Information Processing

UNIT 4 MODELS OF INFORMATION

PROCESSING
“Information is a source of learning. But unless it is organised, processed, and
available to the right people in a format for decision making, it is a burden, not
a benefit.”
– William Pollard
Structure
4.0 Introduction
4.1 Objectives
4.2 Waugh and Norman’s Model of Primary and Secondary Memory
4.3 Atkinson and Shiffrin’s the Stage Model
4.4 Level of Recall
4.5 Levels of Processing: Craik and Lockhart
4.6 Self Reference Effect
4.7 A Connectionist Model of Memory: Rumelhart and McClelland
4.8 Let Us Sum Up
4.9 Unit End Questions
4.10 Suggested Readings and References

4.0 INTRODUCTION
Cognition as a psychological area of study goes far beyond simply the taking in
and retrieving information. Neisser (1967), one of the most influential researchers
in cognition, defined it as the study of how people encode, structure, store, retrieve,
use or otherwise learn knowledge. The information processing approach to human
cognition remains very popular in the field of psychology.

Information processing is the change (processing) of information in any manner

detectable by an observer. Within the field of cognitive psychology, information
processing is an approach to the goal of understanding human thinking. It arose
in the 1940s and 1950s. The essence of the approach is to see cognition as being
essentially computational in nature, with mind being the software and the brain
being the hardware.

One of the primary areas of cognition studied by researches is memory. By the

1960s research in memory had reached a high state of activity, and it was about
this time that some formalised comprehensive theories of memory were beginning
to be formulated. There are many hypotheses and suggestions as to how this
integration occurs, and many new theories have built upon established beliefs in
this area. Currently, there is widespread consensus on several aspects of
information processing; however, there are many dissentions in reference to
specifics on how the brain actually codes or manipulates information as it is
stored in memory. This section considers a few of the more viable memory theories
of that time.

44
 Models of Infromation
4.1 OBJECTIVES Processing

After reading this unit, you will be able to:

• define information processing approach;
• discussing the various models of information processing;
• explain levels of recall; and
• describe levels of processing.

4.2 WAUGH AND NORMAN’S MODEL OF

PRIMARY AND SECONDARY MEMORY
The first modern behavioural model to travel down memory lane, and one whose
concept of primary memory has served as a departure point for most modern
theories, was developed by Waugh and Norman (1965). The theory is dualistic;
primary memory (PM), a short-term storage system, is conceptualised as being
independent of secondary memory (SM), a longer-term storage system. Waugh
and Norman borrowed freely from William James’s dichotomy of primary and
secondary memory and illustrated their theory by means of the model shown in
Figure below, which encouraged the memory metaphor of boxes in the head that
soon proliferated in the literature of cognitive psychology.

What Waugh and Norman did that James never attempted was to quantify
properties of primary memory. This short-term storage system was taken to have
very limited capacity, so that loss of information from it was postulated to occur
not as a simple function of time but (once the storage capacity was exhausted)
by displacement of old items by new ones. PM could be conceptualised as a
storage compartment much like a vertical file, in which information is stored in
a slot or, if all the slots are filled, displaces an item occupying one of the slots.

Rehearsal

Stimulus Secondary
Primary Memory
Memory

Forgotton

Fig. 1.4.1: Model of Primary and Secondary Memory (Adapted from Waugh and Norman
(1965)

Waugh and Norman traced the fate of items in PM (primary memory) by using
lists of sixteen digits, that were read to subjects at the rate of one digit per second
or four digits per second. The purpose of presenting digits every second or quarter
second was to determine whether forgetting was a function of decay (presumed
to be due to time) or interference in PM.
45
Information Processing If forgetting was a function of decay, then less recall could be expected with the
slower rate (one digit per second); if forgetting was a function of interference in
PM, then no difference in recall could be expected according to the presentation
rate. The same amount of information is presented at both presentation rates,
which, by Waugh and Norman’s logic, allows the same time for decay to occur.
It might be argued that even at one item per second, subjects would allow extra
experimental information to enter their PM, but later experimentation (Norman,
1966) in which presentation rates varied from one to ten digits (for a given period),
yielded data consistent with a rate of forgetting expected from the original model.
The rate of forgetting for the two presentation rates is similar. Interference seems
to be a greater factor than decay in forgetting in PM.
Waugh and Norman’s system makes good sense. PM holds verbal information
and is available for verbatim recall; this is true in our ordinary conversation. We
can recall that last part of a sentence we have just heard with complete accuracy,
even if we were barely paying attention to what was said. However, to recall the
same information sometime later is impossible unless we rehearse it, which makes
it available through SM.

