information either from the sensory channels or from other parts of the network. Output

units put out the information. They either control the information directly or send those to

other parts in the network. Hidden units communicate with the input and output units

internally. Most of the important work in a connectionist model occurs in the units and

their connections. Furthermore, there is no central control within the connections, and the

processing is spread across the entire network. In fact, all units have their own activation

value. However, each unit itself is meaningless and not active since the cognitive

processing produces output when the connections between units occur. In particular, to

understand the act of reading as connectionism, we should carefully consider Seidenberg

and McClelland's parallel distributed processing model.

Seidenberg and McClelland's Connectionist Model

 Seidenberg and McClelland (1989) developed a connectionist word recognition

model by describing a parallel distributed processing (PDP) system consisting of four

different processors: context, meaning, orthographic, and phonological. In their PDP

model, they developed a general connectionist framework for thinking about how lexical

knowledge is obtained, represented in the brain, and used in processing (Manis et al.,

1999). Simply stated, their reasoning is that reading involves a series of associations or

connections resulting in accumulated lexical knowledge (Clark & Uhry, 1995). Actually,

the concept of PDP is based on the belief that "intelligence emerges from the interaction

of a large number of simple processing units, each sending excitatory and inhibitory

signals to other units" (McClelland et al., 1986, p. 10). In addition, their PDP model

involves connections in both directions between context and meaning and then additional

two-way connections between meaning, orthography, and phonology (see Figure 2-1).