeal to this neural nomenclature, it is in fact misleading as connectionist networks are commonly significantly dissimilar to neurological systems. ... The basic components of a connectionist system are as follows; (1) A set of processing units (2) A set of modifiable connections between units (3) A learning procedure (optional)
Waskan (2010) wrote:
Connectionism is an approach to the study of human cognition that utilizes mathematical models, known as connectionist networks or artificial neural networks. Often, these come in the form of highly interconnected, neuron-like processing units. There is no sharp dividing line between connectionism and computational neuroscience, but connectionists tend more often to abstract away from the specific details of neural functioning to focus on high-level cognitive processes (for example, recognition, memory, comprehension, grammatical competence and reasoning). During connectionism’s ideological heyday in the late twentieth century, its proponents aimed to replace theoretical appeals to formal rules of inference and sentence-like cognitive representations with appeals to the parallel processing of diffuse patterns of neural activity.
Connectionism was pioneered in the 1940s and had attracted a great deal of attention by the 1960s. However, major flaws in the connectionist modeling techniques were soon revealed, and this led to reduced interest in connectionist research and reduced funding. But in the 1980s connectionism underwent a potent, permanent revival. During the later part of the twentieth century, connectionism would be touted by many as the brain-inspired replacement for the computational artifact-inspired ‘classical’ approach to the study of cognition. Like classicism, connectionism attracted and inspired a major cohort of naturalistic philosophers, and the two broad camps clashed over whether or not connectionism had the wherewithal to resolve central quandaries concerning minds, language, rationality and knowledge. More recently, connectionist techniques and concepts have helped inspire philosophers and scientists who maintain that human and non-human cognition is best explained without positing inner representations of the world. Indeed, connectionist techniques are now very widely embraced, even if few label themselves connectionists anymore. This is an indication of connectionism’s success.
See also
Waskan, J., 2010, 'Connectionism', in the Internet Encyclopedia of Philosophy, available at http://www.iep.utm.edu/connect/, accessed 2 February 2013, still available 1 June 2019.
Top