Douglas L. Nelson - US grants

This project will continue to focus on efforts to describe and understand processes associated with the storage and retrieval of recently encoded familiar experiences. The current working model is described by structural, processing, and retrieval assumptions. Structural assumptions characterize the nature of prestored information and identify what can be encoded about the current stimulus. Processing assumptions specify steps involved in encoding a familiar experience, including featural analysis, order of access, and the role of context. Finally, retrieval assumptions characterize the stages involved in achieving access to a recently encoded experience in the presence of some retrieval cue. This conceptualization is being evaluated using both cued recall and recognition procedures. Preliminary results indicate that retrieval is a function of the perceived similarity between the test cue and the encoded representation. It is also a function of the number of related experiences (category size) implicitly activated during either study or test. The larger the number of functionally encoded sensory or semantic representations, the poorer the recall. The number of functionally encoded concepts appears to be determined by an interaction between the number of prestored representations and the context of the encoding.

The general objective is to understand how familiar concepts are encoded in memory during a brief experience and retrieved at later points in time. Our current working model assumes that the presentation of a familiar word activates its prestored representations as well as other related representations. These representations concern visual, phonemic, and meaning information and, therefore, they differ in type. Within types, there are differences in quantity. Familiar words activate few to many related representations, and the number activated is an important determinant of the likelihood of their later recall. The model specifies when and how implicitly activated and related concepts become encoded and how they influence retrieval. Hence, the model and the related data describe how prior knowledge about a familiar concept influences its memorability. The proposed project is focused on several broad areas: (1) Understanding the independent roles that visual, phonemic, and meaning information play in encoding and retrieval, (2) Understanding the role of prior knowledge and its role in remembering, (3) Exploring the influence of encoding context in determining what is encoded about the presented stimulus, (4) Investigating the nature of retrieval processes in recall and recognition. Our findings are relevant to understanding memory in normals, and have implications for theories of memory and for practitioners in the mental health field who rely on memory principles in their dealing with clients.

Experiencing a familiar concept implicitly activates related concepts. This activation processes. However, as the project has shown, such activation also affect memory for direct experience. The goal of this research is to understand how implicitly activated concepts affect the encoding and retrieval of direct experience and to develop a theoretical model that explains this influence. The working model assumes that the encoding of related concepts incorporates them into the learning or testing episode, and that such 'Incorporation can either facilitate of hinder depending on the tasks that must be performed. The model specifies when related concepts are likely to be encoded and how they affect performance. The methodology requires subjects to encode familiar concepts under various learning conditions and to retrieve these concepts under various testing conditions. Variations in learning conditions include manipulations of context, encoding orientation, timing and interference; variations in testing conditions include manipulations of type of retention test, the nature of the retrieval cues and location of testing. All experiments involve manipulations of the number of related concepts activated by directly experienced concepts. In general, concepts that activate larger networks of related concepts are not as likely to be remembered, nor are they as likely to be effective as retrieval cues. These findings hold for both phonemically and meaning- fully related concepts. Implicitly activated concepts can interfere with memory for what was actually experienced. However, there are exceptions to these general patterns, even reversals of effect, and they help explain how implicitly activated concepts affect behavior. Findings will be relevant to understanding memory in normals, and will have direct implications for research in perception, learning, speech, reading, language comprehension, and for practitioners in mental health who rely on cues to help clients retrieve information.

DESCRIPTION (Applicant's Abstract): Experiencing a familiar concept provides immediate access to associated information that can influence later memory. The investigators' methods require subjects to study familiar words and to remember them in the presence of retrieval cues. They vary encoding and test instructions, the presence of contextual cues during study, the nature of the test cue, and test delay. What is most unique, the proposed experiments vary the associative structure of the studied words and the cues used to prompt recall. Associative structure represents connections acquired prior to the laboratory task as a result of world experience. They use association norms to measure this structure and have collected norms on 4,500+ words. These norms indicate that words vary in how many associates they have and in how connected these associates are. The investigators' research has shown that cued recall is more likely for words with fewer as opposed to more associates. These results are obtained regardless of the learning strategy used to encode the items. However, the recall advantage for words having fewer associates is substantially reduced after switching attention to an unrelated math task, when the words are encoded in context, in recognition tests, and when subjects are asked to exclude these words in recall. Their results also show that cued recall is more likely when the associates of a studied word are more connected. In contrast to the effects of number of associates, high levels of associative connectivity also affect recognition. The purpose of the project is to test predictions of a new model for explaining these findings, to determine when pre-existing structures affect the retrieval of a recently encoded component of that structure, and to explore questions concerning the measurement of associative structure. Briefly, the theory assumes that cued recall is affected by variables that influence the probability of sampling related information in long-term working memory, such as how many competing associates are involved, and by variables that influence recognition, such as connectivity, frequency, encoding strategy, and so on. The project will tell how prior knowledge about a familiar concept becomes involved during encoding and retrieval and how such knowledge affects ones ability to recover specifically targeted memories. The work will also be relevant to the role of: 1) attention in maintaining information in long-term working memory; 2) inhibition in dampening the influence of associated information; and 3) intentionality in the utilization of mediated connections. The research is producing a useful normative database and will increase the understanding of what association norms can tell us about associative structure. Finally, this project has implications for health research. For example, the investigators' work shows that people often remember associated memories in place of what was actually experienced. False memories associated with early childhood may be produced by similar processes. Furthermore, the project offers a new way to explore individual differences. The reduction or absence of effects related to associative structures in special populations (e.g., the elderly, stroke victims, substance abusers, deaf, depressed, and so on) could provide useful information about how these individuals process information as well as providing helpful diagnostic information.

DESCRIPTION (provided by applicant): Seeing a familiar word unconsciously activates related associates in memory that affect how well it is remembered later on (e.g., Dime activates penny, nickel, and so on). In this research participants study words, and then their memory is tested by presenting cues to help them remember the words actually seen. We vary how the words are studied and tested and whether attention to the memory task is disrupted before the test. In what is unique about this project we vary the associative structure of the unconsciously activated associates. Associative structure refers to connections among words acquired in everyday life prior to the laboratory task, and these connections have been measured in our lab for thousands of words using normative procedures. These procedures have shown that words vary systematically in terms of how many associates they have and in terms of how organized or connected these associates are. Our memory work shows that memory for a word actually seen is better when its associates are fewer and more densely interconnected, and that these effects depend on both how good the test cue is and whether attention to the memory has been disrupted prior to testing. What is most interesting about this finding is that it indicates that unconsciously activated past experience effects memory for something recently seen. What is known affects memory for what is new. This project will test predictions of a new model for explaining these and many other findings. This work carries implications for health research. For example, the findings show that people often remember associated memories in place of what was actually experienced. False memories may be produced by similar processes. Furthermore, the project offers a new way to explore individual differences. Differences in effects related to manipulations of associative structures in special as compared to normal populations (e.g., the elderly, stroke victims, substance abusers, deaf, depressed) can provide useful diagnostic information about how these individuals process information. The ideas about activating associative structures developed in this project are being actively applied currently to the study of expectations concerning eating disorders, alcohol, and pain perception.