Approaches to Psychology
PS1012 & PS1014. John Beech
The Cognitive Approach
to Psychology
Cognitive Approach
Two sections:
Paradigms of the cognitive
The historical development of the
cognitive approach.
What is Cognitive
“It is concerned with thinking and
the mental processes related to
it.” (Glassman, 2000).
Examples: perception, thinking,
memory and language.
Cognition: you can’t fully explain
behaviour by stimulus-response
Some cognitive paradigms and/or
concepts: A preview
Filters (e.g. Broadbent)
Categorisation (e.g. a “poodle” is within the set
of dogs)
Mental set – priming our expectations
The executive – how we control our own actions.
Different memory stores
Modularity – the mind as a number of
independently operating devices.
Some cognitive
Paradigms: 1. filters
This was one of the early concepts. Filters help the
problem of information overload by reducing information.
Broadbent (1958) was especially attracted to the notion.
In his model the filter took out unwanted information prior
to the single perceptual system. This restricted incoming
The filter can act visually, verbally or in other ways.
Visually it can operate to restrict the field of vision. An
emotional filter might operate so that one can block out
violence parts of a film. In word recognition experiments
presenting taboo words, taboo words are slower at being
recognised. This could be due to a filter. On the other
hand it may just be a word frequency effect.
Work in the 1950s demonstrated that it is possible to affect
emotionally the unconscious processing of information.
Lazarus and McCleary (1951) gave certain nonsense
syllables (e.g. vag, mig, jek) paired with a slight electric
shock at the same time. Other syllables in the list did NOT
have a shock.
Later they presented all the syllables (this time without any
shocks) and those previously paired with shocks evoked
galvanic skin responses, whereas the others didn’t. (In
other words their sweat glands registered that they were
emotionally reacting to these particular syllables, but not to
the others.)
Further work by researchers in the 1990s have repeated
these findings using more sophisticated techniques.
Thus people can acquire emotional conditioned
This later work also showed that stimuli can be
presented subliminally and be paired with a mild shock,
and this too can result in an emotionally conditioned
response. Thus one doesn’t even need to be aware of
the stimulus being presented to acquire an emotional
response to it!
Paradigms: 2.
Categories refer to how we group entities
together. For instance, we understand
there are oaks and beeches that are
trees. Trees are plants, etc.
From the time of Aristotle the classical
view has been that categories are
subjectively formulated and subject to
our society and language. These
categories have critical attributes that
specify membership. Finally, the
boundaries between the categories are
not fuzzy.
Paradigms: 2.
The philosopher Ludwig Wittgenstein (1889-1951)
had challenged this classical view and various
experimental studies began to question it as well.
Brown and Lenneberg (1954) examined our
classification of colour. A first group of
participants (from the USA) were given 24 colours
to name. These varied in the ease with which
they could be named. So they were able to rate
the colours in terms of how easy they were to
Paradigms: 2. categorisation
The next group were shown a small set of colours and
then given a recognition test that included the original
set. They had to say which they’d seen before in the
first set. Found: colours that were easier to name
were recognised better. Concluded that being able to
name colours helped recognition.
Paradigms: 2.
However, Eleanor Rosch Heider (1970) studied the Dani in
New Guinea, who were Stone Age people. They had only
2 colour words: mola for bright warm colours and mili for
dark cold ones.
They had to name 40 different colours and then given a
recognition test. When naming they were in agreement
about the mola and mili colours, but not for the other
However, in the recognition part, unexpectedly, they
performed like the American participants making the same
confusions. Thus naming does not appear to affect our
conceptualisation of colour.
Paradigms: 2.
The classicist view had been that as the lines of
demarcation between colours was arbitrary,
different cultures had imposed different
boundaries and these would be internalised by
individuals within their particular cultures.
The results of the Rosch Heider expt suggested
that the physiology of the coding of colour is more
important. In this new view “prototypes” are a
central member of a category, eg. A robin is a
better prototype of “bird” than “stork”. She also
believed that there could be fuzzy boundaries
between categories. Her position came to be
known as the “natural” view of concepts.
Paradigms: 3. priming or
mental set
“Mental set” was an area of interest for a
long period. S given an expectation
about what is going to be seen. For
instance, in one experiment in the 1930s
words presented briefly on
tachistoscope. The words were
misspelled. One group thought they
were getting a list of names of animals,
and the other a list about travel. The
word sael was read as “seal” by the
animal group and as “sail” by the travel
Paradigms: 3. priming or
mental set
example was
Rubin’s vase
which is seen
as two faces or
a vase,
depending on
how it is
Paradigms: 3. priming or
mental set
Cognitive psychology used the term “priming”.