4.3 ATKINSON AND SHIFFRIN’S THE STAGE

MODEL
Traditionally, the most widely used model of information processing is the stage
theory model, based on the work of Atkinson and Shiffrin (1968). The key
elements of this model are that it views learning and memory as discontinuous
and multi-staged. It is hypothesised that as new information is taken in, it is in
some way manipulated before it is stored. The stage theory model, as shown in
Figure 1.4.2, recognises three types or stages of memory: sensory memory, short-
term or working memory, and long-term memory.

Fig. 1.4.2: A stage model of memory (Adapted from Atkinson and Shiffrin 1969)

In the Atkinson-Shiffrin model, memory starts with a sensory input from the
environment. This input is held for a very brief time – several seconds at most –
in a sensory register associated with the sensory channels (vision, hearing, touch,
and so forth). This occurs in as little as ½ second for visual stimuli (Sperling, 1960),
46
and about 4 or 5 seconds for auditory stimuli (Darwin et al., 1972). The transfer Models of Infromation
Processing
of new information quickly to the next stage of processing is of critical importance,
and sensory memory acts as a portal for all information that is to become part of
memory. There are many ways to ensure transfer and many methods for facilitating
that transfer. To this end, attention and automaticity are the two major influences
on sensory memory, and much work has been done to understand the impact of
each on information processing.
Information that is attended to and recognised in the sensory register may be
passed on to second stage of information processing, i.e. short-term memory
(STM) or working memory, where it is held for perhaps 20 or 30 seconds. This
stage is often viewed as active or conscious memory because it is the part of
memory that is being actively processed while new information is being taken
in. Some of the information reaching short-term memory is processed by being
rehearsed – that is, by having attention focused on it, perhaps by being repeated
over and over (maintenance rehearsal), or perhaps by being processed in some
other way that will link it up with other information already stored in memory
(elaborate rehearsal). Generally 5 + 2 number of units can be processed at any
given time in STM.

Information that is rehearsed may then be passed along to long-term memory

(LTM); information not so processed is lost. When items of information are placed
in long-term memory, they are organised into categories, where they may reside
for days, months, years, or for a lifetime. When you remember something, a
representation of the item is withdrawn, or retrieved, from long-term memory.

Organisations of long-term memory - Each of the memory unit or structures

represented in the mind is distinct and serves a different operational function.
However, it is evident that some type of very specialised categorisation system
exists within the human mind. One of the first to make this idea explicit was
Bruner (as cited in Anderson, 1998). “Based upon the idea of categorisation,
Bruner’s theory states ‘To perceive is to categorise, to conceptualise is to
categorise, to learn is to form categories, to make decisions is to categorise’”.

Tulving (1972) was the first to distinguish between episodic and semantic
memory. “Episodic memories are those which give a subject the sense of
remembering the actual situation, or event” (Eliasmith, 2001). Episodic memory’s
store is centered on personal experience and specific events. It is entirely
circumstantial and it is not generally used for the processing of new information
except as a sort of backdrop. Semantic memory, in contrast, deals with general,
abstract information and can be recalled independently of how it was learned. It
is semantic memory that is the central focus of most current study because it
houses the concepts, strategies and other structures that are typically used for
encoding new information. Most researchers now combine these two in a broader
category labeled declarative.

Other researchers have identified additional organisational types. For example,

Abbott lists declarative and procedural while Huitt (2000), citing the work of
Paivio (1971, 1986) adds imagery to this list. However, Pylyshyn (2002) states
that imagery is not a distinct organisational structure, but follows the rules that
apply to semantic and episodic memory. Abbott (2002) and Huitt (2000) define
declarative memory as that which can be talked about or verbalised. It is, then,
47
Information Processing the sum of stored information that can be readily retrieved and put into words in
conscious thought and sharing. As previously stated, declarative memory can be
subdivided into both semantic and episodic memories. Procedural memory can
be thought of as “how to” knowledge (Huitt, 2000). It is the type of long-term
memory sometimes associated with information that has reached a state of
automaticity, but it not limited to this. This type of memory is defined in terms of
learned skills and the ability to recall instruction-like memory. Paivio (1971,
1986) describes imagery as the memory structure for collecting and storing
information related to pictures. It captures information much like a photograph
and can be extremely useful for context and visual presentation of information.

