| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||
RESEARCH ARTICLE |
a University of Warwick, England
Elizabeth A. Maylor, Department of Psychology, University of Warwick, Coventry CV4 7AL, England E-mail: elizabeth.maylor{at}warwick.ac.uk.
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionAppendix ENDIXReferences |
|---|
AT least since Proust 1981
, there has been a commonly held belief that tastes and odors can powerfully evoke seemingly forgotten memories (e.g., Laird 1935). More recent experimental evidence would appear to support such a view. For example, olfactory memories are not forgotten to the same extent as are auditory and visual memories (Engen, Kuisma, and Eimas 1973; Engen and Ross 1973; see Schab 1991, for a review). Moreover, olfactory stimuli can provide effective contextual cues in recalling or recognizing information encountered in the recent (e.g., Cann and Ross 1989) or distant (e.g., Aggleton and Waskett 1999) past. In this article, we introduce a novel paradigm to explore the effectiveness of olfactory cues in the retrieval of autobiographical memories associated with the cues. Our main aim was to compare olfactory cuing effects in young adults with those in a population known to have impaired sensory and memory abilities, namely, old adults.
One of the first studies to examine the contextual cuing effects of odors in the laboratory was by Cann and Ross 1989. Young adults viewed photographs of faces in the presence of an ambient odor. An old–new recognition test for the faces was presented 2 days later; on this occasion, either the same or a different ambient odor was present. Recognition performance was significantly higher when the same odor was present at both encoding and test than when a different odor was present (see Schab 1990; Smith, Standing, and De Man 1992, for similar results using different measures of memory).
A particularly striking example of context-dependent memory with odors is provided by a recent study by Aggleton and Waskett 1999. Young adults were asked to recall details of a visit to a museum that had taken place approximately 7 years previously. Memory for the exhibits in the museum was more accurate when participants were tested in the presence of the same distinctive odors as those present in the museum during their original visit, in comparison with novel odors or no odors. The authors concluded that the museum odors were effective retrieval cues for information acquired incidentally from a real-world episode.
Such findings are readily interpretable in terms of the encoding specificity principle (see Tulving 1983; Tulving and Thompson 1973), which states that the probability of retrieval varies directly with the compatibility of the stored information and the retrieval information. Thus, it is assumed that environmental or contextual features such as ambient odors are encoded together with target information during learning and that the presence of the same odor at testing functions as an additional retrieval cue to the target information (see Herz and Engen 1996, for a review).
To examine the nature, rather than the accuracy, of memories elicited by odors, Rubin, Groth, and Goldsmith 1984 presented young adults with the odors, names, or photographs of common items (e.g., coffee and onion). The task was to retrieve a single autobiographical memory associated with each cue, to date each memory, and to provide various ratings such as vividness and pleasantness. Memories produced in response to the (unnamed) odors were no less recent than memories evoked by names or photographs. However, memories evoked by odors were rated as being thought of and talked about less often prior to the experiment and were more likely to be rated as never having been thought of or talked about at all, in comparison with memories evoked by names or photographs. Using a similar procedure, Chu 1998 found that odors evoked older and more detailed autobiographical memories than words. Thus, consistent with the Proustian notion, odors appear to be particularly effective in acting as retrieval cues for previously inaccessible (and sometimes older) autobiographical memories.
Rubin and colleagues 1984 study forms part of a long tradition of research into autobiographical memory using the cue-word technique (Crovitz and Schiffman 1974; Galton 1879; Rubin 1982). In particular, this method has been applied to the study of age differences in retrieval from autobiographical memory (see Conway 1990; Rubin 2000, for reviews). Such work has revealed that the distribution of memories across the life span can be described in terms of three separate components: childhood amnesia, reminiscence, and retention (e.g., Rubin, Wetzler, and Nebes 1986). The retention component corresponds to the most recent 20 or so years of life where there is a monotonic decrease in the numbers of memories recalled from the very recent past to 20 years ago, which is qualitatively similar in young and older adults. The reminiscence bump is specific to middle-aged and older adults and refers to the increase in memories from when participants were aged between 10 and 30 years. Several theories have been proposed to explain the reminiscence bump, ranging from cognitive mechanisms to the formation of personal identity (e.g., Jansari and Parkin 1996; Rubin, Rahhal, and Poon 1998; Rubin and Schulkind 1997; Schrauf and Rubin 1998). Finally, both young and older adults display a paucity of memories from early childhood and infancy (see Conway 1990, for discussion).
