Parallel Processing

Parallel processing is a kind of information processing in which multiple activities are carried out simultaneously. The most typical example of this in modern cognitive science is found in artificial neural networks, where it is assumed that more than one processing unit can be active at the same time (Bechtel & Abrahamsen, 2002), hence the term parallel distributed processing or PDP. Parallel processing is often thought to be the opposite of the serial processing that is characteristic of classical information processing. However, classical architectures can be largely parallel -- for instance, all of the productions in a production system scan memory in parallel to determine which production will act next (Newell & Simon, 1972).

In the early chronometric studies of cognitive processing, it was believed that reaction time data could be used to discover the existence of parallel processing. For instance, in memory scanning experiments a latency curve that was flat in terms of the number of items in memory would reveal parallel processing (Sternberg, 1969). However, it was later argued that parallel processing could generate latency curves with positive slopes, provided that parallel processing slowed down as more items were added (Townsend, 1971). Townsend (1990) has gone on to argue that newer approaches are capable of distinguishing parallel from serial processing.

References: