Working Memory
Working Memory
Working Memory
by Saul McLeod
published 2008, updated 2012
Atkinsons and Shiffrins (1968) multi-store model was extremely successful in terms of
the amount of research it generated.
However, as a result of this research, it became apparent that there were a number of
problems with their ideas concerning the characteristics of short-term memory.
Building on this research, Baddeley and Hitch (1974) developed an alternative model
of short-term memory which they called working memory (see fig 1).
Baddeley and Hitch (1974) argue that the picture of short-term memory (STM) provided
by the Multi-Store Model is far too simple. According to the Multi-Store Model, STM
holds limited amounts of information for short periods of time with relatively little
processing. It is a unitary system. This means it is a single system (or store) without any
subsystems. Working Memory is not a unitary store.
Fig 2. The Working Memory Model Components (Baddeley and Hitch, 1974)
The labels given to the components (see fig 2) of the working memory reflect their
function and the type of information they process and manipulate. The phonological
loop is assumed to be responsible for the manipulation of speech based information,
whereas thevisuo-spatial sketch pad is assumed to be responsible for manipulating
visual images. The model proposes that every component of working memory has a
limited capacity, and also that the components are relatively independent of each other.
information is attended to and which parts of the working memory to send that
information to be dealt with.
The central executive decides what working memory pays attention to. For example, two
activities sometimes come into conflict, such as driving a car and talking. Rather than
hitting a cyclist who is wobbling all over the road, it is preferable to stop talking and
concentrate on driving. The central executive directs attention and gives priority to
particular activities.
The central executive is the most versatile and important component of the working
memory system. However, despite its importance in the working-memory model, we
know considerably less about this component than the two subsystems it controls.
Baddeley suggests that the central executive acts more like a system which controls
attentional processes rather than as a memory store. This is unlike the phonological
loop and the visuo-spatial sketch pad, which are specialized storage systems. The
central executive enables the working memory system to selectively attend to some
stimuli and ignore others.
Baddeley (1986) uses the metaphor of a company boss to describe the way in which the
central executive operates. The company boss makes decisions about which issues
deserve attention and which should be ignored. They also select strategies for dealing
with problems, but like any person in the company, the boss can only do a limited
number of things at the same time. The boss of a company will collect information from a
number of different sources.
If we continue applying this metaphor, then we can see the central executive in working
memory integrating (i.e. combining) information from two assistants (the phonological
loop and the visuo-spatial sketch pad) and also drawing on information held in a large
database (long-term memory).
the same time. It is more difficult to perform two visual tasks at the same time because they
interfere with each other and performance is reduced. The same applies to performing two verbal
tasks at the same time. This supports the view that the phonological loop and the sketch pad are
separate systems within working memory.
Humans process information with amazing efficiency and often perform better than highly
sophisticated machines at tasks such as problem solving and critical thinking (Halpern, 2003;
Kuhn, 1999). Yet despite the remarkable capabilities of the human mind, it was not until the
20th century that researchers developed systematic models of memory, cognition, and
thinking. The best articulated and most heavily researched model is the information
processing model (IPM), developed in the early 1950s.
Download Article
The IPM consists of three main components, sensory memory, working memory, and longterm memory (see Figure 1). Sensory and working memory enable people to manage limited
amounts of incoming information during initial processing, whereas long-term memory
serves as a permanent repository for knowledge. In this entry, the information processing
model will be used as a metaphor for successful learning because it is well supported by
research and provides a well-articulated means for describing the main cognitive structures
(i.e., memory systems) and processes (i.e., strategies) in the learning cycle.
SENSORY MEMORY
Sensory memory processes incoming sensory information for very brief periods of time,
usually on the order of 1/2 to 3 seconds. The amount of information held at any given
moment in sensory memory is limited to five to seven discrete elements such as letters of the
alphabet or pictures of human faces. Thus, if a person viewed 10 letters simultaneously for 1
second, it is unlikely that more than five to seven of those letters would be remembered.
The main purpose of sensory memory is to screen incoming stimuli and process only those
stimuli that are most relevant at the present time. For example, drivers on a busy freeway in
heavy traffic are constantly bombarded with visual and auditory stimuli. To maximize
efficiency and safety, they process only information that is relevant to safe driving. Thus, they
would attend to road conditions but not buildings they pass as they drive. Similarly, they
would attend to sounds of other cars, but not to music from the radio or one passenger's
casual conversation with another.
Researchers agree that information processing in sensory memory usually occurs too quickly
for people to consciously control what they attend to. Rather, attention allocation and sensory
processing are fast and unconscious. Information that is relevant to the task at hand, and
information that is familiar and therefore subject to automatic processing, are the most likely
types of information to be processed in sensory memory and forwarded to the working
memory buffer. Information that is highly relevant may receive some degree of controlled,
conscious processing if it is crucial to a task (e.g., attending to salient information such as
animals along the road while driving at high speed). However, controlled processing in
sensory memory would be likely further to reduce the limited amount of information that can
be processed at any given moment.