A notable difference between the performance of young and older adults in studies of autobiographical memory using the cue-word technique that is particularly relevant to the present study is that older adults are slower to produce responses than young adults. For example, Rubin and Schulkind 1997 measured reaction times to produce a single memory to each cue word and found a significant age-related increase in reaction times. Similarly, Rabbitt and Winthorpe 1988 asked participants to produce as many autobiographical memories as possible in response to cue words over periods of 10 min. Older participants retrieved significantly fewer memories in the time available than young participants. To our knowledge, the question of whether autobiographical memory retrieval can be facilitated in older adults by providing additional cues such as odors has not been previously explored. (Facilitation here refers to an increase in the numbers of memories available in a limited time.)
In the present study, we adapted Rubin and colleagues 1984 procedure to compare the speeded production of autobiographical memories evoked by cue words (e.g., coffee) and by cue words accompanied by the appropriate odor (coffee plus coffee odor). In response to each cue, participants were asked to recall as many autobiographical memories as possible in 90 s. We conducted the experiment in two phases. In the first phase, we presented cue words alone; in the second phase, we presented the same cue words but half were accompanied by the appropriate odor. Participants were instructed to produce different memories in the two phases of the experiment. We therefore expected that fluency would decline from Phase 1 to Phase 2. However, on the basis of the literature reviewed above, we predicted that there would be less decline across the two phases for cue words accompanied by odors in Phase 2. Thus, we expected odors to facilitate the retrieval of autobiographical memories that were encoded in the presence of the odor.
As far as we are aware, previous studies of olfactory cuing effects have exclusively employed young adults as participants. Here, we address the question of whether olfactory cues are equally effective in old adults. There are at least two reasons for expecting reduced effects with aging. First, the sense of smell declines significantly with normal aging, with age-related impairments in olfactory sensitivity, odor discrimination, odor identification, odor recognition, and odor memory (e.g., Cain and Stevens 1989; Cain et al. 1995; Doty et al. 1984; Wysocki and Gilbert 1989; but see Rovee, Cohen, and Shlapack 1975, for no age differences in olfactory sensitivity); see Appendix, Note 1. Second, as already discussed, there is age-related decline in memory retrieval (see also Craik, Anderson, Kerr, and Li 1995), although it should be noted that some aspects of memory, such as priming and implicit memory, are relatively spared (see Light, Prull, La Voie, and Healy 2000, for a review).
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionAppendix ENDIXReferences |
|---|
Apparatus and Stimuli
A response booklet contained the instructions for Phase 1 of the experiment, response sheets for Phase 1, the instructions for Phase 2 of the experiment, response sheets for Phase 2, and a response sheet for an odor recognition test. The response sheets for Phases 1 and 2 each consisted of a series of numbered slots. The response sheet for the odor recognition test included a checklist of the six cue words used in the experiment and six numbered slots for responses.
A stimulus booklet contained the cue words (one per page), printed in upper case with a large bold typeface. We selected the cue words from a longer list of items used in a study of olfaction and autobiographical memory by Chu 1998. The cue words were ginger, coffee, whiskey, lemon, onion, and vinegar.
We prepared odors freshly every day by placing small amounts of each product in opaque plastic bottles and covering them with cotton wool and tight-fitting screw-top lids. For the "no odor" conditions, we placed water in the bottom of the bottles.
We used a stopwatch to control the timing of the experiment.
Design and Procedure
We assigned 28 young and 29 old participants to Condition A and the remaining 29 young and 28 old participants to Condition B. For both age groups, there were approximately equal numbers of male and female participants in each condition. We presented cue words in the same order (i.e., ginger, coffee, whiskey, lemon, onion, and vinegar) in both phases of the experiment. Conditions A and B were identical in Phase 1 of the experiment. In Phase 2 of the experiment, the first, third, and fifth cue words (i.e., ginger, whiskey, and onion) were accompanied by their corresponding odors in Condition A, whereas the second, fourth, and sixth cue words (i.e., coffee, lemon, and vinegar) were accompanied by their corresponding odors in Condition B.
We tested participants individually in quiet rooms free of any obvious odors. The first page of the response booklet provided the instructions for Phase 1 of the experiment. We informed participants that they would be presented with a word and that their task was to recall as many of their own personal memories associated with that word as possible in 90 s. Participants had to write down each memory as quickly as possible next to a number on the response sheet. We noted that we did not need to understand the memory description. We asked participants to summarize each memory in no more than two or three words (e.g., "Mum's garden") and to avoid statements or associations that were not their own memories (e.g., "Roses are pink"). We provided further explanation and examples as necessary until we were satisfied that the participant had fully understood what was required.
When participants were ready to proceed, we turned to the first response sheet in the response booklet and to the first cue word in the stimulus booklet and instructed participants to begin. After 90 s, we asked them to stop although they were allowed to finish the response they were writing. We then turned the page in both the response and stimulus booklets for the next item, and so on for each of the six cue words.
