School Psychology Summary
Chapter 5: How are Conceptual Development and the Biological World
Related?
Conceptual knowledge is knowledge about the kinds of things in the world. This is not only
about knowledge about objects, but also about actions, events, and mental states.
Conceptual development consists of inductive learning and categorization. Induction is a
way of reasoning in which one deduces the general from the specific. Generalising on the
basis of a known example is one of the most common forms of inductive reasoning and is
the basis of categorization. With induction we fill gaps in our knowledge. The ability to
reason on the basis of induction is already present early in development. The focus of
research is therefore on the organisation of knowledge that determines the ability of
categories.
What are superordinate, subordinate, and ‘basic level’ categories?
Categorization entails organising things into a category. There are different levels within a
category: superordinate level (animal), basic level (cat), and subordinate level (siamese).
Neisser (1987) describes categorization as the ability to treat a set of things as equal, to put
them in the same pile, to call them the same, or to respond to them in the same way.
Categorization is important because we cannot consider every object as unique and
because it gives us the opportunity to find out more about an unknown object by just looking
at it and classifying it into a specific category. Perceptual information, conceptual information
and beliefs play a role in categorization. The prototype theory is the theory that the normal
level of a category (basic level) gives us the most information about what something is and
how it should be classified.
What are ‘basic level’ categories and prototypes?
The categorization of exemplars as similar implies that a generalised representation or
prototype of a category has been formed. Subsequently present stimuli are then compared
to this prototype. We categorise a lot on the basic level (normal level; from now on we will
refer to the levels as basic, subordinate and superordinate). It is thought that items from the
same basic level category have the most characteristics in common and the fewest
characteristics in common with items that do not fall within the same basic level category.
In Chapter 1 it was discussed that infants can form prototypes of drawn pictures. Eimas and
Quinn (1994) showed that they can also do this with existing animals. They allowed three to
4-month-old infants to inhabit pictures of horses. Then they showed a new picture of a horse
in the test phase together with pictures of a zebra, a giraffe or a cat. The infants looked at
the other animals longer. This shows that they can see the difference between the prototype
of a horse and that of another animal. In the physical world size, sounds and movement
patterns can also help to classify objects or organisms into a prototype. Arterberry and
Bornstein (2001) investigated this. They found that infants of three months old could
,distinguish between categories based on rich visual input just as well as based on
movement cues.
However, these tests provided information about perceptual categories rather than about
conceptual categories. To test conceptual categories, we look at the sorting behaviour of
infants who cannot yet talk. For example, if they assign toy cars to toy cars and toy horses to
toy horses, then they probably have knowledge of conceptual significance.
How can sequential touching be used as a measure for basic level categorization?
However, children do not show spontaneous sorting behaviour before the age of eighteen
months. As a sign of categorization, they use sequential touching. They touch objects that
belong to a certain category in sequence, so first all cars and then all horses. Mandler and
Bauer (1988) investigated this at a basic level (cat and car) and at a superordinate level
(animal and vehicle). Twelve and fifteen-month-old infants distinguished only through
sequential touch on a basic level. Only infants aged 20 months made a distinction at
superordinate level.
There were many individual differences, for example, that 25% of the twelve-month-old
infants were responsive to the superordinate level. In a follow-up experiment, they showed
that basic level categories were easily distinguished from each other if they belonged to two
different superordinate categories, but more difficult if they belonged to the same
superordinate category. Superordinate classification therefore plays a larger role than was
thought.
However, these results can also be explained differently. It could also be due to the
perceptual context that items from the same superordinate class are more difficult to
distinguish from each other. The difference between a car and a dog is easy to see, for
example, because a car has no legs and no head. A difference between a horse and a dog
is more difficult to see, because they do have these characteristics in common. It is therefore
not entirely certain that the results of the study by Mandler and Bauer indicate that it is
impossible to distinguish different basic level categories within a superordinate category.
Furthermore, membership of a basic level often means membership of a subordinate level.
The theory (about the different levels devised by Rosch) states that perceptual equality
correlates with structural equality. The perceptual equality between dogs and horses
represents an underlying structural equality, namely that they are both organisms. Basic
level and superordinate concepts stem from perceptual knowledge.
Pauen (2002) found that organising objects into categories is also based on previous
knowledge that a child has about certain things. She allowed infants to hold on to some
objects of a category and look at them until they were used to it. Then she gave the infant a
new object from the same category or a new item from a different category. Pauen, however,
created a series of objects that were a mix of animals and furniture. She manipulated the
objects in such a way that there were more perceptual similarities between the different
groups than within the different groups. The furniture, for example, had eyes. Yet infants
spent more time looking at an item from a new category than at an item from a familiar
,category. So, they used their previous knowledge of categories to determine in which
category these objects belonged.
What is the matching-to-sample task?
