In neurological terms, memories are groups of neurons that fire together in the same pattern each time they are activated. The links between individual neurons that bind them into a single "memory" are formed through a process called long-term potentiation (LTP) (Carter, 1998). When we refer to memory in listening, we mean both the process of activating relevant memories to assist in comprehension and the process of forming or updating memories during comprehension.

Memory is generally discussed as involving two dimensions: long-term memory, associated with the sum of all of a person’s knowledge and experience, and short-term memory, associated with knowledge that is activated at a particular moment. Cowan (1993) notes that the popular term "short-term memory" (STM) is ambiguous because it is used to refer to either (1) the set of representations from long-term memory stores that are currently and temporarily in a state of heightened activation, or (2) the focus of attention or content of awareness that can be held for a limited period of time. Cowan argues for a more coherent conception of STM that is hierarchical, with the focus of attention a subset of the activated set of neural connections in long-term memory.

[FIGURE 4A: focus of attention, active memory, long-term memory,

based on Cowan, 1997

Over the past century, research on working memory has been dominated by the construct of memory as a structural entity. Characterizations of short-term memory have emphasized storage, with the role of STM described as specialized for information maintenance for retrieval after a brief interval, such as when we try to retain a new phone number before we dial it (). There has been little emphasis on STM as a means of activating or transforming information or as a means of integrating selected portions of long-term memory with new material.

More recent models have challenged this traditional model of a single short-term "store". For example, newer models posit multiple working memories, modules that are associated with different modalities (e.g. speech vs. writing) and with different kinds of representations (e.g. spatial, serial, verbal). (Cowan, 1993).

Another new proposal is a computational model of working memory. Working memory is seen as a "computation space" in which various operations, such as rehearsal, phonological looping of input, and information reductions, generalizations, and inferences occur. A computational version of working memory still has strict temporal span limitations. Cowan (1997) has discussed two phases of short-term memory with very different properties: (1) a brief sensory unresolved afterimage lasting up to two seconds (sometimes called echoic memory) and (2) a more perceptually resolved short-term memory lasting up to 20 seconds. Under this conception, the second phase of short-term memory, lasting 10-20 seconds, is just one of a series of activated features in memory.

Short-term and long-term memory can be associated with active information and inactive information respectively. For purposes of understanding verbal communication, psychologists now consider it preferable to speak in terms of memory activation rather than in terms of memory size.

Schank (1986) uses script theory as the basis for a dynamic model of memory. His model suggests that events are understood in terms of scripts, plans and other knowledge structures as well as relevant previous personal experiences. An important aspect of dynamic memory are the explanatory processes (XPs) that represent stereotyped answers to events when there is a perceived anomaly in the event. Schank proposes that XPs are a critical mechanism of human creativity.

Script theory is primarily intended to explain language processing and the application of higher-level thinking skills to discourse comprehension. A variety of computer programs have been developed to demonstrate the theory. Schank (1991) applies his theoretical framework to story telling and the development of intelligent tutors. Shank and Fano (1995) describe the application of these ideas to educational software.

Example:

The classic example of Schank's theory is the restaurant script. The script has the following characteristics (scenes, events, and "cognitive moves"):

Scene 1: Entering

S PTRANS S into restaurant, S ATTEND eyes to tables, S MBUILD where to sit, S PTRANS S to table, S MOVE S to sitting position

Scene 2: Ordering

S PTRANS menu to S (menu already on table), S MBUILD choice of food, S MTRANS signal to waiter, waiter PTRANS to table, S MTRANS 'I want food' to waiter, waiter PTRANS to cook

Scene 3: Eating

Cook ATRANS food to waiter, waiter PTRANS food to S, S INGEST food

Scene 4: Exiting

Waiter MOVE write check, waiter PTRANS to S, waiter ATRANS check to S, S ATRANS money to waiter, S PTRANS out of restaurant

There are many variations possible on this general script having to do with different types of restaurants or procedures. For example, the script above assumes that the waiter takes the money; in some restaurants, the check is paid to a cashier. Such variations are opportunities for misunderstandings or incorrect inferences.

Note: {TRANS, ATTEND, BUILD refer to cognitive movements of information}