| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
a Department of Psychology, Uppsala University, Sweden
b Clinical Neuroscience (NEUROTEC) and
c Medical Epidemiology, Karolinska Institutet, Stockholm, Sweden
d Division of Social Sciences, Indiana University, New Albany
e Department of Psychology, University of Southern California, Los Angeles
Maria Larsson, Stockholm Gerontology Research Center, Box 6401, S-113 Stockholm, Sweden E-mail: maria.larsson{at}neurotec.ki.se.
Toni C. Antonucci, PhD
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
aged 45–87 years from the Swedish Adoption/Twin Study of Aging were assessed with a Swedish version of the National Geographic Smell Survey. Both detection and identification of olfactory information were impaired with age. Gender had no effect on detection or identification. Hierarchical regressions revealed that proficiency in semantic memory, intensity perception, and personality style (i.e., neuroticism, impulsivity, and lack of assertiveness) were potent predictors for successful odor identification, even when individual variations in chronological age, sex, education, and global cognitive functioning were taken into account.
AGING is often accompanied by impairments in various sensory and cognitive aspects of olfactory functioning. Older adults exhibit a lower sensitivity for odors, as reflected in absolute threshold measurements (e.g., Cain and Gent 1991
; Murphy, Nunez, Withee, and Jalowayski 1985) and in intensity measures of suprathreshold odors (Stevens and Cain 1985). Likewise, older adults' recognition memory for odors is poorer than that observed in young adults (Larsson and Backman 1993; Murphy, Cain, Gilmore, and Skinner 1991), as is their ability to name or identify olfactory information. The latter is true for both free identification measures (Schemper, Voss, and Cain 1981) and tasks where multiple choices of possible odor names are available (Doty et al. 1984).
An important research issue has been to establish whether increased sensory problems, cognitive changes, or both cause the observed age-related deficits in odor recognition memory and odor identification. Recent research suggests that age-related deficits in both odor recognition and odor identification may be largely attributable to cognitive limitations (Larsson and Backman 1997; Murphy et al. 1991). For example, there is evidence that episodic odor memories are mediated by semantic factors (e.g., familiarity and odor identification) and that older adults' difficulties in identifying common olfactory information underlie the aging-related deficits in odor memory (Larsson and Backman 1993, Larsson and Backman 1997; Lehrner, Gluck, and Laska 1999).
Given that age deficits in odor naming are highly prevalent and that they also mediate age-related impairments in episodic odor memory, it is of interest to consider potential explanations for older adults' problems in identifying odors. Of particular interest is to examine the relationship between aging-related deficits in odor identification and performance in other aspects of cognitive functioning. By definition, odor identification is a semantic memory task, in that it refers to an individual's general knowledge or experience with a specific odorant (Schab 1991; Tulving 1993). It is therefore of interest to examine whether proficiency in odor identification relates to performance in other tasks that tap semantic memory (e.g., vocabulary and information). Of further interest is to determine whether odor identification is related to performance in other cognitive domains, such as performance in short-term memory, episodic memory, and visuospatial functioning. Finding reliable relationships between measures of semantic memory and odor identification, and nonsignificant relationships between measures of other cognitive functions and odor identification, would strengthen the hypothesis that complex verbal intellectual abilities and odor identification tap the same cognitive domain.
In addition to cognitive variables, research suggests a possible role for personality measures in the prediction of odor performance. Perceptual abilities are associated with certain personality traits such as extraversion (Stelmack and Michaud-Achorn 1985). Little is known as to whether various personality styles influence olfactory functions, and the sparse evidence presents a mixed pattern of results. In an early study, Koelega 1970 reported that participants who scored high in extraversion also exhibited a higher olfactory sensitivity, whereas neuroticism was unrelated to odor perception. In a related study, which focused on subjective experiences (i.e., intensity, pleasantness, and familiarity) of olfactory information, no relationship between perceptual experience and degree of extraversion was found (Filsinger, Fabes, and Hughston 1987). In contrast, Pause, Ferstl, and Fehm-Wolfsdorf 1998 reported that neuroticism was a stronger predictor for olfactory sensitivity than was extraversion. This latter finding is in congruence with research focused on emotional (personal) styles and olfactory sensitivity. Olfactory sensitivity for octanol was examined in high- and low-anxiety women, and the results showed that highly anxious women had reliably higher thresholds than did women low in anxiety (Rovee, Harris, and Yopp 1973). In a similar vein, Herbener, Kagan, and Cohen 1989 reported that level of shyness (introversion) was related to odor thresholds; participants high in shyness had lower olfactory thresholds for butanol than participants with low shyness scores.
