Chapter 1 – The Science of the Mind

 

The Scope of Cognitive Psychology

Cognitive psychology was first thought of as the scientific study of knowledge: how do we know
what we know? How is knowledge acquired? How is it retained and used (to make decisions or solve
problems)?

Important for everyday:

• Why do you remember some things and forget others? Can we improve our memory in order
to do better at a quiz?
• How much can we control our ability to pay attention and concentrate, and to filter out
unwanted information (e.g. while studying)?
• How do people take decisions?
A huge range of our actions, thoughts and feelings depend on knowledge, and therefore on memory:

Piggy bank story example: many inferences necessary to understand the story rely on our knowledge

• Why does Betsy shake her piggy bank?

• Why does she go to her mother when shaking the piggy bank doesn’t make any sound?
HM examples of consequences of anterograde amnesia:

• Unable to come to terms with his uncle’s death (other than by forgetting it)
• No sense of self (because cannot remember his actions: were they good or bad?)
Knowledge (and memory) is therefore relevant to most our actions, emotions, our social lives, etc.…

The Cognitive Revolution

Cognitive psychology is roughly 60 years old: started in the 1950s-1960s, with an important shift in
approach that has been called the “cognitive revolution”.

This cognitive revolution is based on two ideas:

1. Psychology must study the mental world to understand behaviour

2. The mental world cannot be studied directly

The Limits of Introspection

Wilhelm Wundt (1832-1920) and Edward Titchener (1867-1927): first experimental psychologists.
study of conscious mental events (feelings, thoughts, perceptions, recollections).

No way to experience someone else’s thoughts, so the only way to study one’s mental world is to
introspect, literally “look within”, your own thoughts: observe and record our own mental lives and
sequence of mental experiences.

Introspectors had to be trained: report their sensations and thoughts with a specific vocabulary, with a
minimum of interpretation.

Influential approach for several years, but quickly hit some obvious limitations

• Some (many) thoughts are unconscious: introspection can say nothing about them (by
definition). However, these unconscious mental events play an important role (Example:

  1  
  

what’s your phone number? Apparently instantaneous recall but based on a series of steps
that take place outside of awareness)
• There is a mismatch between our subjective sensations and thoughts and what is actually
the case in the real world (e.g.: we feel that our eyes are moving smoothly, but eye
recordings show that our eyes actually move in small jumps)
• Introspection results are not testable. Since my thoughts can only be observed by myself,
they cannot be independently verified by another observer. Although there are ways
around this to some extent (see study of familiarity, mental imagery, consciousness: in
this case the introspective report is produced by participants, not by the experimenter)
Not all psychologists adopted the introspection technique:

German psychologist Herman Ebbinghaus attempted to study memory using objective methods by
testing his own ability to learn lists of nonsense syllables (e.g. DAK).

American psychologist William James was not impressed with either Wundt or Ebbinghaus’ methods
and preferred to theorize about humans’ everyday experience and emphasize that the human is active
and inquiring.

The Years of Behaviourism

As a reaction to introspection, behaviourists proposed that psychology must only study what is
observable: behaviour/actions and stimuli: both are measurable, recordable physical events.

One can record how stimuli (the environment) changes a person’s behaviour, in other words, how
experience shapes a person’s behaviour.

In contrasts, beliefs, wishes, goals and expectations cannot directly be observed.

Behaviourists attempted to explain all behaviour as a response to external stimuli, as a function of

past experience with these stimuli (e.g.: conditioning).
Behaviorism dominated psychology in the first half of the 20th century. It was responsible for a lot of
important discoveries, in particular, about conditioning.

It also significantly contributed to methodology used in cognitive psychology today: emphasis on the
importance of operational definition, experimental control, etc.…

By the 1950’s however, many psychologists thought that not all behaviour can be explained in this
way: our behaviour to an objective stimulus is also guided by how we interpret the situation. So
different people will react differently to the same stimuli in the same situation if they have different
beliefs, memories, etc.….

Since behaviourist do not take into account these mental variables, because they are unobservable,
they are not able to fully explain behaviour (e.g. to understand the story of the piggy bank, the story
itself is not sufficient: understanding the story depends on memories that are mental states).