E.g. when naming a list of words, if fruit is
followed by orange, this speeds response to
orange. Giving the semantic category has
helped the exemplar.
Kay and Marcel (1981) presented a mix of
words and non-words. If couch was followed by
louch, louch was pronounced as “lowch”. But if
it was preceded by touch, it was pronounced as
“luch”. The pronunciation of the non-word was
being primed by the previous word.
Paradigms: 4. the executive:
The Baddeley & Hitch (1974)
This is about the
hypothesised decisionmaker in the head.
The executive has had
many roles. E.g. in the
Selfridge model there
was the decision
demon. In Baddeley
and Hitch’s model of
working memory there
are two slave systems,
one an articulatory
(phonological) loop,
the other a visualspatial scratch pad.
But he also has an
executive to coordinate the systems.
Paradigms: 4. the
It has been criticised (e.g. by
Tulving, 1972) as modellers
who use this device appear to
give it human qualities.
Paradigms: 4. the executive: The
Baddeley & Hitch (1974) model
Posner (1978): the
problem of the
“homunculus”, the
miniature person
who takes
decisions. But the
behaviour of the
homunculus must
also be explained.
This provides the
problem of infinite
Paradigms: 5.
Memory stores:
Atkinson & Shiffrin
This was a very powerful model at the time that took over from the
influential Broadbent model. It had 3 stores: a “sensory
memory” (SM) “short-term memory” (STM), a “long-term
memory” (LTM). It also has a “rehearsal buffer” linked to STM.
Information in the working store STS decays within 15-30
seconds. But processing can be put in the rehearsal buffer.
The longer it is held in the rehearsal buffer, the greater the
likelihood of transfer to LTM. STM has 7 slots, each for a
chunk of information. LTM is much more permanent and is
essentially limitless.
Paradigms: 5. Memory stores:
Atkinson & Shiffrin (1986)
Paradigms: 5. Memory
stores - Atkinson &
Shiffrin (1968)
LTM appears to have a number of different locations in
the brain. Eg. A stroke victim lost the ability to name
vegetables, but could still describe each vegetable,
suggesting different locations for names and for
knowledge about vegetables.
Although essentially simple, Atkinson and Shiffrin’s
model became immediately adopted and very
successful. It was probably the most successful
information-processing model of the era. However,
citations to it gradually declined.
Paradigms: 6.
Fodor (1983) argued that the mind is better considered as a
number of separate, mainly independent, devices. These
are specialised for particular kinds of analysis in a fast
manner. Only after this rapid processing will
communication between devices take place.
One implication for modularists such as Fodor is that there
is no need for an executive function.
Gardner (1985), who is sympathetic to modularity sees that
there is a danger of Psychology fractionating into the study
of these different functions without any unifying overviews.
Paradigms/concepts – a summary
1. Filters – we examined this concept by Broadbent, but it can
be applied in different contexts – even emotionally.
2. Categorisation – the classical view is that items are
categorised into distinct sets and that our classification
system is driven by language. But anthropological work
suggests that at least for the perception of colours our
classification of these is not determined by language. We
might use prototypes to conceptualise categories and
there could be fuzzy boundaries between concepts.
3. Mental set – or “priming” in modern parlance. If we are
presented with a stimulus beforehand it can affect the way
we process a second stimulus. This can even affect how
we pronounce a non-word (Kay & Marcel, 1981).
4. The executive – we looked at the homunculus problem.
5. Different memory stores – the Atkinson & Shiffrin (1968)
model has been particularly successful.
6. Modularity – no need for an executive according to Fodor
as the mind is a number of independently operating
Early beginnings: Bartlett.
Background influences: computing,
cybernetics, information theory and
Early part of cognitive revolution:
Miller, Broadbent, Neisser,
The 1970s.
Later developments and criticisms.
Early Beginnings
Some (e.g. Neisser, 1967) regard
Sir Frederick Bartlett as a
forerunner of cognitive psychology.
He was based in Cambridge
Psychological Laboratory.
Before Bartlett the study of memory
had been strongly influenced by
Ebbinghaus, who looked at memory
for nonsense syllables (e.g. guk, lif).
Early Beginnings
But Barlett wanted to be more naturalistic.
Perhaps being well away from the influence
of the American behaviorists, he felt relatively
unfettered in what he could study.
“Remembering” was book he published in
1932 looking at memory for stories and
pictures. Participants had to recall “war of the
ghosts” again and again. It got simplified in
successive recalls.
