journal of Gerontology 1975, Vol. 30, No. 3, 319-325
Aging and Interference with Memory1 Anderson D. Smith, PhD2
adults seem to have poorer memories than younger adults. Initially, it was assumed that the memory deficit in old age was the result of short-term memory failure (Welford, 1958). This conclusion was based on the observation that older persons failed to remember recently acquired information while remaining proficient in the recall of events occurring many years in their past. Upon close examination, however, older persons only have problems with memory tasks which exceed the immediate memory span, and the immediate memory span serves as a good empirical estimate of short-term memory capability (Talland, 1968). In addition, quantitative estimates of short-term memory (Craik, 1968) are relatively invariant across the age span. Estimates of long-term memory, on the other hand, do show marked and systematic decreases from young to old. The long-term memory deficit, seen with tasks which exceed the immediate memory span, has been replicated many times in recent years (Botwinick, 1973; Hultsch, 1971). Early investigators suggested that the aged might be more susceptible to interference than younger subjects (Welford, 1958). An early experiment by Cameron (1943) supported this position. Cameron compared groups of older and younger subjects in remembering a series of digits. The difference between the age 1 This investigation was supported by a NIH Research Grant No. HD06885-01 from NICHD. 2 School of Psychology, Georgia Institute of Technology, Atlanta 30332.
groups was greater when an interpolated task, presumably a source of interference, was included between the presentation and recall of the digits. Because of such evidence, Welford (1958) concluded that the aged had an . . . increased liability to interference . . . which could account for many of the cases where true learning or retention deficiencies can be found.
Recent investigators have focused on the specification of interference effects in the aged. Interference can be defined in terms of either proaction (i.e., interference from previously learned material), or retroaction (i.e., interference from tasks interpolated between the input and recall of the to-be-remembered items). There seems to be no support for an increased effect of proactive interference with old age. In one experiment, Fozard and Waugh (1969) presented two three-item lists to groups differing in age and found no interaction between age and list order even though performance for all groups declined on the second list. No evidence of proactive interference was found in other experiments involving successive colornaming tasks (Craik, 1968), multiple free recall tasks (Raymond, 1971), or list learning tasks (Talland, 1968). The relationship between age and retroactive interference is more equivocal. Cameron (1943), cited earlier, found an interaction between retroaction and age, but later investigators have failed to replicate these findings. Gladis and Braun (1958), for example, found 319
Downloaded from http://geronj.oxfordjournals.org/ at James Cook University on March 17, 2015
Input and output interference in both short-term and long-term memory were measured in three age groups (aged 20-39, 40-59, 60-80 years). A paired-associate probe technique was used, which, by factorially combining the positions of the pairs at presentation and recall, allowed an examination of output interference for all positions in the input list. No differential output interference or input interference due to age was found for items assumed to be recalled from long-term memory. Evidence, however, did indicate that adult aging primarily affects long-term memory, since no differences between groups were found for items assumed to be recalled from short-term memory.