A number of studies have indicated that women perform better in olfactory tasks than do men and that the general female superiority is valid throughout the human adult life span (e.g., Dorries 1992; Doty et al. 1984; Ship and Weiffenbach 1993). However, Ship, Pearson, Cruise, Brant, and Metter 1996 found that men might show a more precipitous and earlier decline in smell identification than do women, implicating an age by gender interaction in olfactory performance. It is still unknown why women show this superiority, but gender-related differences in factors such as hormones (estrogens, progesterone), environmental background, and verbal fluency might play a role. Two of our purposes in the present study were to examine further whether women perform better than men in the second half of the life span and also to investigate the possibility of age by gender interactions in odor detection and odor identification.
In the present study we assessed a subsample of 532 participants from the Swedish Adoption/Twin Study of Aging (SATSA; Pedersen et al. 1991) with a Swedish version of the National Geographic Smell Survey (Wysocki and Gilbert 1989) with a particular focus on odor detection and odor identification abilities. All participants were also provided with a series of questionnaires concerning health, personality, and lifestyle variables (e.g., smoking behavior) and were also tested across a number of cognitive domains.
On the basis of these data, we examined: (a) the influence of age and gender on odor detection and odor identification, (b) the relationship between odor identification ability and performance in various cognitive tasks, (c) the influence of personality on olfactory functioning and (d) the relative predictive value of individual differences in demographic, cognitive, perceptual, and personality variables on odor identification.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
). The SATSA study (Pedersen et al. 1991) included both questionnaire and in-person testing assessments (in-person assessments also included a questionnaire). In the current analyses, we used data collected during the third wave of questionnaires (in 1990) and the second wave of in-person testing (1989–91). The in-person testing took place in a location convenient to the participants, such as district nurses' offices, health care schools, and long-term care clinics. The testing session also incorporated a health examination, and the visit lasted approximately 4 h.
The smell survey was mailed with the 1990 questionnaire to the subset of 606 participants who had completed the first wave of in-person testing (1986–88), and 532 (87.8%) completed surveys were returned. Because olfactory abilities are known to be severely affected by dementia (e.g., Larsson et al. 1999; Mesholam, Moberg, Mahr, and Doty 1998), participants suffering from Alzheimer's disease, vascular dementia, or unspecified dementia were excluded 
. Thus, the remaining sample involved 506 adults, ranging in age from 45 to 87 years 
, 57% of whom were female. Characteristics of the sample are provided in Table 1 . Education was rated on a 4-point scale (1 = completion of elementary school, 2 = completion of vocational high school, 3 = "gymnasium" or academic high school, 4 = university studies or higher). To assess global level of cognitive functioning and screen for severe cognitive impairment, all participants completed the Mini-Mental State Examination (MMSE; Folstein, Folstein, and McHugh 1975). We used the SUMILL scale, which surveys the number of organ systems reported to be affected by a chronic health problem, to measure general health. The scale incorporates measures of 13 organ systems, including cardiovascular, respiratory, musculoskeletal, and central nervous system disorders. An individual's score is the summed total of organ systems reported to be affected by at least one chronic health problem (Harris, Pedersen, McClearn, Plomin, and Nesselroade 1992). SUMILL ranged from 0 to 9 in this sample with an average value of about 3. Use of cigarettes, cigars, and pipes was included in the calculation of pack-years (Kendler, Karkowski, and Pedersen in press). Cigar and pipe tobacco consumptions were converted to cigarette equivalents (e.g., one cigar was equivalent to four cigarettes).