Information Processing in Three Stage Model - Atkinson and Shiffrin make an

important distinction between the concepts of memory and memory stores; they
use the term memory to refer to the data being retained, while store refers to the
structural component that contains the information. Simply indicating how long
an item has been retained does not necessarily reveal where it is located in the
structure of memory.

Information processing from one store to another is largely controlled by the

subject. Information briefly held in the sensory register is scanned by the subject,
and selected information is introduced into the STS. Transfer of information
from the STS was regarded as capable of taking place so long as it was held
there. Atkinson and Shiffrin postulated that information might enter the long-
term store directly from the sensory register.

4.4 LEVEL OF RECALL

In a report by P. I. Zinchenko (1962, 1981), a Russian psychologist, the matter of
how a subject interacts with the material to be learned and committed to memory
was introduced. The basic notion was that words encoded by deep means would
be retained in incidental memory better than if encoded by other, superficial
means. Thus the memorability of words was profoundly influenced by the goal
of the subject at the time the material was presented. Different goals were thought
to activate different systems of connections because subjects have different
orientations toward the material.
The thesis was tested in an experiment in which subjects were given ten series of
four words. The first word was to be connected to one of the other words, but the
instructions varied for each of three groups. An example of a series is HOUSE—
WINDOW—BUILDING—FISH. In the first condition the subjects were asked to
identify the word whose meaning was different from the first word (HOUSE—
FISH). In a second condition subjects were asked to make a concrete connection
between the first word and one of the other words [HOUSE—WINDOW). In the
third condition the subjects were asked to make a “logical” connection between
the first word and one of the other three words [HOUSE—BUILDING].
Zinchenko thought that by altering the instructions the subjects would not only
have different goals toward the material but also be required to examine each
item for meaning. After a brief interrupting task, the subjects were asked to recall
the items. In the condition in which subjects formed logical connections between
the first word and another word, recall of the target word occurred with greater
frequency than the other conditions. Recall of the concrete relationship words
48 was greater than the no-meaning condition.
Thus the level of recall (LOR), as Zinchenko called it, is determined by the goal Models of Infromation
Processing
of an action. In the experiment cited, we can see that when subjects were given
a learning set, or instructions to process material at different levels (to use
contemporary jargon), recall of the material was affected greatly. Because the
original paper was published in Russian and not widely distributed, it has not
been incorporated into the larger framework of memory models. Nevertheless,
as we shall see, the experiment presented by Zinchenko, because of its theoretical
importance to the concept of levels of processing, which has had a profound
influence on cognitive psychology, has important consequences for our
conceptualisation of human memory.

4.5 LEVELS OF PROCESSING: CRAIK AND

LOCKHART
It is likely that progress in the early stages of scientific development is made
more by reaction and counterreaction than by the discovery of great immutable
truths. Craik and Lockhart’s (1972) levels-of-processing (LOP) model, as a
reaction against the boxes-in-the-head scheme of memory, is consistent with
that view. They take the position that data can be better described by a concept of
memory based on levels of processing. The general idea is that incoming stimuli
are subjected to a series of analyses starting with shallow sensory analysis and
proceeding to deeper, more complex, abstract, and semantic analyses.

Whether a stimulus is processed at a shallow or deep stage depends on the nature

of the stimulus and the time available for processing. An item processed at a
deep level is less likely to be forgotten than one processed at a shallow level. At
the earliest level, incoming stimuli are subjected to sensory and featural analyses;
at a deeper level, the item may be recognised by means of pattern recognition
and extraction of meaning; at a still deeper level, it may engage the subject’s
long-term associations.

With deeper processing a greater degree of semantic or cognitive analysis is

undertaken. Consider word recognition, for example. At the preliminary stages,
the visual configuration may be analysed according to such physical or sensory
features as lines and angles. Later stages are concerned with matching the stimuli
with stored information—for example, recognition that one of the letters
corresponds to the pattern identified as A. At the highest level, the recognised
pattern “may trigger associations, images or stories on the basis of the subject’s
past experience with the word” (Craik & Lockhart, 1972).

The significant issue, in Craik and Lockhart’s view, is that we are capable of
perceiving at meaningful levels before we analyse information at a more primitive
level. Thus, levels of processing are more a “spread” of processing, with highly
familiar, meaningful stimuli more likely to be processed at a deeper level than
less meaningful stimuli.

That we can perceive at a deeper level before analysing at a shallow level casts
grave doubts on the original levels-of-processing formulation. Perhaps we are
dealing simply with different types of processing, with the types not following
any constant sequence. If all types are equally accessible to the incoming stimulus,
then the notion of levels could be replaced by a system that drops the notion of
49
Information Processing levels or depth but retains some of Craik and Lockhart’s ideas about rehearsal
and about the formation of memory traces.

