4220 chapter 9
Semantic verification
Spence, Nicholls, and Driver (2000) required that participants quickly indicate whether a signal (either a light, a touch, or a tone) occurred on their left side or right side. They found that response times were slower when the previous signal was from a different modality than when it was from the same modality. Pecher, Zeelenberg, and Barsalou (2003) therefore reasoned that if people simulate information, they should be slower when they have to shift modalities when verifying properties in a conceptual task.
Verification of semantic statements
A popular procedure for studying the organization of semantic knowledge is to ask people to quickly verify semantic statements. The experimenter might present a statement such as "A bird is an animal" and ask the subject to respond true or false as quickly as possible.
Schema
A schema is a cluster of knowledge that represents a general procedure, object, percept, event, sequence of events, or social situation (Thorndyke, 1984). Schema theory refers to a collection of models presuming that we encode such knowledge clusters into memory and use them to comprehend and store our experiences.
Spreading activation model
A model that accounts for response times by formulating assumptions about how activation spreads in a semantic network
Bartlett's schema theory
A fundamental assumption of Bartlett's schema theory is that all new information interacts with old information represented in the schema.
feature comparison model
A model proposing that items are categorized by matching the items features to category features assumes that instances are classified by comparing the features, or attributes, of the two nouns representing the member and the category. To verify that a robin is a bird, a person would compare the features of robin with the features of bird .
hierarchical network model
A model proposing that items are categorized by using the hierarchical relations specified in a semantic network. assumes that category information is stored directly in memory by means of associations. robin is associated with bird and bird is associated with animal
THE PERCEPTUAL SYMBOLS MODEL
All theories described thus far in this chapter are amodal because they do not directly represent the perceptual experiences encountered in learning about concepts. A dramatically different theoretical approach is the percep-tual symbols system proposed by Barsalou (1999), in which perceptual experiences are directly stored in LTM. A perceptual symbols system is a modal approach because it stores sensory experiences such as audition,vision, taste, smell, and touch. Figure 9.9 shows the distinction between the previously discussed amodal approaches and the modal approach proposed by Barsalou (1999). Amodal systems represent perceptual information indirectly as a list of features or as a list of associations in a semantic network.In contrast, perceptual symbol systems retrieve information from LTM by reenacting or simulating perceptual experiences.
The Hierarchical Network Model
Each word in the network is stored with pointers (arrows) showing how it is related to other words in the network. By following the pointers,we know that ostrich and canary are examples of birds and that bird and fish are examples of animals. We also know that a canary, an ostrich, a shark, and a salmon are animals because the pointers connect these instances with the superordinate category animal
Building Semantic Networks
Oneway to organize information is to construct a semantic network. Semantic networks show how concepts are related to each other. Networks are typically represented by diagrams in which the concepts are called nodes and the lines showing the relationship between two concepts Characteristics are the features or properties of a concept.
The Hierarchical Network Model
Furthermore, the level in the network where the properties are stored influences response times. Another interesting prediction based on the network model concerns facilitating retrieval from memory (Collins & Quillian, 1970). Facilitation occurs when the retrieval of information is made easier because the previous question required retrieval of similar information. For example, it should be easier to verify a property of a canary if the previous question was also about a canary.The network model, however, allows us to make a more specific prediction.
The Spreading Activation Model
However, spreading activation should only activate semantically related words if associations exist among those words. A child who does not have strong associations between the word needle and the words thread, pin, eye , and sewing should be less likely to falsely recognize needle . An implication of this hypothesis is that children should make more false recognitions as they learn more associations among words.
Script
Knowledge about what occurs during routine activities Because the sequence of events is quite standard, a natural way of organizing scripts is according to the temporal order in which the events occur.Imagine that you have a flat tire and need to put on the spare tire. Your memory might consist of an organized sequence of actions, beginning with what you do first and ending with what you do last.
Modern schema Theory
Let's take a familiar routine for students: registering for courses. What kinds of knowledge would you like to have before signing up for a course?You would probably want to know what the prerequisites are, whether it meets some requirement, how many credits you will receive, when and where it meets, and perhaps who is teaching it. An advantage of having schematic knowledge is that we can sometimes rely on default knowledge — that is, likely values that enable us to make intelligent guesses in the absence of specific knowledge. For example, most lecture courses offer the same number of credits, so you could probably guess how many credits you would receive for a cognitive psychology course before looking it up. You might also have guessed that the prerequisite was introductory psychology and might even have guessed the name of the instructor if the same person usually teaches this course.