Pass the salt example: the same command eliciting the same behaviour can be given in very different
ways. In contrast, if ‘Pass the salt’ is included in some other context, it might elicit a completely
different behaviour. Similarly, just swapping a few letters completely changes the meaning even
though the commands are still quite similar, so physical similarity of stimuli does not lead to similar
behaviour.

Behaviourism has trouble explaining linguistic understanding.

  2  
  

The Roots of Cognitive Revolution

Explaining behaviour therefore seems to require to describe mental states. Accessing mental states
however seems to require introspection, which is difficult to reconcile with scientific objectivity.

Solution suggested by Immanuel Kant’s transcendental method (or inferential method): from
observations, we can try to infer/reconstruct the cause, even if this cause is unobservable/invisible: we
can assume the existence of unobservable variables and see if they explain the observations. This is
how physicists work, when the infer the existence of particles (e.g. electrons) that are not observable.

In the scientific method, we can reproduce the same observations by using the same experimental
situation to obtain more robust results.

Mental processes might be unobservable, but they have observable consequences: one can therefore
observe behaviour in reproducible conditions and try to infer the mental states that lead to this
behaviour: In a typical cognitive psychology experiment, some behaviour is measured (reaction time,
error rate, pattern of error) and hypotheses are made about the mental processes that might have led to
this particular behaviour. These hypotheses can then be tested in variations of the same experiment.

The Path from Behaviorism to the Cognitive Revolution

The inferential logic has been applied to many different behaviours to explain how people remember,
make decisions, pay attention, solve problems, etc.…

Of course, there are many ways of hypothesizing the unobservable processing steps and the goal of
cognitive psychology (as of any science) is to try and select the best hypothesis that explains as many
observations as possible.

The need to postulate mental events was also felt by behaviourist psychologists and some were
involved in the birth of cognitive psychology.

Edward Tolman (1886-1959) proposed that learning was not simply a change in behaviour, but
involved the acquisition of knowledge (a mental, unobservable entity).

He showed this by placing rats in the same maze for 10 consecutive days: every day the rats would
wander and explore the maze randomly. A classical behaviourist would have declared that the rats did
not learn anything because there was no change in behaviour.

On the 11th day however, some food was placed in a certain location of the maze. The next day, the
rats immediately ran to that location, showing that they had perfect knowledge of the maze: this
proves that learning occurred without an observable change in behaviour during the first 10 days: rats
now had an internal map of the maze. To explain the behaviour of the rat on the 12th day, we have to
postulate unobservable knowledge of the maze, which leads to observable behaviour.

In contrast, rats placed in a maze with some food would take several days to learn how to reach this
location.

Chomsky’s rebuttal of Skinner on language: Behaviourist approach cannot account for how language
is acquired: language is too complex to be learned purely by association between stimulus and
response. Instead he emphasized the mental processes needed to understand and produce language
and proposed that some of these processes are innate (contradicting the behaviourist position that
language acquisition can be entirely explained by learning principles).

European Roots of the Cognitive Revolution

Europe was less influenced by behaviourism in the first half of the 20th century and several schools of
experimental psychology during this period influenced cognitive psychology:

  3  
  

Gestalt psychology: behaviours, ideas and perceptions cannot be understood part by part
(analytically), but each element only makes sense within the whole. This led to an emphasis on the
role of the perceiver in interpreting stimuli, and not simply as a passive receiver as in behaviourism
Gestalt: Humans have basic tendencies to organize what they see (perceive). Example of the oval and
two segments perceived as a face: the whole is more than the sum of its parts. They tried to describe
laws that explain how we group different elements into a whole

Gestalt psychologist strongly opposed Wundt’s introspection method because it separated subjective
experience into separate components. Criticized behaviourists for the same reason (behavior analysed
as stimulus-response units).

