Cognitive Psychology: Chapter 9- Concepts and Generic Knowledge
• We need concepts in order to have knowledge, and we need knowledge in order to
• Wittgenstein (1953) 20 century philosopher, that that philosophers had been trying
for thousands of years to define terms like “virtue” and “knowledge”
• Definitions often have exceptions (most dogs are four-legged, but can be three-legged)
• Even simple terms denoting concepts that we use easily and often resist to being defined
• We need a way of identifying concepts that highlights what the various members of a
category have in common (eg. What all dogs have in common) while simultaneously
allowing exceptions to the proposed rule.
• We can do so by keeping the context of our definitions but being more flexible to our
use of the definitions
• For instance; a dog is an animal that probably has fur, 4 legs, and barks
• Wittgenstein proposed that members of a category have a family resemblance to
• there are features that are common in the family, and so, if we consider family
members, 2 or even 3 at a time, we can fine shared attributes
• there may be no features that are shared by all dogs, or all games, just as there are no
features that are shared by everyone in your family
• a rigid definition is not possible
• we can identify “characteristic features” for each category; features that most
category members have
• the more of these features an object has, the more likely we are to believe it is in the
• family resemblance is a matter of degree, not all-or-none
Prototypes and Typicality Effects
• definitions set the “boundaries” for a category • if a test case has certain attributes, then it is inside the category
• Prototype theory: perhaps the best way to identify a category or to characterize a
concept is to specify the “center” of a category rather than the boundaries
• Example: prototype dog= the ideal dog
• Therefore, all judgments about dogs are made with reference to this ideal
• In some cases, a prototype may literally represent the ideal for the category : example, the
prototype of diet soda might have 0 calories but still taste great.
• More commonly, a prototype will be an average of the various category members you
have encountered. Example: the average color of the dogs you have seen, the average size
of the dogs you have seen, etc.
• Different people have different prototypes; people may disagree about what the ideal for
a category is
• Also cultural differences
• Despite this, the prototype serves as an anchor, or benchmark for our conceptual
• When we reason about a concept or use our conceptual knowledge, our reasoning is
done with reference to the prototype
Fuzzy Boundaries and Graded Membership
• What it means to “know” a concept is simply to have some mental representation of the
• Things that have fewer attributes in common with the prototype will probably cause you
uncertainty about their identity
• Since the category is characterized by its center (the prototype) and not by its
boundaries, there is no way we can say whether something is inside of the category or
• (to be inside or outside, you need a definite boundary to be inside or outside of)
• Each category has what is called a fuzzy boundary: with no clear specification of
category membership and non-membership • Objects closer to the prototype are, in effect, “better” members of the category than
objects farther from the prototype
• Thus, categories that depend on a prototype have graded membership.
• Graded membership: the idea that some members of a category are “better” members
and therefore are more firmly in the category than other members
Testing the Prototype Notion
• Sentence verification task: research participants are presented with a succession of
sentences. Their job is to indicate (by pressing the appropriate button) whether each
sentence is true or false)
• In most experiments, we are interested in how quickly participants can do this task, and
in fact their speed depends on several factors.
• Response speed depends on the number of “Steps” the participants must traverse to
confirm the sentence
• Participants also response more quickly to true sentences than for false, and also more
quickly for familiar categories.
• According to a prototype perspective, participants make judgments by comparing the
thing mentioned to their prototype for that category
• When there is much similarity between the test case and the prototype, participants can
make decisions quickly, judgments about items more distant from the prototype take
• Production task: ask people to name as many birds or dogs as they can.
• According to a prototype view; they will do this production task by first locating their
bird or dog prototype in memoery and then asking themselves what resembles this
• Start with the center of the category (prototype) work their way outward
• Therefore, birds closest to the prototype should be mentioned first, birds farther from the
prototype, later on. • The first birds mentioned in the production task should be the birds that yielded the
fastest response times in the verification task because what matters in both tasks is
proximity to the prototype.
• Members of a category that are “privileged” on one task (eg. Yield the fastest response
times) turns out to also be privileged on other tasks (eg.Are most likely to be mentioned).
• Various tasks converge in the sense that each task yields the same answer/ indicates the
same category members as special.