|
; Russell et al. 1993; Wysocki and Gilbert 1989). In the Swedish version, a translation of the questions was attached to the original National Geographic scratch-and-sniff forms. The smell survey comprised six scratch-and-sniff microencapsulated odorants: androstenone, amyl acetate, GalaxolideTM, eugenol, mercaptans, and rose. Two of these odorants are food related: amyl acetate, which has a banana- or pear-like fruity odor, and eugenol, a major constituent of clove oil. Mercaptans is a highly unpleasant sulfurous compound that is added to natural gas as a warning odor. Synthetic rose is a pleasant, familiar non food odor. GalaxolideTM is a widely used synthetic musk that is often added to perfumes. Finally, the compound androstenone is a volatile steroid metabolite produced by many mammals, including humans. Men produce more androstenone than women do, but the compound may be extracted from the urine and sweat of both sexes.
Respondents were asked to scratch and sniff each odor panel and to answer a number of questions, including: "Did you smell something (yes or no)?" "Did it smell good or bad 
?" "How intense is this odor 
?" For odor identification, respondents were instructed to select one of the following alternatives: floral, musky, urine, foul, ink, spicy, woody, fruity, burnt, sweet, other (if other: What?). Survey designers chose the descriptors to provide generic, nonoverlapping designations of odor quality (Russell et al. 1993; Wysocki and Gilbert 1989). In the present study, we used the standards provided by Wysocki and Gilbert 1989 to score odor identification. If a respondent provided a veridical label of the odor, the response was counted as correct. Also, if a respondent chose a near-miss label such as banana candy for amyl acetate, it was coded as correct. If a respondent detected an odorant but was unable to identify it, then the identification response was scored incorrect.
For the analyses, data were summarized across odorants and four measures were obtained: number of odorants detected (odor detection), number of odorants correctly identified (odor identification), mean pleasantness rating, and mean intensity rating.
Cognitive measures.
The cognitive battery in SATSA was selected to provide representation of both the domains of fluid (figure logic, Koh's block design, card rotations) and crystallized (information, synonyms, analogies, Thurstone's memory) intelligence (Horn 1982). In addition, the battery included measures tapping specific cognitive domains such as short-term memory (digit span forward and backward), perceptual speed (digit symbol, figure identification), and episodic memory (names and faces). A more detailed description of the cognitive tests included in SATSA is available in Pedersen, Plomin, Nesselroade, and McClearn 1992.
Personality tests.
Five scales taken from different personality inventories were used in the SATSA battery: extraversion, neuroticism, lack of assertiveness, impulsivity, and openness to experience. The personality traits extraversion and neuroticism were measured with a short form of the Eysenck Personality Inventory (Floderus 1974). Each scale score is based on the sum of 0 (no) and 1 (yes) responses to nine items. Two subscales from the Karolinska Scales of Personality were also included in the SATSA battery: the Lack of Assertiveness scale and the Impulsivity scale (Schalling 1986; Schalling, Edman, and Asberg 1983). The Lack of Assertiveness scale is composed of 10 items (e.g., "If I'm treated badly at a restaurant I don't like to complain" and "When someone teases me I can never think of a good answer until afterwards"). The Impulsivity scale score consists of the sum of responses to each of 10 items on a 5-point Likert scale. The impulsivity items mainly reflect acting on the spur of the moment and rapid decision making (e.g., "I often rush into new things" or "When I make decisions I usually do it quickly"). Finally, a shortened version of the Openness to Experience scale (25 of the original 48 items) from the NEO Personality Inventory was used (Costa and McCrae 1985; Leong and Dollinger 1990). The Openness domain taps proactive seeking and appreciation of experience for its own sake, on the basis of characteristics such as openness to feelings, new ideas, flexibility of thought, and readiness to indulge in fantasy (McCrae and Costa 1985).
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
and old adults 
. The total number of odors detected and the total number of correct identifications for each individual were analyzed with separate 2 (Age: young-old, old) x 2 (Gender: male, female) analyses of variance.