Early Beginnings
In other expts one person’s recall was given to another
and that was passed on, and so on. This serial
reproduction was like the party game of “Chinese
whispers” and produced the same result.
Found: “condensation” involving the compression of
information leaving out the irrelevant parts; “elaboration”
where we add in extra bits that would be plausible, but
weren’t actually included.
Suggested: we have a “schema (-mata, pl) of what we’re
listening to (ie. we have a mental model or skeleton).
This schema is retrieved when we recall. Irrelevant
details may not be retrieved. Recall is a reconstructive
process, where the details are generated from the
original skeleton, or schema.
Early Beginnings
In an experiment on the concept of schema
Brewer and Treyens (1981) had participants
wait in an academic’s study. Then they were
called into another room and asked to
describe the contents of the room they’d
been waiting in. Several people recalled
there being books in the room, when in fact
the room had had no books in it. Thus they
had reconstructed from their schemata what
the room should have looked like.
Early Beginnings
Problems with such work are to do with
judging detail vs. schema is a matter of
Bartlett also found evidence of emotional
similarities in stories. This resulted in
connections from one story to another.
Details would thus pass from one story
to another.
Early Beginnings
In more recent times, schema is
called “frame of reference”. This is
the set of beliefs about a topic.
When recounting an event, you
incorporate what you believe to be
relevant to the frame of reference.
When reading on a topic, you
incorporate your frame of
reference, into your interpretation
of what you are reading.
Early Beginnings
Experiments in Cognitive
Psychology started mainly after
Initially on “information
processing”. We input information,
process and store it. But there
were several influences.
Background influences to
development of Cognitive
Science: computing
Alan Turing (1912-54) was a British
mathematician who committed
suicide in his early 40s (see the
web). He developed the idea of a
simple machine (later called a
“Turing machine”). This read a
symbol, performed a simple
operation on it and then wrote the
answer back.
Background influences
to development of
Cognitive Science
Symbols can be combined using a set of rules.
This means that they can be used to perform
deductive reasoning.
A Turing machine could potentially perform a
set of procedures to produce the solution to a
problem. These procedures are known as
Turing showed that a binary code (1s and 0s)
could be used to perform in combination very
complex programs.
Background influences
to development of
Cognitive Science
‘Turing machine test’. A person
converses with computer. But
responses via this computer are actually
from a computer or a person. They have
to try to guess which is which. Turing
thought that should be possible to
develop so that the person would be only
at chance level. Influential – human
thought could be mimicked.
John von Neumann: introduced notion of
storage of program within memory.
Cybernetics &
Norbert Weiner, mathematical
prodigy, at MIT in 30s and 40s,
worked on servomechanisms
that guided missiles. Feedback
and self-regularisation.
Give feedback and a machine
adjusts performance. They are
He wrote “Cybernetics” in 1948.
He was interesting because he
linked the human nervous
system with the developing
Neuronal modelling
McCulloch and Pitts (1943): looked
at how nerve cell operates and
connects to other nerve cells. A
nerve cell firing could be viewed as
Given the interconnections
between cells, there could be a set
of propositional statements coded
in the network.
Neuronal modelling
The Turing machine could mimic
networks, or else, produce abstract
programs that describe mental
Hubel and Weisel (1962) later
examined the firing of individual
cells in the visual cortex of cats
demonstrating firing to limited
visual features.
Development of
WW1 & WW2: brain damage due
to shrapnel and missiles.
Development of
Alexander Luria in Russia
worked on brain-injured patients.
Feeling that the time had come
to investigate the mind more
thoroughly. It was time to take a
new direction from behaviorism.
Many were impatient.
Lorenz and Tinbergen observed
animals in natural habitats. (Nobel
prize for both in 1973.)
Developed in 30s and 40s.
The movement led to looking at
more natural settings for humans
as well.
Information theory
Shannon and Weaver developed
this in 40s.
Concerned with coding,
transmission, storage and retrieval
and decoding of information.
Information theory
One aspect of the theory was that text varied in
the surprise value in its information. For
instance, a headline “Prime minister killed” would
generate a shock value because it would be very
unexpected. However, “Prime minister opens
toy factory” would not surprise.
This can be explained in terms of chance. The
odds of a really shocking event are usually much
lower. It is a rare occurrence.
Information theory
In experimental terms this could affect
decision-making. If there is a choice
between 2 alternatives, this is relatively
easy. But if there are 4 alternatives
involved, the chances of any light
occurring is less, compared with when
only 2 alternatives. It has been found
that we tend to slow down when the
event is less expected. Information
theorists proposed that 2 alternatives
involved 1 “bit” of information, 4 = 2 bits
and 8 = 3 bits. Bits were equivalent to
Information theory
Two choices
Four choices
Other important
The “Hixon Symposium” 1948: meeting of a group of
eminent scientists from several disciplines, which proved a
pivotal point.