• Category members mentioned early in a production task (robin bird.Applefruit) are
also “privileged” in a picture-identification task
• Picture-identification task: a task in which a person is shown simple pictures (often line
drawings) and must indicate, as rapidly as possible, what the picture shows (e.g. “a dog”,
“a truck”) Responses tend to be faster if the objects shown are typical of their category.
• Rating task: Atask in which research participants must evaluate some item or category
with reference to some dimension, usually expressing their response in terms of some
number. For example, participants must be asked to evaluate birds for how typical they
are within the category of birds, using a 1 response to indicate “very typical” and a 7
response to individual “very atypical”
• Typicality: The degree to which a particular case (an object, situation, event) is typical
for its kind
• Rosch and others have argued that there is a “natural” level of categorization, neither
too specific nor too general, that we tend to use in our conversations & reasoning
• The special status of this basic-level categorization can be demonstrated in many
• Basic-level categorization: represented in our language via a single word (ex. chair)
• Specific categories: identified via a phrase (lawn chair, kitchen chair, etc)
• The importance of basic level categories also shows up in our memory errors • In one study, participants read a story, and then after a delay their memory for the
story is tested. If the story contained specific terms, participants often (falsely)
recalled that they had heard something more general
• “she noticed that her jeans were stained” remembered as “she noticed that her pants
• If the story contained general terms these were misremembered as being more
specific than they actually were. (ex. participants remembered hearing about dogs
when they actually heard about animals)
• The errors almost always tend to “revise” the story in the direction of basic-level
• Exemplar-based reasoning: reasoning that draws on knowledge about specific category
members, rather than drawing on more general information about the overall category (a
specific remembered instance)
Analogies from remembered exemplars
• In some cases categorization can draw on knowledge about specific category members
rather than on more general information about the overall category
• Example: categorization is supported by memories of a specific chair, rather than
remembered knowledge about chairs in general
• The exemplar-based approach is in many ways similar to the prototype view
• You categorize objects by comparing them to a mentally represented “standard”
• The difference between the views lies in what that standard is
• For prototype theory: the standard is the prototype; an average representing the entire
• For exemplar theory: the standard is provided by whatever example of the category
comes to mind
Explaining typicality data with an exemplar model
• An exemplar-based approach can also explain the graded-membership pattern • If you frequently encounter something the memory of it becomes well primedfaster
memory searchpattern of what is more readily available in your memory
Exemplars preserve information about Variability
• Exemplar-based views can easily explain typicality effects
• Drawback of prototypes is that they are averages; much information is lost in an
• An average does not tell you how variable a set of data is
• Aset of exemplars preserves information about variability; a prototype does not
Acombination of Exemplars and prototypes
• Different settings, or different perspectives would trigger different memories and bring
different exemplars to mind
• Some evidence favours the idea that part of our conceptual knowledge involves
• Must not lose sight of the fact that prototypes also have advantages
• Prototypes: represent in an efficient and economical manner what is typical for a category
• Our conceptual knowledge includes BOTH prototypes and exemplars
• An object triggers some information in memory (either a specific instance, according to
exemplar theory, or the prototype, according to prototype theory)
• In both cases, you assess the resemblance between the conceptual knowledge, supplied
by memory, and the novel object
• Triggering of memoryjudgment of resemblance conclusion based on resemblance
Odd number, even number
• Judgments of typicality and judgments of category membership both derive from the
same source: resemblance to an exemplar or to a prototype • If the resemblance is great, then a test case will be judged to be typical, and it will also be
judged as a category member
• If the resemblance is small, then the test case will be judged atypical and probably not a
• However, we can also find situations in which there is no relation between typicality and
• Study by Armstrong, Gleitman, and Glaitman (1983) asked participants to do several
of the concept tasks ( sentence verification, making explicit judgments about whether
individuals were typical or not of the category)
• Researchers used categories with a clear definition—for example odd numbers
• Asked to rate each of the numbers as to how good of an example it is for the category
• Still judged some to be more typical than others, even though they are all odd
• So there is NOT always a link between typicality and membership
Lemons and Counterfeits
• Categories show the same pattern (echo the message)
o In each case, people find it easy and “natural” to decide whether a particular object is
“typical” for the category or not, but then the same people judge category member- ship
in a way that’s entirely independent of typicality.