A model that is closer to their original idea is shown in Figure 1.4.3. This figure
depicts the memory activation involved in proofreading a passage as contrasted
with that involved in reading the same passage for the gist of the material.
Proofreading, that is, looking at the surface of the passage, involves elaborate
shallow processing and minimal semantic processing.

Reading for gist, that is, trying to get the essential points, involves minimal
shallow processing, or “maintenance rehearsal” (held in memory without
elaboration), but elaborate semantic processing. Another example of this latter
kind of memory activity would be a typist who concentrates on responding to
letter sequences but has very little understanding of the material being typed.

As a result of some studies (Craik & Watkins, 1973; and Lockhart, Craik, &
Jacoby, 1975), the idea that stimuli are always processed through an unvarying
sequence of stages was abandoned, while the general principle that some sensory
processing must precede semantic analysis was retained.

Proofreading Gist

Maintenance rehearsal
Deeper Levels
Elaborat
rehearsal

Fig. 1.4.3: Memory activation in two kind of reading. (Adapted from Solso, 2006)

Levels of Processing versus Information Processing. Information-processing

models of memory have generally stressed structural components (for example,
sensory store, STM, and LTM) dealing with processing (for example, attention,
coding, rehearsal, transformation of information, and forgetting) as operations
that are tied (sometimes uniquely) to the structural components. However, another
approach is to postulate process and then to formulate a memory system in terms
of these operations. Craik and Lockhart have taken just such a position, and their
implicit criticism of the information processing model (along with Neisser, 1976)
suggests that it is falling on hard times.

Where information-processing models of memory stress the sequence of stages

through which information is moved and processed, this alternate viewpoint
argues that memory traces are formed as a by -product of perceptual processing.
Thus, the durability of memory is conceptualised as a function of the depth of
processing. Information that is not given full attention and is analysed only to a
shallow level is soon forgotten; information that is deeply processed—attended
to, fully analysed and enriched by associations or images—is long lasting. The
levels-of-processing model is not free of criticism (see Craik & Tulving, 1975;
and Baddeley, 1978). The criticism includes that (1) it seems to say little more
than that meaningful events are well remembered, a mundane conclusion; (2) it
50
is vague and generally untestable; and (3) it is circular in that any events that are Models of Infromation
Processing
well remembered are designated “deeply processed,” with no objective and
independent index of depth available.

One clear difference between the boxes-in-the-head theory (Waugh and Norman,
and Atkinson and Shiffrin) and the levels-of-processing theory (Craik and
Lockhart) is their respective notions concerning rehearsal. In the former, rehearsal,
or repetition, of information in STM serves the function of transferring it to a
longer-lasting memory store; in the latter, rehearsal is conceptualised as either
maintaining information at one level of analysis or elaborating information by
processing it to a deeper level. The first type, maintenance rehearsal, will not
lead to better retention.

Craik and Tulving (1975) tested the idea that words that are deeply processed
should be recalled better than those that are less so. They did this by having
subjects simply rate words as to their structural, phonemic, or semantic aspects.
Craik and Tulving measured both the time to make a decision and recognition of
the rated words. The data obtained are interpreted as showing that (1) deeper
processing takes longer to accomplish and (2) recognition of encoded words
increases as a function of the level to which they are processed, with those words
engaging semantic aspects better recognised than those engaging only the
phonological or structural aspects. Using slightly different tasks, D’Agostino,
O’Neill, and Paivio (1977); Klein and Saltz (1976); and Schulman (1974) obtained
similar results.

4.6 SELF REFERENCE EFFECT

New light was shed on the levels-of-processing concept when Rogers, Kuiper,
and Kirker [1977) showed that self-reference is a powerful method variable.
Using a method similar to that of Craik and Tulving (1975), they asked subjects
to evaluate a list of forty adjectives on one of four tasks hypothesised to vary in
depth, or semantic richness. Included were structural, phonemic, semantic, and
self-reference tasks.

As in the Craik and Tulving study, it was assumed that words more deeply coded
during rating should be recalled better than those words with shallow coding.
After the subjects rated the words, they were asked to free-recall as many of the
words they had rated as possible. Recall was poorest for words rated structurally
and ascended through those phonemically rated and semantically rated. Self-
reference words were recalled best.

The Narcissistic Trait Modifications of the original experiment have been

conducted in several laboratories with the results being about the same. Some
have argued that self-reference tasks are stored in some special memory system.