320
SMITH
Downloaded from http://geronj.oxfordjournals.org/ at James Cook University on March 17, 2015
found only with items from the last half of the input list. Because of the immediate recall for these items, it is possible that the effect reflected the nature of short-term memory since the last half of the input list contained the recency items. Recency items are the last few items seen, and if recall for these items is immediate (with short retention intervals), the level of recall is high because of their availability in short-term memory. If recall is delayed later in the output sequence, however, the level of recall is reduced as the recency items in short-term memory are displaced by other materials. The Smith (1974) experiment, on the other hand, used a long-term memory paradigm since the number of items to be remembered exceeded the immediate memory span and since input position was counterbalanced so that recency would not influence the data. In this experiment, no differential interference effects were found. This suggests that the phenomenon of output interference might depend on the relative influence of short-term and long-term recall on the data. To clarify the relationship between age and output interference, a closer examination of input position effects, together with an empirical separation of short-term memory and long-term memory seems necessary. The paired-associate probe technique (Tulving & Arbuckle, 1966) has been used to examine output interference with college-aged subjects and, at the same time, to examine short-term and long-term recall separately. Using only colWith categorized materials, another experi- lege-aged subjects, Arbuckle (1967) presented ment failed to find differential output inter- ten paired-associate pairs at input and cued or ference due to age (Smith, 1974). While the probed at recall with each stimulus item in number of words recalled declined across the sequence. The input and output positions of category output sequence for three different each pair were counterbalanced to allow an adult age groups, no age-related differences in examination of each input-output combination. the slopes of the interference functions were Output interference was found only for the last items in the input list. These items are recency found. items and are assumed to be recalled from One possible explanation of the conflicting short-term memory. Initial and middle items results is the differential influence of short-term in the list were unaffected by the order of outmemory in the two procedures. Recall is put. These results were interpreted that output usually inferred to> involve short-term memory interference has an effect only in short-term when it is immediate (with short retention recall. intervals) and when the amount of information At the end of presentation, the last few items to be retained is small (within the immediate memory span). The Taub experiments typ- are stored in short-term memory. If the output ically used short-term memory paradigms such of these recency items occurs immediately after as reproducing the order of letter sequences. presentation, recall is high. If, however, the In addition, output interference was often recall of other items is interpolated between the no age-correlated retroactive interference effects when the age groups were equated on the basis of vocabulary and learning ability. The interpolated activity between presentation and recall necessary to produce retroactive interference can be further divided into the presentation of further material (input) and the recall of other information from memory during the interval (output). Tulving and Arbuckle (1966) refer to these effects as input interference and output interference. While this distinction is important to clarify the nature of possible age interactions with retroactive interference, few experiments have been conducted which consider the effects separately. One group of experiments conducted by Taub and his associates does suggest that aged individuals are less able than young subjects to maintain items in memory while simultaneously making a response, i.e., output interference (Taub, 1968; Taub & Grieff, 1967; Taub & Walker, 1970). In these experiments, a comparison was made between recall in the first half and second half of the recall sequence. The results suggested that older persons are more susceptible to output interference than younger subjects. No age effects were found in the first half of the recall sequence while effects were found for the second half. The comparison, however, between the two' halves of the recall order does not allow a precise measure of the nature of output interference. A better design would separate the recall sequence into smaller portions.
AGING AND INTERFERENCE WITH MEMORY
the subjects. One of the categories was randomly selected and served as a practice list to insure that each subject understood the instructions and could adequately perform the task. The order of the remaining eight categories was systematically varied across subjects. A single random order of instances from each category was determined and paired with the stimulus items. The stimuli for all lists were the digits, 1-8, which always appeared in sequential order in each input list. The subject was placed in a separate room from the experimenter and equipment. The subject room was separated from the equipment room by a translucent screen. A Kodak Ektagraphic projector was used to present the materials on the screen. The sequence of the experiment and the instructions were controlled by a magnetic tape recording.