British psychologist Frederic Bartlett (1886-1969), although not a Gestalt psychologist, also
emphasized how people shape and organize their own experience, with the concept of mental
framework, or schema. He was studying memory and showed that when people have to remember and
recall stories, they make systematic errors that reflect how we transform the story to make it
consistent with our own personal experience

Computers and the Cognitive Revolution

All these schools of thought emphasized the need to postulate/hypothesize internal states or processes
to explain behaviour, but how should these internal processes be theorized?
New approach to these questions was introduced in the 40’s and 50’s with progress in electronic
information processing: newly developed computers were capable of information processing and
storage and people knew how they worked (since they had designed them). Computer scientist even
thought that computers would soon be able to display intelligent behaviour.

Cognitive psychologists began to propose theories/hypotheses (particularly on memory) in terms of

information processing, borrowing from the computer science vocabulary: A given behaviour could
be explained by a sequence of processing steps, or in other words, an algorithm. For instance, in the
Atkinson and Shiffrin model of memory, information is from one storage to another in discrete steps
(sensory memory, short-term/working memory, long-term memory)
This framework is still used today, but others have been proposed: e.g. parallel/connexionist
approach, neuroscientific approaches taking account the brain, etc.…

Research in Cognitive Psychology: The Diversity of Methods

To test a particular hypothesis, cognitive psychologists have to collect more data in different
experiments: they try to derive a new prediction from their working hypothesis and test this prediction
in a new experiment: if the same hypothesis can explain the new data, the hypothesis can be kept and
tested further. If the hypothesis cannot explain the new data, then a new hypothesis must be proposed
that explains both the original data and the new ones.

Hypotheses in Cognitive Psychology can be tested using various methods. Cognitive psychology is
not limited to a particular experimental procedure. What unifies it is the use of the scientific method
to test hypotheses about mental processes.

Types of measurements used:

• How well do people perform in a particular task, e.g. in a memory test, how complete and
how accurate their recollection is. By examining the pattern of errors, one can make
inferences about the mental processes involved
• Response time: mental operations are fast but take a certain time: measuring the time taken to
by mental processes also allows inferences about them

  4  
  

What is also important is in what situation the measurements are taken:

• How do the measurements change when the input (stimuli) are changed: this is what allows
psychologists to make inference on mental processes: establishing relationships between the
input (stimuli) and the output (behaviour) helps characterizing the mental system that
transforms inputs into outputs
• But’s it’s not just the stimuli that can be manipulated: differences in behaviour can be
observed when the context in which the stimuli are presented changes, when a person’s
circumstances or cognitive/emotional state changes, or for different groups of people. All of
these (experimental) manipulations enable us to probe the way the cognitive system processes
information

Response time and performance are the traditional measurements, but there are other types of
measurements used, in particular brain measurements.

Although cognitive psychology is not directly concerned with how the brain implements cognitive
operations, insights from cognitive neuroscience or neuropsychology provide constraints on the
architecture of the cognitive system.

Another discipline that has influenced cognitive psychology a lot is artificial intelligence. In artificial
intelligence, researchers try to mimic human behaviour or performance using computers. They
therefore need to design algorithms that can solve problems and so they need to make hypotheses
about how these problems can be solved using a series of steps.

Pure AI: design the algorithm to accomplish a task as efficiently as possible, even if it ends up
performing it more efficiently. The algorithm might use different strategies than a human would (e.g.
chess playing program).
Computer simulations: design the algorithm so that it mimics human performance, including its
limitations and mistakes: this is the approach that is most influential in cognitive psychology: the goal
is to describe an algorithm that plausibly explains human behaviour by comparing the performance of
the model to human performance

One type of computer simulation that offers an alternative to the serial information processing
approach is the Parallel distributed approach (or connexionist approach): In this approach, the
cognitive system is understood as a network of units (neuron-like) that can perform operations in
parallel.

It is based on the observation from neuroscience that the brain has an architecture that is different
from that of a computer: whereas every operation (information manipulation) occurs serially in a
computer, the network of neurons forming the brain is capable of billions of simultaneous operations
and it is likely that cognitive processing occurs in a distributed manner throughout the brain.

Computers, despite being fundamentally serial, can however simulate parallel neural networks in
order to test distributed processing hypotheses

  5