As a result, a test case can be a category member without being typical or can be
typical without being a category member.
o Example 1: It is true that robins strike us as being closer to the typical bird than penguins
do. Nonetheless, most of us are quite certain that both robins and penguins are birds
o Example 2: Moby-Dick was definitely not a typical whale, but he certainly was a whale
Observations are easily confirmed in the laboratory For example, research
participants judge whales to be more typical of the concept “fish” than sea
lampreys are, but the same participants respond (correctly) that sea lampreys are
fish and whales are not o It seems that category judgments are often not based on typicality but rather defined in
ways that refer to “deep properties,” not to mere appearances.
Consider a painted, sugar injected and flattened lemon. Is it still a lemon after all
that abuse? Apparently it is. (No matter how different the lemon is from its
prototype, it will still remain a lemon).
However, the near identical prototypes do not work with money because
“counterfeit” money is not real money.
• Keil’s Studies Keil studied how preschool children build concepts and what concepts they do
or do not include (these kids do not understand the technical meaning of the concepts).
o In one study, kids were asked what makes something a “coffeepot,” a “raccoon,” and etc.
as a way of examining their concepts. They were also asked whether it would be possible
to turn a toaster into a coffeepot, to which many children said yes.
o On the flip side, children denied that a raccoon could be transformed into a skunk. The
children understood that the skunk’s fur could be dyed, and it could be taught to behave
like a raccoon but at the end of the day, it will still be a skunk.
This suggests that we reason differently about naturally occurring things like
raccoons and manufactured products like coffeepots (deep properties matter).
The Complexity of Similarity
• Both the prototype and the exemplar views depend on judgments of resemblance
o Cases like the damaged lemon and the counterfeit bill, suggest that resemblance may not
be the key here. The damaged lemon does not resemble the lemon prototype, but it is still
a lemon. The counterfeit bill does resemble the prototype for real money, but it still isn’t
legal. These simple observations suggest that we cannot base category membership on
• The resemblance notion can be preserved: An abused lemon may not resemble the lemon
prototype perceptually, but it does resemble the lemon prototype in terms of its DNApattern.
o If these essential properties are focused on, and the superficial attributes are ignored, we
could maintain the claim that category membership depends on properties shared with the
prototype or shared with exemplars. In this way, we could preserve the claim that
category membership depends on resemblance.
• How do we decide which features to ignore when assessing resemblance and which features to
o Judge by categorizing
Concepts as Theories
• Theory of concept needs elements other than prototypes and exemplars in order to guide our beliefs and knowledge, of our use of prototypes.
• Heuristic strategy one that gives up the guarantee of accuracy in order to gain some efficiency.
Example: which would you prefer: to be right all the time but needing 18 hours for every
judgment you make, or to be right most of the time but able to make your judgments instantly?
o Categorization via resemblance is a heuristic strategy, an efficient way to think about
Theories and Category Coherence
• Aperson’s implicit “theories” about concepts influence him in many ways—and govern, for
example, how easily or how quickly he can learn new concepts.
o Imagine sorting items into categoryAand B. CategoryAincludes all regular surfaced
metal objects, while B includes objects made of metal, with irregular surfaces. This
sorting task would be difficult unless you are given another piece of information: namely,
that categoryAincludes objects that could serve as substitutes for a hammer. With this
clue, you can immediately draw on your other knowledge about hammers, and in this
way you can see why the features are as they are.
The Use of Theories in Categorization
• Categorizing things based on your theory. For example, we have certain beliefs about how drunks
behave; we have, in essence, a “theory” of drunkenness.
• The proposal that people will reason differently about naturally occurring things and artifacts
because they have different beliefs about why categories of either sort are as they are.
o In general, people tend to assume more homogeneity when reasoning about biological
kinds than when reasoning about artifacts
o Researchers have used fMRI scans to examine what parts of the brain are activated when
people are drawing on one aspect or another of their conceptual knowledge.
These neuroimagery data confirm the differences among types of concepts, with
different brain areas activated when people are thinking about living things and
when they are thinking about nonliving things
Concepts: Putting the Pieces Together
• There is likely to be a prototype for each concept, as well as a set of remembered exemplars that
people use for a range of fast and easy judgments about the relevant category
• Much of people’s knowledge reflects on their understanding of cause-and- effect relationships,
which influences how they categorize items and also how they reason about the objects in their world.
-Galton asked various people to describe their mental images and rate them for
vividness, this used introspection and self-report data.
