EPAM Model

EPAM-mechanisms
EPAM mechanisms
(link to source)

Developed by Feigenbaum and Simon, this chunking mechanism, Elementary Perceiver and Memorizer (EPAM), is a computational model that expresses the essences of chunking theory.

The Components of EPAM and The Flow of Information

EPAM consists of the following elements: a limited short-term memory (STM), a discrimination network, and attention mechanisms. The model itself might seem somewhat confusing. The figure and the information below to accompany the figure should clear it up somewhat.

The idea behind the flow of information in the EPAM model is understood through various steps:

  • A stimulus is perceived and then converted into a set of features.
  • The features are then arranged within the discrimination network in such a way that a pointer toward an area in LTM is found.
  • When the features are being arranged in the discrimination network, there is also a process occurring simultaneously in the network that is deciding whether learning needs to occur.
  • Finally, the system takes action or the next external stimulus occurring is retrieved.

It is pretty interesting what occurs in this computational model. Basically, in the distribution network, there is something called a leaf node, which is a part of the distribution network that contains an internal representation of the image of the external stimuli. The leaf node is the essential element that decides whether learning occurs in the system or not. According to Gobet et al., 2001, learning can only occur by comparing the information that the leaf node contains to the information in the stimulus-perfect match, then no learning; subset of stimulus, then additional components are added to the image; no match, then the distribution network is changed in such a way that the leaf node now becomes the internal representation and the stimulus and old image are now used as the foundations for the new leaf node.

What Does The Figure Actually Mean?

A little about the figure to apply it to the explanation of how the EPAM model works: The first part of the figure, part (a) shows an example of a discrimination network. In this part of the figure, the darkened circles represent the leaf nodes, mentioned in the above paragraph. The solid lines in the figure represent the links between the pairs of nodes. The links are test links, meaning that they are being decided whether learning needs to occur or not, so they are testing the links. The test letters are also presented in the figure. The ellipses represent the leaf node images. In part (b) of the figure, the phrase the dog is presented. The presentation of the dog leads to familiarization. Lastly, part (c) of the figure shows that presenting the cat leads to discrimination, which means that extra nodes and links are added to the network. Also, the part in red within the figure is the only part of the figure that shows that alteration has occurred.

It might seem intimidating, but understanding the computational model aspect of chunking theory is very helpful in applying the theory to a network type of structure. Check out Evidence for Chunking Theory to see a direct relationship between the computational model and an application that uses the chunking theory.

(Gobet et al., 2001)

Valid XHTML 1.0! Valid CSS! Level Double-A conformance icon, 
          W3C-WAI Web Content Accessibility Guidelines 1.0 Bobby 508 Compliant! © Copyright Victor Long, Chandra Singh, and David Snitkof 2005
All Rights Reserved