The older age group detected fewer of the odorants, 
, and gender had no effect on odor detection (p > .40). The interaction between age and gender was not significant (F < 1). The oldest participants identified fewer of the odorants correctly, 
, and men and women did not differ in identification performance (p > .10). The interaction between age and gender also was not significant (F < 1).
Separate correlations between age and odor detection for each odor, with gender and pack-years partialed out, indicated that increasing age was associated with an impaired ability to detect mercaptans 
-
and GalaxolideTM 
. No other correlations were significant. There were two significant correlations between age and odor identification across odor stimuli, suggesting that aging may be associated with a better ability to identify the smell of amyl acetate 
and lesser ability to identify mercaptans or gas 
.
Diseases and Olfactory Performance
To investigate the relationship between a number of diseases and performance in the olfactory tasks, we performed a series of correlations with age, sex, and pack-years partialed out. The general health measure—SUMILL—was unrelated to both odor detection and odor identification. Separate analyses for a number of diseases (i.e., central nervous system (CNS) disorders, diabetes, epilepsy, metabolic disorders, respiratory disorders, and stroke) revealed that only CNS disorders were reliably related to impairments in odor detection ( p < .05) and that epilepsy was associated with an impaired ability to identify olfactory information (p < .05), which is consistent with earlier observations (Savic, Bookheimer, Fried, and Engel 1997).
Cognitive Functioning and Olfactory Performance
The relationships between performance in the cognitive measures and the olfactory tasks, partialed by pack-years, are displayed in Table 2 . None of the cognitive tests were related to odor detection performance. However, odor identification was reliably and positively related to performance in tasks tapping crystallized intelligence and semantic memory functions (i.e., analogies, synonyms, information, Thurstone's memory), and one measure of perceptual speed (i.e., digit symbol). No significant relationships were observed for any of the measures of fluid intelligence, short-term memory, and episodic memory.
|
|
|
The shared variance between odor identification and the semantic memory measures suggests that these measures tap the same cognitive domain. Also, all beta weights, except the one for synonyms, were positive, indicating that higher proficiency in semantic memory tasks was related to better performance in odor identification.
Perceived pleasantness of an odorant was unrelated to identification, whereas the degree of perceived intensity was strongly and positively related to identification performance 
. Interestingly, personality traits contributed significantly to the variance 
despite the previous inclusion of demographic and cognitive variables. Neuroticism, impulsivity, and lack of assertiveness were reliable predictors for odor identification, whereas extraversion and openness to experience were not.
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; Schiffman 1997). An important finding is the reliable positive relationship between general semantic knowledge and proficiency in odor identification. The observation that performance in fluid intelligence, short-term memory, and episodic memory was unrelated to odor identification provides a strengthening of the view that odor identification and semantic memory proficiency in general tap the same cognitive domain (Larsson 1997). This finding is highlighted by the fact that semantic memory proficiency still remained a significant predictor for odor identification performance, despite statistical control for the effects of chronological age, sex, education, and global cognitive functioning (according to the MMSE). These outcomes indicate the pivotal role played by semantic memory aptitude for successful odor identification. Also, none of the cognitive measures were related to proficiency in odor detection, suggesting that the two olfactory tasks measure two separate entities.
In contrast to earlier findings of gender differences in olfactory sensitivity and identification (e.g., Barber 1997; Doty et al. 1984; Lehrner 1993), our results reveal no reliable influence of gender on the two olfactory measures, although women were slightly better at odor identification. Furthermore, contrary to the findings reported by Ship and colleagues 1996, we did not find any evidence of age by gender interactions in odor detection or odor identification abilities. However, the observations by Ship were based on longitudinal data, whereas our findings were cross-sectional, making it impossible to disentangle true aging effects from cohort membership.
One factor that has been proven to influence olfactory functioning is health status (Schiffman 1992). Our results suggest that general health, as measured by SUMILL, is unrelated to both olfactory measures. However, separate analyses for the different diseases showed that CNS disorders affected odor detection negatively and that participants suffering from epilepsy exhibited greater difficulties in identifying the odorants, in accordance with a number of earlier reports (e.g., Doty 1991; Savic et al. 1997). A related question that needs to be further explored in future research concerns pharmaceutical use and its potential impact on olfactory functioning.