Recall of Hierarchical Information
One advantage of a well-organized memory is that it helps us retrieve information in a systematic way. If the information was not systematically organized, the task would be very difficult, and your level of achievement would not be very impressive.
Limitations of the feature comparison model
One interpretation of this finding is that characteristic features are more salient and directly observable than defining features (T. P. McNamara &D. L. Miller, 1989). Characteristic features such as mean and gun are observable, whereas defining features such as taking without permission are more conceptual. Young children initially emphasize the directly observable features and have to learn to shift their emphasis to the more conceptual features.In conclusion, the feature comparison model has some advantages over the hierarchical network model, but it also has some limitations.
Limitations of the feature comparison model
Presumably, these similarity judgments are made by comparing the features of the example and the category concept, but there is little direct support for this assumption. A second criticism of the feature comparison model is its proposal that all classifications require computations — that we use the features of concepts to compute their degree of similarity. A third criticism of the feature comparison model is the argument against necessary, or defining, features
The Spreading Activation Model
Recall that priming is the facilitation in the detection or recognition of a stimulus by using prior information. One controversy regarding the spreading activation model (Ratcliff &McKoon, 1988) is whether activation spreads beyond a single node as predicted by the model. Although BREAD activates the word BUTTER , would it activate a word like POPCORN that is associated with BUTTER , but not with BREAD? According to the model, activation should spread from BREAD to BUTTER to POPCORN , but POPCORN should be less activated than BUTTER because activation decreases in strength as it spreads outward.
Modern Schema Theory
Rumelhart argued that schema are the building blocks of cognition.According to Rumelhart, a schema theory is basically a theory about how knowledge is represented and about how that representation facilitates the use of knowledge in various ways. Schemas are used to interpret sensory data, retrieve information from memory, organize action, and solve problems.
Schema Theory
Semantic networks provide a convenient way of organizing knowledge, but the emphasis is on showing how two nodes are related, and not on showing how ideas are grouped together to form larger clusters of knowledge. The advantage of the subnode model is that it offers a way of forming larger clusters by grouping related ideas around subnodes.
The Feature Comparison Model
The feature comparison model has two stages. The first stage compares all the features of two concepts to determine how similar one concept is to the other. For example, to determine whether a robin is a bird, we would compare the features of robin with the features of bird. If the comparison reveals that the two concepts are either very similar or very dissimilar, we can respond true or false immediately. The second stage is necessary when the degree of similarity is between the two extremes. The answer isn't obvious in this case, so the model proposes that we examine only the defining features to determine whether the example has the necessary features of the category.
typicality effect
The finding that the more typical members of a category are classified more quickly than the less typical category members
The Feature Comparison Model
The probability that the second stage is necessary increases as the similarity between the category concept and the example decreases. The model therefore predicts that the more typical members of a category (such as robin, sparrow, blue jay should be classified more rapidly than the less typical members ( chicken, goose, duck ) because evaluating the defining features during the second stage slows the classification. E. E. Smith and colleagues (1974) found that people could classify instances that are typical of the category faster than they could classify instances that are not typical of the category.
The Feature Comparison Model
This model assumes that the meaning of words can be represented in memory by a list of features and that classifications are made by comparing features rather than by examining links in a network (see Figure 9.4). The features can be used to define categories,but they vary in the extent to which they are associated with a category.
Recall of Hierarchical Information
When the associated words were linked together, people recalled many more words than when the same words were randomly linked together (for example, when cat and bread were linked to yellow ). Semantic organization of the material improved recall,even though the organization did not consist of a hierarchy. The difference between the organized and random conditions was more striking for hierarchical organization,
Autobiographical Memory
autobiographical memories do not consist of abstract, homogenized information. Rather, they contain information from sensory, language, emotion, and other systems that process different kinds of information. Because, there are many memory systems that need to interact, memory becomes highly constructive, with the constructions guided by specific schemas associated with the different systems.Autobiographical memory shares similarities with episodic memory dis-cussed in Chapter 5. Like episodic memory, autobiographical memory has a context. Events in our lives occur at a particular place and time.