At the end of Phase 1, we gave participants a short break of 2 min. We then asked them to read the instructions for Phase 2 of the experiment. These explained that words would again be presented but this time participants would also be required to smell an odor from a small bottle at the same time as recalling memories. We warned them that they might be unable to smell some of the odors but that they should continue with the experiment nevertheless. We instructed participants not to repeat any of the memories that they had already recalled in the first phase of the experiment.
The procedure for Phase 2 was the same as for Phase 1 except that the appropriate bottle was unscrewed and presented to participants before beginning each 90-s period. For participants in Condition A, the first, third, and fifth bottles contained the appropriate odor, whereas the second, fourth, and sixth bottles contained no odor; these assignments were reversed for participants in Condition B.
Approximately one third of the old participants reported difficulty in writing quickly and we, therefore, allowed them to report their memories verbally for us to write down on the response sheet. (Two young participants also gave verbal responses because of arm injuries.) We did not regard this difference in procedure as significant for two reasons. First, it should have counteracted, to some extent, the known age decline in writing speed (Birren and Botwinick 1951). Second, the main comparison of interest (i.e., odor vs. no odor) was a within-participants manipulation.
We gave participants another short break before the final test of odor recognition. We presented bottles containing the six experimental odors in a random order; we asked participants to identify each odor, using the checklist provided. We then thanked and fully debriefed participants as to the design and purpose of the experiment. During informal conversation after the experiment, we encouraged participants to describe some of their reported memories in more detail. From this, we were able to verify that participants had complied with the instructions to report their own specific memories. Testing sessions lasted approximately 40 min.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionAppendix ENDIXReferences |
|---|
Overall Analysis
If participants retrieved the same memory in Phases 1 and 2, we discounted the second occurrence. Such repetition errors were relatively infrequent in both age groups, and we, therefore, did not analyze them further. First, we analyzed the total numbers of memories retrieved in response to the three cue words that were accompanied by odors in Phase 2, and to the three cue words accompanied by no odors in Phase 2. The means are presented in Fig. 1 as a function of phase and age group. It is clear that the young group retrieved more memories than the old group. Not surprisingly, the odor and no odor conditions were very similar in Phase 1, reflecting the fact that the two sets of three items were counterbalanced across the different conditions. Memory retrieval was expected to decline from Phase 1 to Phase 2 because we did not allow participants to repeat memories from the earlier phase. This was indeed the case for cue words accompanied by no odor. However, there was no decline from Phase 1 to Phase 2 for cue words accompanied by the odor, thus providing evidence that olfactory cues can facilitate memory retrieval. The difference between the no odor and odor conditions in Phase 2 was larger for the young group (an increase of 5.7 memories) than for the old group (an increase of 3.5 memories), possibly suggesting that olfactory cues are less effective in old age.
|
Although the three-way interaction was significant, indicating a smaller absolute effect of odor in the old than in the young, it is apparent from Fig. 1 that the relative effects were quite similar. Thus, the additional odor cue in Phase 2 increased the numbers of memories retrieved by approximately the same proportion in the old as in the young. A more appropriate measure of olfactory cuing that takes account of overall group differences in memory retrieval is therefore the effect of an odor in Phase 2 relative to baseline for those items in Phase 1 (i.e., Odor [Phase 2]/Odor [Phase 1]) divided by the effect of no odor in Phase 2 relative to baseline for those items in Phase 1 (i.e., No odor [Phase 2]/No odor [Phase 1]); see Appendix, Note 3. For the young group, this relative olfactory cuing effect, calculated from the overall means in Fig. 1, is (10.44/9.05)/(4.75/9.03) = 2.19. For the old group, the corresponding effect is (6.60/6.67)/(3.10/6.82) = 2.18. In other words, although olfactory cuing was significantly greater for the young than for the old in absolute terms, olfactory cuing was almost identical for the young and old in relative terms. Thus, for both age groups, just over twice as many memories were retrieved with the odor than without the odor.
The relative olfactory cuing effect as defined above was calculated for each individual participant in order to perform a statistical comparison between the two age groups. The effects were slightly larger than those based on the overall group means, but there was no significant difference between the young (M = 2.94, SD = 2.20) and the old (M = 2.77, SD = 1.74), t(108) = 0.44, p = .66 (see Appendix, Note 4). In addition, the correlation between exact age and olfactory cuing for the old group alone (aged 69–98 years) was close to zero, r(51) = -.05, p = .71. Together, these results confirm that olfactory cuing, expressed as the relative facilitation of memory retrieval by the presence of an odor, is not reduced in old age.