METHOD
Subjects Seventy-two male alumni of the Georgia Institute of Technology participated in the experiment. The subjects were divided into three groups according to age: Group 1 included the ages from 20-39 years; Group 2 included the ages 40-59; and Group 3 included the ages 60-80. The subjects were solicited by mail, and volunteers were paid a small sum for participation. All subjects resided in the metropolitan Atlanta area, representing similar socioeconomic and occupational backgrounds. To insure the comparability of the age groups, two subtests of the Wechsler Adult Intelligence Scale (Wechsler, 1955), the forward digit span and vocabulary subtests, were administered to all subjects prior to their participation in the experiment. Stimulus Materials and Apparatus Nine different categories were selected from the Battig and Montague (1969) category norms (e.g., countries), and the first eight words with response counts of 150 or less were chosen from each category (e.g., Mexico, Japan, Sweden, etc.). Words which had ambiguous category membership were eliminated and replaced by the next word from the norm listing. These words served as the responses for the nine paired-associate lists given to the subjects in the experiment. Different categories were used to insure easy list differentiation by
Procedure After standard paired-associate instructions and practice, each subject received a single trial on each of eight different experimental lists by the study-test method of paired association. The eight pairs in each list were first presented one-at-a-time followed by recall of the response members when the stimuli were presented in a predetermined sequence. The input sequence for the pairs in each list was determined by the sequential ordering of the digit stimuli. The output sequence of the pairs was determined from the rows of an 8 X 8 balanced Latin square as seen in Table 1. The rows of the table represent the eight lists, the columns represent the output positions of the pairs at the time of recall, and the numbers represent the positions of the specific pairs in the input sequence during presentation. The factorial design of the experiment insured that each input position was represented at each output position. Each subject received all eight lists, Table 1. Sequence of Input and Output. Output Sequence 7 4 6 8 5 1 4 6 5 3 7 2 4 5 7 6 8 3 6 8 5 1 7 4 1 7 8 6 2 5 2 8 1 7 3 1 3 2 4 6 8 4 7 2 3 5 1 5 4 8 3 2 6 "Numbers in table refer to the serial position of the pair in the input sequence. 1 1" 2 3 4 5 6 7 8
2 2 3 4 5 6 7 8 1
3 8 1 2 3 4 5 6 7
Downloaded from http://geronj.oxfordjournals.org/ at James Cook University on March 17, 2015
presentation of the recency items and their subsequent output, the probability of recall is as low as for other items in the list. The present experiment used the pairedassociate probe technique to examine the nature of retroactive interference in three different adult age groups. Since some memory deficits occur at age 40 (Hultsch, 1971) and other effects seem not to occur until age 60 (Fozard, Nuttall, & Waugh, 1972), three age groups were included in the present design. The procedure also allowed the experimental separation of short-term memory and longterm memory by the factorial combination of the input and output positions of the items (Glanzer, 1971). The recency items can be looked at separately and compared to the other input positions in the list.
321
SMITH
322
Table 2. Mean Performance of Three Age Groups. Mean Mean Digit Age Vocabulary Span Recall Age Range Group 1 31.8 7.6 4.38 27.9 20-39 31.2 7.2 3.99 2 48.9 40-59 32.7 7.3 2.87 68.5 60-80 3 p=.7O Ftest P
Aging and Interference with Memory1 Anderson D. Smith, PhD2
adults seem to have poorer memories than younger adults. Initially, it was assumed that the memory deficit in old age was the result of short-term memory failure (Welford, 1958). This conclusion was based on the observation that older persons failed to remember recently acquired information while remaining proficient in the recall of events occurring many years in their past. Upon close examination, however, older persons only have problems with memory tasks which exceed the immediate memory span, and the immediate memory span serves as a good empirical estimate of short-term memory capability (Talland, 1968). In addition, quantitative estimates of short-term memory (Craik, 1968) are relatively invariant across the age span. Estimates of long-term memory, on the other hand, do show marked and systematic decreases from young to old. The long-term memory deficit, seen with tasks which exceed the immediate memory span, has been replicated many times in recent years (Botwinick, 1973; Hultsch, 1971). Early investigators suggested that the aged might be more susceptible to interference than younger subjects (Welford, 1958). An early experiment by Cameron (1943) supported this position. Cameron compared groups of older and younger subjects in remembering a series of digits. The difference between the age 1 This investigation was supported by a NIH Research Grant No. HD06885-01 from NICHD. 2 School of Psychology, Georgia Institute of Technology, Atlanta 30332.
groups was greater when an interpolated task, presumably a source of interference, was included between the presentation and recall of the digits. Because of such evidence, Welford (1958) concluded that the aged had an . . . increased liability to interference . . . which could account for many of the cases where true learning or retention deficiencies can be found.
