CHAPTER 9: KNOWLEDGE One of the key properties of a connectionist network is
that a specific category is represented by activity that
Back propagation A process by which learning can is distributed over many units in the network. This
occur in a connectionist network, in which an error contrasts with semantic networks, in which specific
signal is transmitted backward through the network. categories are represented at individual nodes. (9)
This backward-transmitted error signal provides the
information needed to adjust the weights in the Definitional approach to categorization The idea that
network to achieve the correct output signal for a we can decide whether something is a member of a
stimulus. (9) category by determining whether the object meets the
definition of the category. See also Family
Basic level In Rosch’s categorization scheme, the level resemblance. (9)
below the global (superordinate) level (e.g., “table” or
“chair” for the superordinate category “furniture”). Error signal During learning in a connectionist
According to Rosch, the basic level is psychologically network, the difference between the out- put signal
special because it is the level above which much generated by a particular stimulus and the output that
informa- tion is lost and below which little is gained. actually represents that stimulus. (9)
See also Global (superordinate) level; Specific
(subordinate) level. (9) Exemplar In categorization, members of a category
that a person has experienced in the past. (9)
Categorization The process by which objects are
placed in categories. (9) Exemplar approach to categorization The approach
to categorization in which members of a category are
Category Groups of objects that belong together judged against exemplars, examples of members of the
because they belong to the same class of objects, category that the person has encountered in the past.
such as “houses,” “furniture,” or “schools.” (9) (9)
Family resemblance In considering the process of
Category-specific knowledge impairment A result of categorization, the idea that things in a par- ticular
brain damage in which the patient has trouble category resemble each other in a number of ways.
recognizing objects in a specific category. (9) This approach can be contrasted with the definitional
approach, which states that an object belongs to a
Cognitive economy A feature of some semantic category only when it meets a definite set of criteria. (9)
network models in which properties of a cat- egory that
are shared by many members of a category are stored Graceful degradation Disruption of performance due
at a higher level node in the network. For example, the to damage to a system that occurs only gradually as
property “can fly” would be stored at the node for parts of the system are damaged. This occurs in some
“bird” rather than at the node for “canary.” (9) cases of brain damage and also when parts of a
connectionist network are damaged. (9)
Concept A mental representation used for a variety of
cognitive functions, including memory, reasoning, and Hidden units Units in a connectionist network that are
using and understanding language. An example of a located between input units and output units. See also
concept would be the way a person mentally Connectionist network; Input units; Output units. (9)
represents “cat” or “house.” (9)
Connection weight In connectionist models, a Hierarchical model As applied to network models of
connection weight determines the degree to which knowledge representation, a model that consists of
signals sent from one unit either increase or decrease levels arranged so that more specific concepts, like
the activity of the next unit. (9) canary or salmon, are at the bottom and more general
concepts, such as bird, fish, or animal, are at higher
Connectionism A network model of mental operation levels. (9)
that proposes that concepts are rep- resented in
networks that are modeled after neural networks. This Hierarchical organization Organization of categories
approach to describing the mental representation of in which larger, more general categories are divided
concepts is also called the parallel distributed into smaller, more specific categories. These smaller
processing (PDP) approach. See also Connectionist categories can, in turn, be divided into even more
network. (9) specific categories to create a number of levels. (9)
Connectionist network The type of network proposed
by the connectionist approach to the representation of High prototypicality A category member that closely
concepts. Connectionist networks are based on neural resembles the category prototype. See also
networks, but are not necessarily identical to them. Prototypicality. (9)
that a specific category is represented by activity that
Back propagation A process by which learning can is distributed over many units in the network. This
occur in a connectionist network, in which an error contrasts with semantic networks, in which specific
signal is transmitted backward through the network. categories are represented at individual nodes. (9)
This backward-transmitted error signal provides the
information needed to adjust the weights in the Definitional approach to categorization The idea that
network to achieve the correct output signal for a we can decide whether something is a member of a
stimulus. (9) category by determining whether the object meets the
definition of the category. See also Family
Basic level In Rosch’s categorization scheme, the level resemblance. (9)
below the global (superordinate) level (e.g., “table” or
“chair” for the superordinate category “furniture”). Error signal During learning in a connectionist
According to Rosch, the basic level is psychologically network, the difference between the out- put signal
special because it is the level above which much generated by a particular stimulus and the output that
informa- tion is lost and below which little is gained. actually represents that stimulus. (9)
See also Global (superordinate) level; Specific
(subordinate) level. (9) Exemplar In categorization, members of a category
that a person has experienced in the past. (9)
Categorization The process by which objects are
placed in categories. (9) Exemplar approach to categorization The approach
to categorization in which members of a category are
Category Groups of objects that belong together judged against exemplars, examples of members of the
because they belong to the same class of objects, category that the person has encountered in the past.
such as “houses,” “furniture,” or “schools.” (9) (9)
Family resemblance In considering the process of
Category-specific knowledge impairment A result of categorization, the idea that things in a par- ticular
brain damage in which the patient has trouble category resemble each other in a number of ways.
recognizing objects in a specific category. (9) This approach can be contrasted with the definitional
approach, which states that an object belongs to a
Cognitive economy A feature of some semantic category only when it meets a definite set of criteria. (9)
network models in which properties of a cat- egory that
are shared by many members of a category are stored Graceful degradation Disruption of performance due
at a higher level node in the network. For example, the to damage to a system that occurs only gradually as
property “can fly” would be stored at the node for parts of the system are damaged. This occurs in some
“bird” rather than at the node for “canary.” (9) cases of brain damage and also when parts of a
connectionist network are damaged. (9)
Concept A mental representation used for a variety of
cognitive functions, including memory, reasoning, and Hidden units Units in a connectionist network that are
using and understanding language. An example of a located between input units and output units. See also
concept would be the way a person mentally Connectionist network; Input units; Output units. (9)
represents “cat” or “house.” (9)
Connection weight In connectionist models, a Hierarchical model As applied to network models of
connection weight determines the degree to which knowledge representation, a model that consists of
signals sent from one unit either increase or decrease levels arranged so that more specific concepts, like
the activity of the next unit. (9) canary or salmon, are at the bottom and more general
concepts, such as bird, fish, or animal, are at higher
Connectionism A network model of mental operation levels. (9)
that proposes that concepts are rep- resented in
networks that are modeled after neural networks. This Hierarchical organization Organization of categories
approach to describing the mental representation of in which larger, more general categories are divided
concepts is also called the parallel distributed into smaller, more specific categories. These smaller
processing (PDP) approach. See also Connectionist categories can, in turn, be divided into even more
network. (9) specific categories to create a number of levels. (9)
Connectionist network The type of network proposed
by the connectionist approach to the representation of High prototypicality A category member that closely
concepts. Connectionist networks are based on neural resembles the category prototype. See also
networks, but are not necessarily identical to them. Prototypicality. (9)
