Memory for scripts in young and older adults

LEAH L. LIGHT 0 0 PitzerCollege , Claremont, California This study examined the question of whether young and older adults differ in their representation or utilization of the generic knowledge contained in scripts. In Experiment 1, young and older adults generated scripts for routine daily activities, such as grocery shopping, going to the doctor, and writing a letter to a friend. No evidence was found for age-related differences in the way that stereotypical action sequences are represented in semantic memory. In Experiment 2, young adults were found to recall and recognize new instantiations of scripts better than did older adults. However, adults in both age groups displayed similar effects of action typicality on retention, suggesting that there are no age-related differences in drawing inferences from generic knowledge. The implications of these findings for processing-resource hypotheses about memory and aging are discussed. - differences on a delayed test for both explicitly stated and implied information but no age differences when people were tested immediately after having read pas sages. Belmore's results, taken together with those of Burke and Yee (in press), suggest that older adults do not have problems in accessing real-world knowledge during comprehension but rather that the problem lies in a more general memory impairment. This would explain Cohen's (1979, Experiment 2) finding that older adults were less accurate than young adults in detecting when short passages contained statements that con tradicted common knowledge about the world (e.g., that a blind man read a newspaper or that a housewife who had no bread made sandwiches). In this study, there was no measure of memory for information in the passages; it is therefore possible that failure to detect anomaly resulted from problems in memory for explicit informa tion rather than from failure to access or utilize prag matic information in long-term memory. In this paper, we report an experiment that bears on the question of whether young and older adults differ in the extent to which they utilize real-world knowledge in recall and recognition of discourse. Young and older adults were exposed to a story about several days in the life of a character named Jack, who engages in a set of mundane activities such as writing a letter to a friend and going grocery shopping. Schank and Abelson (1977) suggested that we represent our knowledge of such stereotypic action sequences as "scripts." These scripts specify conventional roles, props, action sequences, reasons for engaging in an activity, and expected out comes for ordinary activities. For example, we expect to find waiters and menus in restaurants, we go to restaurants because we are hungry, we eat, and then we leave after having paid the bill. There is evidence that scripts play an important role in comprehension and memory for particular instances of conventional activi ties. Understanding requires people to match what they see or hear to a previously stored representation of actions. Thus, new information is understood in terms of old information. Details that are omitted from a particu lar instantiation of a script are inferred on the basis of the stereotypic sequence that defines well-known situa tions. For this reason, college students are likely to infer that highly typical actions have taken place even when these have not been explicitly stated and to later intrude these typical actions in recall and to make false alarms to them in recognition (e.g., Bower, Black, & Turner, 1979; Graesser, Gordon, & Sawyer, 1979; Graesser, Woll, Kowalski, & Smith, 1980). Such findings can be explained in terms of the "script pointer plus tag" hypothesis of Schank and Abelson (1977) or the partial copy model of Bower et al. (1979). These hypotheses have in common the notion that when an instantiation of a particular activity is encountered, its generic script is activated in semantic memory. In addition, the spe cific activities mentioned (Bower et al., 1979), or at least those that deviate from the generic scripts (Schank & Abelson, 1977), are recorded in episodic memory. Highly typical actions are likely to be recalled or (falsely) recognized even when they have not been stated ex plicitly, because they are activated as part of the generic script. Hence, when scripts are invoked, typical actions are discriminated less well than are atypical ones. Comparing the performance of young and older adults on memory for scripts provides a testing ground for hypotheses about why older adults remember dis course less well than do young adults. Here we will consider two hypotheses about age-related differences in memory for scripts. These hypotheses share the premise that older adults have reduced processing resources; they differ in the hypothesized consequences of reduced attentional capacity. The first hypothesis we examine is that of Cohen (1979), who argued that, due to reduced capacity in working memory, older adults have problems in inte grating new information with prior knowledge. This hypothesis is related to the more general claim that older adults have a deficit in semantic processing (e.g., Craik & Simon, 1980; Eysenck, 1974; Simon, 1979) that results in poorer comprehension and memory. If older adults are less likely to access general world knowl edge, that is, generic scripts, during comprehension of discourse, they should be less likely than young adults to infer falsely that highly typical actions take place when these are not stated explicitly. As a result, older adults should exhibit smaller typicality effects than those that young adults exhibit and should be more accurate than are young adults in discriminating typical actions that have been mentioned from those that have not been mentioned. A second, and rather different, set of predictions is derived from the work of Rabinowitz, Craik, and Ackerman (1982). These authors proposed that reduced processing resources in the elderly result in decreased likelihood that the specific meanings of events in con text are encoded, because encoding information in context is effortful, whereas encoding of generic mean ing is relatively automatic (Hasher & Zacks, 1979). Although Rabinowitz et al. dealt with memory for single words, their hypothesis has implications for script memory as well. Saying that the general meaning of a script is encoded is theoretically equivalent to saying that the generic script has been activated in semantic memory. Similarly, failure to encode the meaning of a word in context can be viewed as being equivalent to failure to record the details of the instantiation of a script on a particular occasion. Because of reduced capacity, then, older adults would be expected to en code all instantiations of a script in more or less the same stereotyped way and to be less likely to encode each instantiation in a distinctive, contextually specific way. Thus, the recall protocols of older a (...truncated)