Certainly, if you are asked to evaluate a personality trait as being self-descriptive,

such as greedy, loving, or angry, you are engaging a very powerful self-schema,
an organised system of internal attributes that is constellated around the topic of
“I, me, mine.” We also call this the narcissistic trait. Since we all know a great
deal about ourselves (and are emotionally, if not intellectually, deeply invested
in ourselves) we have a rich and elaborate internal network available for storing
self-information. Because of these complex internal self structures we can more
51
Information Processing easily organise new information as it might refer to ourselves than other, more
mundane information (see Bellezza 1992 for several important studies on this
theme). Whether or not these self-rating memories are stored in different parts of
the brain remains a question, but it is a good hunch that plenty of precious brain
space is given over to the narcissistic trait.

4.7 A CONNECTIONIST (PARALLEL DISTRIBUTED

PROCESSING) MODEL OF MEMORY:
RUMELHART AND MCCLELLAND
Many people have been associated with this model of human cognition, but
David Rumelhart and James McClelland have done the most to formalise the
theory.

Essentially, the model is neutrally inspired, concerned with the kind of processing
mechanism that is the human mind. Is it a type of von Neumann computer – a
Johniac – in which information is processed in sequential steps? Alternatively,
might the human mind process information in a massively distributed, mutually
interactive parallel system in which various activities are carried out
simultaneously through excitation and/or inhibition of neural cells? PDPers opt
for latter explanation.

“These [PDP] models assume that information processing takes place through
the interactions of a large number of simple processing elements called units,
each sending excitatory and inhibitory signals to other units” (McClelland,
Rumelhart, & Hinton, 1986). These units may stand for possible guesses about
letters in a string of words or notes on a score. In other situations, the units may
stand for possible goals and actions, such as reading a particular letter or playing
a specific note. Proponents suggest that PDP models are concerned with the
description of the internal structure of larger units of cognitive activity, such as
reading, perceiving, processing sentences, and so on.
The connectionist (or PDP) model attempts to describe memory from the even
finer-grained analysis of processing units, which resemble neurons. Furthermore,
the connectionist model is based on the development of laws that govern the
representation of knowledge in memory. One additional feature of the PDP model
of memory is that it is not just a model of memory; it is also a model for action
and the representation of knowledge.
A fundamental assumption of the PDP model is that mental processes take place
through a system of highly interconnected units, which take on activation values
and communicate with other units. Units are simple processing elements that
stand for possible hypotheses about the nature of things, such as letters in a
display, the rules that govern syntax, and goals or actions (for example, the goal
of typing a letter on a key board or playing a note on the piano). Units can be
compared to atoms, in that both are building blocks for more complete structures
and combine with others of their kind to form larger networks. A neuron in the
brain is a type of unit that combines with other neurons in a parallel processing
mode to form larger systems.
Units are organised into modules, much as atoms are organised into molecules.
The number of units per module range from thousands to millions. Each unit
52
receives information from other modules and, after processing, passes information Models of Infromation
Processing
to other modules. In this model, information is received, is permeated throughout
the model, and leaves traces behind when it has passed through. These traces
change in the strength (sometimes called weight) of the connections between
individual units in the model.

A memory trace, such as a friend’s name, may be distributed over many different
connections. The storage of information (for example, friend’s name) is thought
to be content addressable—that is, we can access the information in memory on
the basis of its attributes. You can recall your friend’s name if I show you a
picture of him, tell you where he lives, or describe what he does. All of these
attributes may be used to access the name in memory. Of course, some cues are
better than others.

Even though the theory is abstract, it touches real-life activities. To continue

with the example of your friend’s name, suppose I ask, “What is the name of the
man you play tennis with?” Such an inquiry gives at least two content-addressable
cues: man and tennis partner. If you play tennis with only one man (and you
know his name), then the answer should be easy. If you have many partners who
are men, then the answer may be impossible.

Additional information (for example, the man with the beard, the left handed
player, the guy with red tennis shorts, the dude with the rocketlike serve, the
chap with the Boston terrier, and so forth) may easily focus the search. You can
imagine how very narrow the search would be if all of these attributes were
associated with only one person: the man you play tennis with has a beard, is
left-handed, wears red tennis shorts, has a hot serve, and has a terrier.

In real life, each of these attributes may be associated with more than one person.
You may know several people who have a hot serve or have a beard. If that is the
case, it is possible to recall names other than the intended one. However, if the
categories are specific and mutually exclusive, retrieval is likely to be accurate.
How can a PDP modular concept of memory keep these interfering components
from running into each other?