Important paper by Karl Lashley who challenged the
behaviorists. The behaviorists had advocated study of
only observable behaviour. But this excluded areas such
as problem solving, practical skills, imagery and language.
Other important
Karl Lashley had suggested the
behaviorist theory of associative
stimulus-response chains could not work
for fast motor sequences, as when
playing music. There would not be
enough time for feedback.
Lashley proposed that the nervous
system was always active and arranged
in a hierarchy. The centre of the system
determined control. Lashley’s paper,
which was a substantial challenge to
current thought, was very well received.
The early years of the
cognitive science
Allan Newell and Herbert Simon:
the first proof of a theorem from a
computer program alone.
Naom Chomsky developing his
approaches to grammar and
linguistic transformations.
The early years of the
cognitive science
Miller (1956): important paper on
the human memory and its limited
capacity. Our memory of
phonemes, numbers, words, or
whatever unit seemed to change at
about 7 items, plus or minus 2.
This was the point beyond which
errors occurred.
The early years of the
cognitive science
Miller brought together different studies
and combined in one simple concept. It
also bridged with information theory.
This limit was not limited just to the
auditory system.
Hilary Putnam suggested an interesting
distinction between computer programs,
lists of statements, and the wiring of
computers. Similarly, one can study
neural networks, analogous to wiring.
Patterns of thought would be analogous
to the computer programs.
The early years of the
cognitive science
There was early cognitive work in selective attention such
as by Broadbent and Cherry. In the UK, the influence of the
behaviorists was not so strong.
Early on the emerging discipline was known as “information
processing”. This emphasised link between computer
science and information theory.
Neisser (1967) used the term “cognitive psychology” in his
book. This expanded horizons to areas such as problem
solving. He considered human cognition to be about the
continual analysis and synthesis of information from the
moment light reached the retina.
The early years of the
cognitive science
Neisser’s book was tremendously
influential at this time. He
acknowledged the contributions of
information theorists and computer
scientists, but pointed out their
limitations. E.g. bits were relevant
to telephones, but not to attention
within a human system. Computer
programs (at that time) were too
The early years of the
cognitive science
New journals were created:
Cognitive Psychology (1969),
Cognition (1972) and Memory
and Cognition (1973). Then
textbooks on Cognitive Psychology
followed. Later Applied Cognitive
Psychology, edited in this
department by Graham Davies.
By the 1970s
There were by now a very
substantial number of papers
being published in these new
areas driven by a consensual
view of mechanistic explanations
of cognition.
The Sloan foundation, a NY
based private foundation decided
to invest $15M into the young
By the 1970s: Sternberg
Saul Sternberg’s classic experiments
(1966-9): Participants shown
series of nos., each for 1.2 s. E.g.
given 9 3 5 6 2 and then they were
given probe no. (e.g. 5) and had to
say if present in the set of numbers
in their STM. Each trial gave them
a new set with a different size of 16 items. See fig for results.
By the 1970s: Sternberg
Sternberg found
that RT increased
with set size.
Processing rate
was 38 msec/item.
Also no difference
in slope for the
‘yes’ or ‘no’ items.
Strange result as
scanning is serial
and exhaustive
rather than selfterminating.
By the 1970s: Sternberg
One would expect that when a “Y” required that
when found S would immediately make a
response. If this had happened the slope would
have been half that for the “N” responses. For the
“N” responses, all items would have to be
But this is not what Sternberg got. He could
conclude that scanning STS is serial and that it is
By the 1970s: Sternberg
Similar results found for faces, colours and words with a
different slope for different types of items. Differences in
children vs. adults.
Sternberg proposed: encoding the stimuli, memory search to
find match, response selection (Y or N) and then response
execution. He assumed these stages each took an added
amount of time of the RT.
However, the foundations of such experiments began to
In the case of Sternberg, his experiments can be also be
explained by parallel processing. Variations in the paradigm
produced challenges to his model.
By the 1970s
If an item is repeated (e.g. 9 3 5 3 2),
RTs to it (3 in this case) are shorter
suggesting other influences.
People became increasingly sceptical.
The models were elegant, but did not
stand up to challenge.
Also, this kind of work did not seem to
be “ecologically valid”. There was a drift
away from these types of experiments.
A later development:
Early work in Cognitive Science
had not worried particularly about
the biological origin of the mind.