Age deficits in olfactory abilities were not uniform across odorants. Available knowledge on heterogeneous age effects for different odors is sparse. However, Wysocki and Gilbert 1989 reported that the shape of the age-response curve in both odor detection and identification varied across different odorants. Our data suggest that the ability to identify amyl acetate improves with age; however, this finding is in contrast with earlier evidence (Wysocki and Gilbert 1989), and additional work is needed to clarify the validity of this finding. Nevertheless, replicating earlier findings, we found evidence that age takes a particular toll in the ability both to detect and to identify the smell of mercaptans, the warning agent in natural gas (Cain, Gent, & Cometto-Muniz, 1993; Wysocki and Gilbert 1989). This is in congruence with recent neurological data showing that brain responses for natural gas odor, as measured by olfactory event-related potentials, change across the life span. Specifically, older adults display lower N1/P2 amplitudes, which suggest a diminished sensitivity, and increased P3 latencies, which suggest a slower cognitive processing for gas odor (Madowitz and Geisler 1998). Failure to perceive and to identify the smell of mercaptans may be a potential public safety problem, and the present data indicate that older householders are selectively affected.
As noted earlier, not much is known concerning relationships between personality factors and olfactory abilities (Pause et al. 1998). Surprisingly, odor identification performance was associated with personality dimensions such as neuroticism, impulsivity, and lack of assertiveness. Controlling for individual variation in demographic and cognitive abilities did not negate the impact of these personality factors on odor identification.
Participants with high scores in neuroticism (emotionality) also identified more odors correctly. Olfactory neuroimaging research has revealed that olfactory information processing relies heavily on neuroanatomical structures within the limbic system, such as the amygdala, enthorinal cortex, and insula (Levy et al. 1997; Zald, Donndelinger, and Pardo 1998). In a similar vein, it has been proposed that emotionally highly reactive individuals show a higher activation within the limbic system and that anxious individuals show higher reactivity of the Behavioral Inhibition System, which relates to limbic structures (Eysenck and Eysenck 1985; Gray 1990). Given the strong dependency between limbic structures and olfactory functions, it is not surprising that individuals high in emotional reactivity also would excel in olfactory ability (Pause et al. 1998).
The predictive power of personality styles such as impulsivity and lack of assertiveness on odor identification are intriguing. Both dimensions were negatively related to odor identification. More impulsive individuals may be more negligent in the task situation, whereas persons exhibiting lack of assertiveness may be indecisive in answering, which ultimately may affect their level of performance. However, full understanding of these findings remains unclear and needs further exploration.
The strong influence of intensity perception on odor identification (Stevens and Cain 1985) is of note. Perceived intensity played a major role in predicting odor naming, even when age-related variation was accounted for. This outcome suggests that individual variation in intensity perception is a major constituent for successful odor identification and should be taken into account in studies evaluating proficiency in odor identification.
In summary, this work has replicated and extended the understanding of age-related deficits in odor identification. In agreement with earlier findings, the present results indicate an age-related deterioration in both odor detection and odor identification. However, in contrast to odor identification, odor detection was unrelated to cognitive parameters and personality traits. Four potent factors for successful odor identification were identified: age, semantic memory aptitude, intensity perception, and personality style. The fact that proficiency in odor identification was unrelated to fluid intelligence, short-term memory, and episodic memory strengthens the notion that semantic memory (i.e., crystallized intelligence) and odor identification tap the same cognitive domain.
|
Acknowledgments |
|---|
Received for publication August 16, 1999. Accepted for publication February 2, 2000.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  D. Finkel, N. L. Pedersen, and M. Larsson Olfactory Functioning and Cognitive Abilities: A Twin Study J. Gerontol. B. Psychol. Sci. Soc. Sci., July 1, 2001; 56(4): P226 - 233. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