Analysis Across Items
Fig. 2 shows the relative olfactory cuing effects for the six cue words calculated from the overall means for each age group. Regardless of the particular cue word, approximately twice as many memories were retrieved with the odor than without the odor. For coffee, lemon, and vinegar, the olfactory cuing effects were almost identical for the two age groups; for ginger and whiskey, the olfactory cuing effects were slightly greater for the young than for the old; and for onion, the olfactory cuing effect was greater for the old than for the young. Thus, with the possible exception of onion, it appears that there were no marked differences between the six different items in terms of their olfactory cuing effects and no striking age differences (see Appendix, Note 5).
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionAppendix ENDIXReferences |
|---|
) and theories of context-dependent memory (see Herz and Engen 1996, for a summary). Contrary to expectations, the olfactory cuing effect is almost identical in magnitude across adult groups with mean ages of 21 and 84 years.
First, it should be noted that the old adults produce significantly fewer memories overall than the young adults, as expected from the literature on normal aging and memory retrieval (e.g., Craik et al. 1995). Moreover, the age deficit is comparable with that observed by Rabbitt and Winthorpe 1988 and also in other studies from our laboratory that required participants to travel independently to testing sessions at the University of Warwick. For example, in a verbal fluency task, old adults with a mean age of 83 years retrieved 14.7 words beginning with the letter "s" in 4 min, whereas young adults with a mean age of 20 years retrieved 20.3 words—an old to young proportion of 0.72. This compares with the present data in which the total numbers of memories retrieved in Phase 1 are 13.5 and 18.1 for old and young, respectively—a proportion of 0.75. In other words, there is little indication that the present method of recruitment leads to a more cognitively impaired sample of old adults than is typically the case in aging studies. However, the observed overall age deficit is useful in ruling out one possible explanation for the preserved olfactory cuing effect, namely, that the old group is unusually able. Note also that the experiment is sufficiently powerful to reveal a significant age difference in olfactory cuing when analyzed in terms of absolute effects.
We suggest that the absence of an age-related decline in olfactory cuing (as measured by the relative benefits from odors) is consistent with several recent studies demonstrating that although older adults are particularly impaired in their recall of contextual information (see Spencer and Raz 1995, for a summary), they can be unimpaired in their use of contextual information to facilitate memory. For example, Light, La Voie, Valencia-Laver, Albertson Owens, and Mead 1992 observed similar modality effects in young and older adults in perceptual identification tasks such that both age groups showed greater repetition priming when study and test modalities matched than when they differed. The authors concluded that contextual information was equally available to both age groups on this indirect test of memory, whereas older adults were significantly worse than young adults in explicitly retrieving modality information (direct test). Similarly, Naveh-Benjamin and Craik 1995 found that older adults benefited at least to the same extent as young adults when the perceptual-contextual characteristics of stimulus words (i.e., the font in which they appeared, or the voice through which they were spoken) were reinstated at testing, although older adults were impaired on an explicit memory test for these characteristics. Finally, Schramke and Bauer 1997 compared state-dependent learning in younger and older adults by requiring participants to rest or exercise prior to encoding and retrieval. In both age groups, recall was significantly greater when the activities prior to encoding and retrieval matched than when they differed, again suggesting that older adults can spontaneously use contextual information (in this case, the participants' physiological state) to aid recall. The present study, therefore, extends these findings by demonstrating for the first time that aging does not impair the use of olfactory contextual information in facilitating memory retrieval.
Finally, there remains the question of whether odors are particularly special (see Herz and Engen 1996; Schab and Crowder 1995, for discussion), or whether the present findings could be extended to other ambient cues. For example, it may be easier to recall the details of a summer holiday when tested in a warm room than in a cold room. The present findings suggest that old and young adults should benefit equally, at least in proportional terms, from such context reinstatement.
|
Acknowledgments |
|---|
|
Appendix ENDIX |
|---|
|
TOPAbstractMethodsResultsDiscussionAppendix ENDIXReferences |
|---|
2. An initial four-way ANOVA included gender as an additional between-participants variable. There were no significant effects or interactions involving gender; it was therefore dropped from subsequent analyses.
3. Because of the overall similarity between odor and no odor conditions in Phase 1, this could be simplified to Odor (Phase 2)/No odor (Phase 2). However, the original formula is preferred, particularly for statistical analysis, because odor and no odor conditions in Phase 1 involved different cue words for individual participants and, therefore, baselines were not necessarily similar.
4. Four old participants could not be included in this analysis because they retrieved no memories in the no odor condition in Phase 2.
5. Two separate ANOVAs were conducted on items 1, 3, and 5 (ginger, whiskey, and onion), and items 2, 4, and 6 (coffee, lemon, and vinegar), with A/B group (Condition A vs. Condition B) and age group as between-participants variables, and phase and item as within-participants variables. In both cases, there was highly significant olfactory cuing as indicated by interactions between A/B group and phase. There were no significant interactions involving either age group or item.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionAppendix ENDIXReferences |
|---|
| ||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
