connectionism, history of
 
 
Construed broadly, connectionism maintains that cognitive processes are (implemented in) processes taking place in networks of nerve cells. Thus construed, the history of connectionism spans a wide range of research in numerous disciplines over the course of centuries. See connectionism, symbolicism.
 

Details:
 
A central concern for connectionists is the idea that we must "take the brain seriously" in our psychological theorizing. This idea, of course, dates to ancient Greek speculations concerning the action of animal spirits in the nervous system. Later, notable speculations may be found in Rene Descartes' Treatise of Man (composed circa 1633, first published posthumously in 1662) and David Hartley's Observations on Man (1749).
 
During the 19th Century, histologists developed staining and fixing techniques that enabled them to study the microscopic structure of the brain. With this information, the idea of taking the brain seriously took the form of relating psychological processes to neuronal processes. The existence of the nerve cell came to be widely accepted over the course of the 1890's. At this time, it was widely believed that psychological processes were associative processes. A long tradition of associationism was thus linked to theories of neural networks by the supposition that the biological correlate of the psychological process of association was the formation or strengthening of synaptic connections between neurons. Versions of late 19th Century neural network theorizing may be found in Herbert Spencer's Principles of Psychology, 3rd edition (1872), Theodore Meynert's Psychiatry (1884), William James' Principles of Psychology (1890), and Sigmund Freud's Project for a Scientific Psychology (composed 1895). Neural network theories with an associationist flavor were quite common around the turn of the century, up into the 1930's.
 
Beginning in the 1910's, Karl Spencer Lashley began a series of animal experiments aimed at determining the effects of brain lesions. Lashley supposed that lesions would selectively remove the neurons dedicated to specific tasks, hence destroy an animal's ability to perform those tasks. Rather than complete, selective loss of abilities, Lashley found that degradation of task performance was proportional to the amount of tissue removed. (See, for example, Lashley, 1929.) These results led to a decline in the popularity of neural network theories.
 
During the 1930's, Nicolas Rashevsky proposed to use differential equations and physical concepts, such as energy minimization, to describe how the behavior of nerves and networks of nerves that might be related to psychological processes, such as Pavlovian conditioning. (See, for example, Rashevsky, 1931a, 1931b, 1935). Rashevsky's work was part of a larger project of developing a mathematical biophysics that would mirror the methods of mathematical physics.
 
Rashevsky was instrumental in bringing together Warren S. McCulloch and Walter Pitts, who in 1943 published a seminal contribution to many fields, "A Logical Calculus of Ideas Immanent in Nervous Activity". This work described how networks of binary threshold neurons might be described in terms of sentences of first-order logic. Although popularly remembered for having contributed the idea that networks might carry out logical inferences, McCulloch and Pitts were themselves more interested in the description of networks containing closed loops. This interest stemmed in part from McCulloch's work in tracing neural pathways using strychnine neuronography. Neuroscientific findings by Rafael Lorente de No also increased the interest in closed neural circuits.
 
In 1949, Donald Olding Hebb, a student of Lashley's, proposed a cell assembly theory of cognition. This version of connectionism was meant to circumvent Lashley's problematic lesion results by supposing that the brain contains numerous redundant neural pathways spatially distributed over relatively broad regions of the brain. In roughly the same years, there was considerable neuroscientific investigation of so-called 'post-tetanic potentiation'. It had been found that a large tetanic stimulation of a nerve innervating a muscle would lead to an enhanced effect of the nerve on the muscle. This provided a kind of confirmation of the hypothesis that learning at the psychological level might be realized by facilitation at the biological level. Sir John Eccles, among others, pursued this line of research for over two decades. This research is in some respects a precursor of the current immense interest in so-called "long-term potentiation" and "long-term depression". (See, for example, Baudry & Davis, 1994).
 
During the 1950's and 60's, Frank Rosenblatt investigated the properties of mathematically described neural networks with modifiable connections, discovering the so-called "perceptron convergence procedure" that could train a two-layer network to compute any two-layer-network-computable function. In 1969, Marvin Minsky and Seymour Papert published Perceptrons, a critique of this sort of neural network model. This work showed that two-layer networks were limited in the functions they could compute, thereby providing a major cause for a decline in neural network research during the 1970's.
 
During the late 1970's, Geoffrey Hinton, James McClelland, David Rumelhart, Paul Smolensky, and other members of the "Parallel Distributed Processing Research Group" became interested in neural network theories of cognition. Their landmark Parallel Distributed Processing: Explorations in the Microstructure of Cognition, (1986), marks the return of connectionism as a significant theory of cognition. Where previous strains of connectionism were more or less explicitly linked with associationism or logical inference, the new connectionism in PDP guise has been more cognitively eclectic, suggesting that cognitive processes might be constraint satisfaction processes, energy minimization processes, or pattern recognition processes. It has also been, in large measure, defined, both by its supporters and opponents, in relation to a computational theory of cognition. In this regard, contemporary advocates of connectionism often endorse a "subsymbolic paradigm" in contrast to the "symbolic paradigm" of the computational theory of cognition.
 
 
Ken Aizawa