There were later seen to be limits
in the computer as a serial
A later development:
The connectionist approach (Rumelhart
& McClelland, 1982) that was
subsequently developing suggested
thousands of connections between
hundreds of units working in a parallel
The numbers were limited simply
because it would take far too long to run
proper simulations in proportion to the
no. of connections in the brain. The
simulations are laboriously done on a
computer in serial, but they actually
simulate units as if they have parallel
A later development:
Processing was now about the changing
weightings of connections among the units. It
seems to have been an approach that has
produced new solutions in the areas of perception
and thought.
Some suggest a distinction between the “orthodox
cognitive psychologist” (OCP) and the “cognitive
The OCP is more interested in the intellectual side
and is not concerned with biology.
Criticisms of the cognitive
scientific approach
An important concept is that the mind is
considered to be mechanistic. That is, a
mechanical system that operates by
physical laws.
(Remember the historical concept of
mechanism we examined under
In practice, this means that a
computational system is able to simulate
the mind. It leads to the idea that
rationality and intention is essentially
mechanistic. Many cognitive scientists
believe that this can be the only
conception of mind.
Criticisms of Cognitive Psychology
Costall and Still (1991): Cognition is too restricted
to the individual and to the intellect. The mind is
seen as a series of operations that is worthy of
study. Social aspects are largely ignored. There
is however “cognitive social psychology” which
uses similar tools to the cognitive scientist. This
examines social attitudes and stereotypes.
What seems to be missing from cognitive social
psychology is the concept that social influences
help to form the human mind. The social
constructionalists believe that mind is constructed
foremost from the influences of society. Berger
and Luckmann’s (1967) book was influential in
this area.
Criticisms of the cognitive
scientific approach
In a controversial thought experiment,
Hilary Putnam (1981) imagines a “brain
in a vat”. It is isolated and gets inputs
from a computer connected to its nerve
endings. The thought processes in this
brain are hallucinations even though they
are the same as those of a normal
person using their senses to examine
their environment. Thus a mental state
alone is not enough for seeing and
understanding. We are affected by what
is in our environment.
Criticisms of the cognitive
scientific approach
Meaning is partly social. For instance, I may use
words such as “beech” and “elm”, even though I
may be hazy about what these trees are actually
like. I rely on knowledge of others, particularly
experts, to differentiate these two categories of
tree. I am relying on a linguistic community.
Another criticism of “orthodox cognitive
psychology” is that it focuses on higher mental
functions and has disregarded the body.
Criticisms of the cognitive
scientific approach
This is perhaps historically partly due to the rejection of
behaviorism, which advocated observing the responses
of the body.
Thus in cognitive terms modelling the operation of the
lexicon should be little different from modelling the skills
involved in piano playing. But these perform very
different functions.
But our body is important. Eg. for perception to work we
move our eyes, we grasp and we walk around. We need
innumerable skills that cannot be spelled out in terms of
propositional knowledge. Mind is not locked in the head,
it works in a culturally formed environment.
Criticisms of the cognitive
scientific approach
After the tremendous excitement of the
“cognitive revolution”, there was an
experience of something of a downcycle.
Neisser 10 years later in (1976) was
critical of the computer metaphor in
cognition, as well as the indifference to
cultural influences.
He disliked what he called the
“mechanistic information processing
models”. For instance, in perception he
wanted us to understand how we are
able to see a stable world while we
ourselves moved around it.
Criticisms of the cognitive
scientific approach
Glass, Holyoak and Santa
(1979) state: “Cognitive
Psychology is not getting
anywhere; … in spite of our
sophisticated methodology,
we have not succeeded in
making a substantial
contribution toward the
understanding of the human
mind.” (p. ix).
Criticisms of the cognitive
scientific approach
Gardner (1985, p134) remarks that there was perhaps
excessive excitement about the cognitive revolution. To
him it is perhaps unsurprising that this has given way to a
less ecstatic reaction. He doesn’t criticise experiments
under artificial conditions, but these findings can be
unstable, when the conditions are changed a little.
Gardner’s own observation is that when a new interesting
finding is reported, many experimenters then work hard to
uncover its weaknesses. In the end, the finding is found
wanting in some way. This gives the impression of
advances going backwards rather than forwards.
Gardner would like to see broader, more relevant
problems tackled with the sophisticated methods that
have been developed by cognitive psychologists.
Early beginnings: Bartlett.
Background influences: computing,
cybernetics, information theory and
Early part of cognitive revolution:
Miller, Broadbent, Neisser,
The 1970s.
Later developments and criticisms.