Recent investigators have focused on the specification of interference effects in the aged. Interference can be defined in terms of either proaction (i.e., interference from previously learned material), or retroaction (i.e., interference from tasks interpolated between the input and recall of the to-be-remembered items). There seems to be no support for an increased effect of proactive interference with old age. In one experiment, Fozard and Waugh (1969) presented two three-item lists to groups differing in age and found no interaction between age and list order even though performance for all groups declined on the second list. No evidence of proactive interference was found in other experiments involving successive colornaming tasks (Craik, 1968), multiple free recall tasks (Raymond, 1971), or list learning tasks (Talland, 1968). The relationship between age and retroactive interference is more equivocal. Cameron (1943), cited earlier, found an interaction between retroaction and age, but later investigators have failed to replicate these findings. Gladis and Braun (1958), for example, found 319
Downloaded from http://geronj.oxfordjournals.org/ at James Cook University on March 17, 2015
Input and output interference in both short-term and long-term memory were measured in three age groups (aged 20-39, 40-59, 60-80 years). A paired-associate probe technique was used, which, by factorially combining the positions of the pairs at presentation and recall, allowed an examination of output interference for all positions in the input list. No differential output interference or input interference due to age was found for items assumed to be recalled from long-term memory. Evidence, however, did indicate that adult aging primarily affects long-term memory, since no differences between groups were found for items assumed to be recalled from short-term memory.
320
SMITH
Downloaded from http://geronj.oxfordjournals.org/ at James Cook University on March 17, 2015
found only with items from the last half of the input list. Because of the immediate recall for these items, it is possible that the effect reflected the nature of short-term memory since the last half of the input list contained the recency items. Recency items are the last few items seen, and if recall for these items is immediate (with short retention intervals), the level of recall is high because of their availability in short-term memory. If recall is delayed later in the output sequence, however, the level of recall is reduced as the recency items in short-term memory are displaced by other materials. The Smith (1974) experiment, on the other hand, used a long-term memory paradigm since the number of items to be remembered exceeded the immediate memory span and since input position was counterbalanced so that recency would not influence the data. In this experiment, no differential interference effects were found. This suggests that the phenomenon of output interference might depend on the relative influence of short-term and long-term recall on the data. To clarify the relationship between age and output interference, a closer examination of input position effects, together with an empirical separation of short-term memory and long-term memory seems necessary. The paired-associate probe technique (Tulving & Arbuckle, 1966) has been used to examine output interference with college-aged subjects and, at the same time, to examine short-term and long-term recall separately. Using only colWith categorized materials, another experi- lege-aged subjects, Arbuckle (1967) presented ment failed to find differential output inter- ten paired-associate pairs at input and cued or ference due to age (Smith, 1974). While the probed at recall with each stimulus item in number of words recalled declined across the sequence. The input and output positions of category output sequence for three different each pair were counterbalanced to allow an adult age groups, no age-related differences in examination of each input-output combination. the slopes of the interference functions were Output interference was found only for the last items in the input list. These items are recency found. items and are assumed to be recalled from One possible explanation of the conflicting short-term memory. Initial and middle items results is the differential influence of short-term in the list were unaffected by the order of outmemory in the two procedures. Recall is put. These results were interpreted that output usually inferred to> involve short-term memory interference has an effect only in short-term when it is immediate (with short retention recall. intervals) and when the amount of information At the end of presentation, the last few items to be retained is small (within the immediate memory span). The Taub experiments typ- are stored in short-term memory. If the output ically used short-term memory paradigms such of these recency items occurs immediately after as reproducing the order of letter sequences. presentation, recall is high. If, however, the In addition, output interference was often recall of other items is interpolated between the no age-correlated retroactive interference effects when the age groups were equated on the basis of vocabulary and learning ability. The interpolated activity between presentation and recall necessary to produce retroactive interference can be further divided into the presentation of further material (input) and the recall of other information from memory during the interval (output). Tulving and Arbuckle (1966) refer to these effects as input interference and output interference. While this distinction is important to clarify the nature of possible age interactions with retroactive interference, few experiments have been conducted which consider the effects separately. One group of experiments conducted by Taub and his associates does suggest that aged individuals are less able than young subjects to maintain items in memory while simultaneously making a response, i.e., output interference (Taub, 1968; Taub & Grieff, 1967; Taub & Walker, 1970). In these experiments, a comparison was made between recall in the first half and second half of the recall sequence. The results suggested that older persons are more susceptible to output interference than younger subjects. No age effects were found in the first half of the recall sequence while effects were found for the second half. The comparison, however, between the two' halves of the recall order does not allow a precise measure of the nature of output interference. A better design would separate the recall sequence into smaller portions.