WORKING MEMORY
After stimuli enter sensory memory, they are either forwarded to working memory or deleted
from the system. Working memory is a term that is used to refer to a multi-component
temporary memory system in which information is assigned meaning, linked to other
information, and essential mental operations such as inferences are performed. A number of
different models of working memory have been proposed (Shah & Miyake, 1999). However,
the three-component model developed by Baddeley (1998, 2001) is the most common, and
will be discussed shortly.
Another key term is automaticity, which refers to being able to perform a task very quickly
and efficiently due to repeated practice (Stanovich, 2003). Automated activities usually
require few cognitive resources; thus, even a complex skill such as driving a car at 75 miles
per hour can seem effortless. Effective information processing in sensory memory requires a
high degree of automaticity with regard to recognition of familiar stimuli such as spoken or
printed words, faces, and sounds.
A third key term is selective processing, which refers to the act of intentionally focusing one's
limited cognitive resources on stimuli that are most relevant to the task at hand. For example,
when driving in snow, one might allocate more of one's limited cognitive resources to
watching the center line in the highway than one would allocate on a clear summer day. In
contrast, on an extremely windy day, one would pay little attention to the whereabouts of the
center line but pay special attention to any flying debris that could cause an accident. In
essence, selective processing enables learners to be optimally efficient by putting all of their
cognitive eggs in one basket. It is no coincidence that highly effective learners succeed
because they identify what is most important to learn and allocate limited attention to relevant
information.
Baddeley's 2001 model of working memory consists of three components, the executive
control system, articulatory loop, and visual-spatial sketch pad. The role of the executive
control system is to select incoming information, determine how to best process that
information, construct meaning through organization and inferences, and subsequently
transfer the processed information to long-term memory or choose to delete that information
from the memory system altogether (e.g., a telephone number that is no longer needed). Most
models of working memory assume that the central executive is the place where humans
make conscious meaning of the information they process (Shah & Miyake, 1999). The role
of the articulatory loop is to maintain and further process verbal information. The role of the
visual-spatial sketch pad is analogous to the articulatory loop in that it maintains and further
processes non-verbal and visual information. Information is lost quickly from working
memory (i.e., 5 to 15 seconds) unless some type of mental rehearsal occurs. Barring rehearsal
(e.g., repeating a telephone number), information is either forwarded to long-term memory or
is deleted from the system.
Baddeley's model makes several critical assumptions about the processing of information in
working memory. One is that each of the three subsystems possesses its own
assumption is that the executive control system regulates the articulatory loop and visualspatial sketch pad.
LONG-TERM MEMORY
Unlike sensory and working memory, long-term memory is not constrained by capacity or
duration of attention limitations. The role of long-term memory is to provide a seemingly
unlimited repository for all the facts and knowledge in memory. Most researchers believe that
long-term memory is capable of holding millions of pieces of information for very long
periods of time (Anderson, 2000). A great deal of research has gone into identifying two key
aspects of long-term memory: (a) what types of information are represented, and (b) how
information is organized. These two questions are addressed in the next section of this entry.
For present purposes, there is universal agreement that qualitatively different types of
information exist in long-term memory and that information must be organized, and therefore
quickly accessible, to be of practical use to learners.
Figure 1 shows that working memory and long-term memory are connected
by encoding andretrieval processes. Encoding refers to a large number of strategies that move
information from temporary store in working memory into long-term memory. Examples
include organization, inference, and elaboration strategies, which will be discussed later.
Retrieval refers to processes that enable individuals to search memory and access information
for active processing in working memory. Both encoding and retrieval greatly facilitate
learning when information in long-term memory is organized for easy access.
A comparison of the three components of the IPM indicates that both sensory and working
memory are relatively short term in nature (see Table 1). Their main roles are to screen
incoming information, assign meaning, and relate individual units of information to other
units. In contrast, the main role of long-term memory is to serve as a highly organized
permanent storage system. Sensory and working memory process few pieces of information
within a short time frame. Automaticity of processing and selective allocation of limited
cognitive resources greatly increases the efficiency of information processing. Long-term
memory is assumed to be more or less permanent and unlimited in terms of capacity. The
main processing constraint on long-term memory is the individual's ability to quickly encode
and retrieve information using an efficient organizational system.
The information processing model provides a conceptual model which explains the different
functions and constraints on human memory. The IPM also has had a major impact on
instructional theory and practice. Sweller and Chandler's 1994 work developed cognitive load
theory to explain how different instructional and learner constraints affect optimal
information processing. The crux of their argument is that each task imposes some degree of
cognitive load, which must be met either by available cognitive resources or learner-based
strategies such as selective attention and automaticity. Reducing cognitive load enables
individuals to learn with less overall mental effort. Cognitive load theory has been especially
helpful in terms of planning instruction and developing learning materials. Others researchers
such as Mayer and Moreno (2003) have developed frameworks to increase learning by
systematically reducing cognitive load through better design of learning materials and more
strategic use of limited resources by students.
In summary, the information processing model postulates a three-component model of
information processing. The IPM is consistent with empirical findings and provides an
excellent framework for understanding principles of effective learning, which are considered
later in this entry. Sensory and working memory are limited with respect to capacity and