In the store model, STM and LTM are distinct memory systems, whereas in the working memory view, “STM” is simply the most active information in LTM. Putative differences between coding, retrieval mechanisms, and forgetting processes have lead proponents of the store model to treat STM and LTM as separate systems; however, subsequent research has shown that such differences may not in fact exist. For example, it has been thought that information in LTM was semantically encoded and information in STM was acoustically encoded. Yet, it was found that deaf subjects confuse visually similar letter strings held in STM. This finding suggests that information in STM is not exclusively acoustically encoded.

It was also thought that displacement is a primary source of forgetting in STM, while interference played a critical role in LTM. However, it has been shown that repeatedly asking subjects to store sets of categorically related items in STM results in an apparent decrease in memory capacity: fewer words are recalled accurately, and there are more intrusions from previous trials. This finding suggests that items from previous trials are maintaining activation in LTM and inhibiting recall of words on the current trial. In other words, interference from active information in LTM can play a role in forgetting in STM.

Findings such as these gave rise to the conclusion that LTM and STM do not function as distinct systems. As an alternative to this view, the notion of working memory was proposed. Working memory can be seen as an aggregate of several components which function in concert, including a central executive component, which allocates WM resources to slave systems such as the phonological loop and the visuo-spatial sketchpad.

In sum, WM differs from STM in terms of its hypothesized components, and also in terms of its relation to LTM. The components of WM are treated by some researchers essentially as temporarily highly active forms of information in LTM.