-Participants reported that they could 'inspect their images much as they would inspect
a picture. Scenes were represented as if viewed from a certain position and certain
distance. They also reported being able to identify colours and textures.
-However, the participants varied greatly.
-Many described images with photographic clarity, while others reported very sketchy
images or none at all. They could think about the objects but not visualize them.
-Methodological concern: Perhaps all of Galton's participants had the same imagery skill
but some were cautious in how they chose to describe their imagery while others were
-This data might reveal differences in how people talk about their imagery rather than
differences in the imagery per se.
- Because of the previous results in self-report, researches no longer ask participants to
-To gain objective data, these experiments ask people to do something with their images
so we can examine how accurately and quickly people respond. We can use this to
testing the nature of imagery.
-These chronometric (Time-measuring) studies give us much more accurate examples
of mental imagery then self-report data.
-These studies allow us to ask what sorts of information are prominent in a mental
image and what is not.
- When asking people to describe an object, features that are distinctive to that object
will be more prominent, where as when asking people to depict an object, prominence
lies with the parts of the object that are larger or positioned in the forefront. (Ex.
"Describe a cat." Claws and whiskers are prominent because they are distinctive to cat.
"Draw a Cat". Head is more prominent because it will be first thing drawn, and large.)
-Similar effects are shown when people are asked questions about the cat. When
specifically asked to create a mental image of a cat questions such as "Does the cat
have a head" are answered more quickly than "Does the cat have claws". Whereas,
when participants were asked to think about a cat, with no mention of imagery, "Does the cat have claws" would be answered more quickly than "does the cat have a head".
-When the mode of representation changes, so does the pattern of information
-In a different experiment, participants were asked to memorize a fictional map and to
memorize the location of various landmarks. They would then be asked to form an
image of the map in their 'mind's eye'. They would be asked to imagine one of the
landmarks. Another landmark was mentioned, and they were asked to draw a line from
the first land mark to the second and hit a button once they were done (stopping a timer
which measured the 'scanning times'). They continued this using different landmarks.
-Results showed that the image scanning procedure was consistent. If the space
between the two landmarks doubled, so did the scanning times, and so on.
-The same result was found when asking participants to zoom in or out of an image. the
zoom time was correlated with the amount at which they had to zoom in/out
-All of this collected data lined up with the results of asking participants to complete the
tasks with real images.
-This tells us a great deal about the nature of mental imagery: images represent a scene
in a fashion that preserves all of the distance relationships within that scene.
-Similar results are clear in the transformation of mental images.
-In an experiment, participants were shown two different shapes. The participant had to
decide f the second image was in fact different, or just a rotation of the first image. Is it
possible to rotate the image mentally?
-To perform this mental rotation task, participants first had to imagine one of the forms
and rotate it into alignment with the other. The amount of time it took to rotate it
depended on how much rotation was needed.
-Thus, again we see that the imagined movement resembles actual movement.
-Another way of interpreting all of the data above is that participants simply control the
timing of their responses in order to recreate the 'normal' pattern. (Knowing that moving
through the world takes time, and the greater the distance, the longer it takes to travel)
-Instead of a participant immediately scanning the imaged in their minds eye they could
be thinking about what they were asked to do and recognizing that it would take a
certain amount of time to travel from one point to the next, so in he desire to show they
are listening to instruction, wait a specific amount of time to hit the button.
-Participants are sensitive to demand character : the cues that might signal how they are
'supposed' to behave. InteractionsbetweenImageryandPerception
-Seeing the parallels between visual images and visual stimuli, we now ask what is the
relation between imagining and perceiving?
-In a study, participants were asked to detect faint signals- either dim visual stimuli or
soft tones. They had to say whether a signal was given or not for each trial. Participants
did this in either of two environments: wither while forming a visual image before their
'mind's eye' or while forming an auditory image before their 'mind's ear'.
-Hypothesis: that if the participants were busy imagining they would not be open to
perceiving. So, we should expect to see competition if participants were asked to do
-Exactly what they found Their results indicate that forming a visual image interferes
with seeing and that forming an auditory image interferes with hearing.
-What happens if participants try to imagine stimulus related to the one they are trying to
-Participants asked to think about either 'H' or 'T'. Then shown either an H or T at a very
low contrast, making the image hard to see. Visualizing an H made it much easier to
see H, and visualizing T made it easier to see T.