A child's categorical capacity can also be determined with a matching-to-sample test. A
target is shown and then they are asked to select from a number of newly shown stimuli
which stimulus forms a pair with the target. Bauer and Mandler (1989b) had children perform
a matching-to-sample test based on basic level pairs and super-ordinate pairs. The children
(19, 25 and 31 months old) were slightly better in basic-level objects, but they also scored
very well on super-ordinate objects. So, from the age of nineteen months there is a
sensitivity for both basic level categories and superordinate categories.
Rosch argued that categorization starts at the basic level and then develops into
categorization at the superordinate and subordinate level. However, different studies indicate
something else.
What is the core developmental role of the superordinate level?
Mandler and Quinn et al. both suggest that categorization first takes place at a superordinate
level and thereafter at a basic level. Mandler says this is due to perceptual (basic level)
categorization and conceptual categorization. However, Quinn et al. argue that the
perceptual learning processes are sufficient for categorization at superordinate level first and
subsequently at basic level.
Mandler (1991) demonstrated his idea by allowing children of 19, 24 and 31 months of age
to distinguish between two basic level categories through sequential touching. He paired
dogs with horses (low perceptual contrast), dogs with rabbits (medium perceptual contrast)
and dogs with fish (high perceptual contrast).
Children of 31 months old could keep the dogs and horses apart. In the medium perceptual
contrast, children between the ages of 24 and 31 were able to see the difference well and
with regard to dogs and fish, any age group could make a distinction. All groups could also
see the difference between dogs and vehicles (different superordinate categories). Mandler
concluded that the children start at a superordinate level and are increasingly refining their
categories into basic level categories. However, with these results it cannot be concluded
with certainty that being unable to distinguish between a dog and a horse means that the
basic level categorization for dogs and horses is not present in the child.
Mandler and McDonough (1993) demonstrated that infants of seven, nine and eleven
months old do not make a distinction between dogs and fish in an object examination task,
but do make a distinction between animals and vehicles. This shows that they can
distinguish at a superordinate level, but not at a basic level. In this study, toy animals were
used and toy animals only show certain characteristics of an animal, but do not present the
animal in terms of smell, size and so on. These characteristics could be a key component in
distinguishing between different types of animals at the basic level. In addition, even in this
study it cannot simply be said that, if a child does not make the distinction, categorization
does not actually exist.
, Quinn has a theory which is called the global to basic sequence (first global categorization
and then basic level). This theory is based on perception. It derives from connectionist
modelling. This is a mathematical model of learning via neural networks. Every unit in the
network has an output that is a simple numerical function of its input. Cognitive skills, such
as language, are represented by patterns of activity at different units. Information therefore
enters the input nodes and relevant information about identifying characteristics is filtered
and passed on to the output nodes. So, it starts with global categorization, but during
learning, more and more nodes are being developed that will distinguish between multiple
details. This makes categorization on a basic level possible. The more of these nodes
develop, the more categorization takes place at the basic level and ultimately this is
preferred.
What are examples of cognitive neuroscience studies of infant categorization?
More recently, infant researchers have turned to EEG to measure early categorization
behaviour. In this literature, infant ERPs are used to provide an index of discrimination
between stimuli, offering an alternative to object examination procedures, and enabling
younger participants to be tested. ERP studies have produced evidence consistent with both
the basic level being discriminated first, and with the global level being discriminated first.
Quinn, Westerlund and Nelson (2006) contrasted pictures of cats with pictures of dogs, a
basic level distinction, in the first ever study of neural markers of infant categorization. EEG
was recorded as six-month-olds watched a series of 18 different pictures of cats, and then
were shown either more cat pictures, or some dog pictures. Quinn and his colleagues
reported a novel ERP, the Nc (a negative central component), related to viewing the novel
category (dogs). Therefore, for the basic level of categorization, a distinct neural marker
appears to be associated with categorical discrimination.
Subsequent studies have supported the Nc component as relevant to infant categorization at
the subordinate level as well. Meanwhile, Pauen and her colleagues used the Nc to
investigate infant discrimination of superordinate categories, using the animal versus
furniture paradigm also used in their object exploration work.
Although this early ERP evidence for both basic and global level categorization appears
confusing, neural studies of adult behaviour in perceptual categorization studies have found
that average feature values are automatically computed by the brain for any set of images
that is viewed (Hochstein et al., 2015). The ERPs recorded by researchers in infant
categorization paradigms are likely to depend heavily on the choice of images for a particular
experiment and on the nature of the experimental controls for perceptual similarity.
What evidence exists from sorting and match-to-sample paradigms with
preschoolers?
Poulin-Dubois and her colleagues have produced evidence consistent with Rosch’s original
developmental proposals, using both a sorting task and a match-to-sample paradigm.
Wright, Poulin-Dubois and Kelley (2015) argued that while a range of implicit paradigms
such as those discussed above have been used with infants and toddlers, there is little
explicit evidence showing that children are forming categories that involve naïve theories. To