AGING AND INTERFERENCE WITH MEMORY
the subjects. One of the categories was randomly selected and served as a practice list to insure that each subject understood the instructions and could adequately perform the task. The order of the remaining eight categories was systematically varied across subjects. A single random order of instances from each category was determined and paired with the stimulus items. The stimuli for all lists were the digits, 1-8, which always appeared in sequential order in each input list. The subject was placed in a separate room from the experimenter and equipment. The subject room was separated from the equipment room by a translucent screen. A Kodak Ektagraphic projector was used to present the materials on the screen. The sequence of the experiment and the instructions were controlled by a magnetic tape recording.
METHOD
Subjects Seventy-two male alumni of the Georgia Institute of Technology participated in the experiment. The subjects were divided into three groups according to age: Group 1 included the ages from 20-39 years; Group 2 included the ages 40-59; and Group 3 included the ages 60-80. The subjects were solicited by mail, and volunteers were paid a small sum for participation. All subjects resided in the metropolitan Atlanta area, representing similar socioeconomic and occupational backgrounds. To insure the comparability of the age groups, two subtests of the Wechsler Adult Intelligence Scale (Wechsler, 1955), the forward digit span and vocabulary subtests, were administered to all subjects prior to their participation in the experiment. Stimulus Materials and Apparatus Nine different categories were selected from the Battig and Montague (1969) category norms (e.g., countries), and the first eight words with response counts of 150 or less were chosen from each category (e.g., Mexico, Japan, Sweden, etc.). Words which had ambiguous category membership were eliminated and replaced by the next word from the norm listing. These words served as the responses for the nine paired-associate lists given to the subjects in the experiment. Different categories were used to insure easy list differentiation by
Procedure After standard paired-associate instructions and practice, each subject received a single trial on each of eight different experimental lists by the study-test method of paired association. The eight pairs in each list were first presented one-at-a-time followed by recall of the response members when the stimuli were presented in a predetermined sequence. The input sequence for the pairs in each list was determined by the sequential ordering of the digit stimuli. The output sequence of the pairs was determined from the rows of an 8 X 8 balanced Latin square as seen in Table 1. The rows of the table represent the eight lists, the columns represent the output positions of the pairs at the time of recall, and the numbers represent the positions of the specific pairs in the input sequence during presentation. The factorial design of the experiment insured that each input position was represented at each output position. Each subject received all eight lists, Table 1. Sequence of Input and Output. Output Sequence 7 4 6 8 5 1 4 6 5 3 7 2 4 5 7 6 8 3 6 8 5 1 7 4 1 7 8 6 2 5 2 8 1 7 3 1 3 2 4 6 8 4 7 2 3 5 1 5 4 8 3 2 6 "Numbers in table refer to the serial position of the pair in the input sequence. 1 1" 2 3 4 5 6 7 8
2 2 3 4 5 6 7 8 1
3 8 1 2 3 4 5 6 7
Downloaded from http://geronj.oxfordjournals.org/ at James Cook University on March 17, 2015
presentation of the recency items and their subsequent output, the probability of recall is as low as for other items in the list. The present experiment used the pairedassociate probe technique to examine the nature of retroactive interference in three different adult age groups. Since some memory deficits occur at age 40 (Hultsch, 1971) and other effects seem not to occur until age 60 (Fozard, Nuttall, & Waugh, 1972), three age groups were included in the present design. The procedure also allowed the experimental separation of short-term memory and longterm memory by the factorial combination of the input and output positions of the items (Glanzer, 1971). The recency items can be looked at separately and compared to the other input positions in the list.
321
SMITH
322
Table 2. Mean Performance of Three Age Groups. Mean Mean Digit Age Vocabulary Span Recall Age Range Group 1 31.8 7.6 4.38 27.9 20-39 31.2 7.2 3.99 2 48.9 40-59 32.7 7.3 2.87 68.5 60-80 3 p=.7O Ftest P