-This again shows that visualizing and perceiving draw on the same mechanisms, so
one activity can serve as a prime to the other.
-Biological evidence also shows that the brain structures specific to vision are also used
-Visual perception relies heavily on tissue located in the occipital cortex (the same relied
on to examine visual stimulus)
-V1 and V2 in visual cortex are involved in earliest stages of visual perception- active
when participants are visualizing highly detailed images.
-MT/MST highly sensitive to motion in ordinary visual perception, and activated when
participants are asked to visualize a moving image.
-Parts of the brain used to identify faces, is active when asked to imagine a face.
-In the case of brain damage, the same trends continue. Brain damage resulting in the
inability to see colour, results in not being able to visualize scenes in colour either.
-UsingTranscranial Magnetic Stimulation (TMS) (creating a temporary disruption in a
specific brain region), you can see further proof of this.
-So, the 'neural machinery' os needed for imagery overlaps with that needed for
perception. If the machinery is occupied with one, it is not available for the other. If the
machinery is disrupted then both are compromised. The same brain structures are
involved in visualizing and in vision, thus there are intimate relationships between imagery and perception.
-The parallels here, implies that imagery and perception should function in similar ways,
research indicates this is true.
-Consider visual acuity- the ability to see fine detail. In vision, acuity is much greater in
the centre of visual field than it is peripherally.
-Studies show a remarkable similarity between a person's acuity in vision and in
-In both cases, acuity fell off abruptly if the dots were not in the centre of the visual field.
-A number of studies have examined people who have been blind from birth as well.
-Images that needed to be visualized were made into sculptures for the participants to
be explored with the hands. Once this is done, the test continues the same as it did for
those with vision. Scanning across images and forming a rotation.
-The data was quite similar to those obtained with sighted research participants.
Response times were proportionate to the 'distance' traveled, and so on.
-Their 'spatial imagery' might be represented in the mind in terms of a series of imagined
movements, so that it is body or motion imagery rather than visual imagery.
Alternatively, perhaps spatial imagery is not tied to any sensory modality but is part of
our broader cognition about spatial arrangements and layout.
-Visual Imagery: represents spatial layout in terms of how things look.
-Spatial Imagery: cab represent spatial layout in terms of movements or body feelings.
-Blind people seem to use spatial imagery, where sighted people seem to be able to use
-Visual imagery relies on brain areas that are also needed for vision, and so damage to
these areas disrupts both imagery and vision. Spatial imagery, relies on different brain
areas and so damage to visual areas will not interfere with this from of imagery and
damage to brain sites needed for this imagery won't interfere with vision.
-Most people have the ability to use both visual and spatial types of imagery. This leads
to ask When do people use what kind of imagery?
-Ex: to think about colour you will need to use visual imagery (it will not be enough to
think about position in space). In tasks like mental rotation you can use either form of
-The choice between which type of imagery to use will depend on an individual's ability
levels- you may be good at one and poor at another.
-Roughly 10% of population will declare themselves entirely unable to see mental
-However, Studies have compared 'non imagers' with 'vivid imagers' and shown that
there is no difference between vivid and non imagers in how they do mental rotation, or how quickly and accurately they scan across an image.
-Notice though, that the studies just mentioned are all studies that can use spatial
imagery. Therefore, there is no relationship between the self-report and performance of
these tasks. There's no reason to think that how vivid someone's visual memory is will
be related to their performance.
-There should be differences though, with non and vivid visualizers on tests that require
visual tasks. Studies show that there is a destination between these two groups of
people when completing a task that requires information about what an image looks like.
-This shows that self- report can have meaning and can reveal individual differences.
-We need to be clear that our percepts (our mental representation of the stimuli we are
perceiving) are in come ways similar to pictures, but in other ways different. Like
percepts, pictures are depictions, representing key aspects of the 3D layout of the world.
Percepts are different from pictures because they are organized and unambiguous in a
fashion that pictures are not. (Think: Necker cube).
-Why are discoveries from images sometimes possible and sometimes not?
-Images are organized depictions, and so they are understood a certain way. they have
a certain top and bottom, figured foreground and background.
-Therefore, we should think of mental images as a package that includes both depiction
itself and this 'perceptual reference frame. Thus discoveries must be made about the
-On this basis, discoveries about the imaged form will be guided by both depiction and
-A simple hin