According to this model, information is represented in memory in terms of

numerous connections with other units. If an attribute is part of a number of
different memories and is activated (for example. What was your friend’s name
...?), then it will tend to excite all the memories in which the attribute is a part.
One way interfering components are kept from overrunning the system is to
conceptualise the relationship between units as being subject to inhibitory laws.
Thus, when we identify the person you play tennis with as a man, in theory we
inhibit all searches for people who are women. When we add that he has a Boston
terrier, then we do not search for the names of people with whom you do not
play tennis and who do not own a Boston terrier.

Consider the following example of prototype learning, suggested by McClelland

and Rumelhart (1986). A small boy sees many different dogs, each only once
and each with a different name. All the dogs have slightly different features but
can be considered a variation of the prototype dog, the epitome of “dogness.”
The boy forms a memory for a prototypical dog on the basis of experience with
exemplar dogs. As in the case of faces, the boy is likely to recognise the prototype
dog as a dog, even if he has never seen it. Of course, the boy is not likely to 53
Information Processing remember the names of each of the individual dogs, though the most recently
seen dog may still be in memory.

The rationale offered by the connectionist model for prototype formation in the
case of the boy and his (prototype) dog is that each time the boy sees a dog, a
visual pattern of activation is produced over several of the units in the module.
In contrast, the name of the dog produces a reduced pattern of activation. The
combined activation of all exemplar dogs sums to the prototype dog, which may
be the stable memory representation. Thus, the model, more detailed than
presented here, seems to account for this form of memory quite nicely.

The connectionist model of memory has won many disciples in the past few
years. Its popularity is due in part to its elegant mathematical models, its
relationship to neural networks, and its flexibility in accounting for diverse forms
of memories.

4.8 LET US SUM UP

In summary, there are many different theories and models of information
processing that focus on different aspects of perceiving, remembering, and
reasoning. There are many constant themes of information processing regardless
of the specific theory to which one subscribes. Almost all ideas related to how
information becomes stored in memory agree that the learner more deeply and
meaningfully processes information that is presented in a context-rich manner.
One of the most important agreements is that elaboration is a key to permanently
storing information in a way that facilitates its quick retrieval when it is needed.
Most theories hold that the mind contains some type of framework into which
new information is placed. This structure is multi-leveled and has varying degrees
of specificity. New information can be matched with, compared to, contrasted
to, joined with, or modified to fit with existing structures. The formation of and
continual building of these structures, then, is critical in order to process
information. This in-place structural system allows for differing levels of
complexity of information processing.

4.9 UNIT END QUESTIONS

1) Describe the Attkinson-Shiffrin model of information processing in detail.
2) Compare information processing and level of processing models of memory.
3) How does a connectionist (PDP) model handle memory?
4) What is meant by level of recall, level of processing, and self-refrence effect?
5) What are the basic principles and models of information processing?
6) How does organisation in long-term memory take place?
7) Design an experiment to compare the maintenance (shallow) and elaborate
level of information processing

4.10 SUGGESTED READINGS AND REFERENCES

Abbot, B. (2002). Human Memory. Fort Wayne: Indiana University-Purdue
University at Fort Wayne, Psychology Department., from http://users.ipfw.edu/
abbot/120/LongTermMemory.html
54
Atkinson, R., & Shiffrin, R. (1968). Human memory: A proposed system and its Models of Infromation
Processing
control processes. In K Spence & J Spence (Eds.). The Psychology of Learning
and Motivation: Advances in Research and Theory (Vol. 2). New York: Academic
Press.

References

Bransford, J. (1979). Human Cognition: Learning, Understanding, and

Remembering. Belmont, CA: Wadsworth.

Galotti, K.M. (2008). Cognitive Psychology: Perception, Attention, and Memory.

London: Cengage.

Goldstein, E. H. (2008). Cognitive Psychology: Connecting Mind, Research and

Everyday Experience. London: Thomson Learning.

http://chiron.valdosta.edu/whuitt/col/cogsys/infoproc.html

http://www.well.com/user/smalin/miller.html]

Huitt, W. (2000). The Information Processing Approach. Educational Psychology

Interactive. Valdosta, GA: Valdosta State University.

Solso, R.L. (2006). Cognitive Psychology. New Delhi: Pearson Education.

Sternberg, R.J. (2009). Applied Cognitive Psychology: Perceiving, Learning,

and Remembering. London: Cengage.

55