Chem. Senses 28: 523-526,
2003
© Oxford University Press 2003
Influences of Feedback and Ascending and Descending Trial Presentations on Perithreshold Odor Detection Performance
Smell and Taste Center and Department of Otorhinolaryngology: Head and Neck Surgery, University of Pennsylvania Medical Center, Philadelphia, PA 19104, USA
Correspondence to be sent to: Richard L. Doty, Director, Smell and Taste Center, University of Pennsylvania Medical Center, 5 Ravdin Building, 3400 Spruce Street, Philadelphia, PA 19104, USA. e-mail: doty{at}mail.med.upenn.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
The influences of feedback and ascending and descending trial sequences on the ability of 135 college-aged subjects to detect phenyl ethyl alcohol odorant concentrations ranging from 10–9 to 10–5.5 v/v were examined in a two-alternative forced-choice test paradigm. At the highest concentrations, ascending trial sequences produced better performance than descending trial sequences; the reverse was true at the lowest concentrations. There was a tendency for feedback to improve performance marginally at the lowest two odorant concentrations presented. In the region associated with a traditional detection threshold calculation (i.e. at the 75% performance point in a two-choice detection task), no influences of feedback or direction of trial sequence were apparent. These data indicate that the effects of explicit feedback and trial sequence direction depend upon the segment of the peri-threshold stimulus concentration continuum evaluated.
Key words: threshold, phenyl ethyl alcohol, psychophysics, feedback, method of limits
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
A popular means for assessing olfactory function is to measure the lowest odorant concentration that is detectable to a subject (i.e. the so-called detection threshold). A number of methods are currently used to provide such a measure, including the method of limits, various staircase procedures and, on rare occasions, the classical method of constant stimuli [for review, see (Doty and Laing, 2003
)]. In
general, methods that initially employ ascending presentations of odorant
concentrations are more common, since they are presumed to evoke less
adaptation than procedures that employ descending stimulus presentations and,
consequently, produce lower and possibly more reliable threshold values.
Empirical data in support of such a notion, however, is largely lacking. In
the only study directly assessing this issue in the chemical senses, Pangborn
et al. (Pangborn et al.,
1964) reported lower thresholds for ascending than for descending
trials. However, this work employed only five subjects and a non-forced choice
trials paradigm, and no statistical analysis was performed on the data. In
contrast, there is limited evidence against the notion of superior performance
on tasks employing mainly ascending trials and two-alternative forced-choice
responses. Thus, greater detection performance in two-alternative
forced-choice staircase detection threshold tasks occurs after long descending
trial runs (Doty, 1991), and
perithreshold detection performance seems to increase within trial blocks of
the same concentration in signal detection paradigms employing either humans
or rats (Doty et al.,
1981; Doty and Ferguson-Segall,
1989). Hence, despite intuitive appeal, the question as to whether
ascending trials result in better threshold performance than descending trials
remains open.
The present study examined the influences of ascending and descending trial
sequences, as well as the influence of feedback, on the ability of subjects to
detect, in a two-alternative forced-choice paradigm, perithreshold
concentrations of the rose-like smelling odorant phenyl ethyl alcohol (PEA).
To our knowledge, the role of feedback in odor detection performance has never
been formally explored, although it is well established in animal chemosensory
psychophysical work that feedback decreases the time required to learn a task
(Dorries et al.,
1991). Phenyl ethyl alcohol was chosen as the stimulus because of
its wide use in olfactory psychophysics and the fact that, relative to most
odorous chemicals, it has comparatively low intranasal trigeminal reactivity
(Doty et al.,
1978).
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Subjects
One hundred thirty-five college-aged subjects [mean age (SD) = 20.74
(2.21); 71 men and 64 women] were recruited from advertisements placed on
campus bulletin boards at the University of Pennsylvania. A deliberate effort
was made to recruit minority students. All subjects were non-users of tobacco
products. The ethnic composition of the study group was as follows:
Afro-American, 12; Caucasian American, 100; Hispanic Americans, 3; Asian
American, 19. The ethnic background of one subject was not recorded. All
subjects reported being in good health and none reported having problems
smelling or tasting. None were taking medications at the time of testing.
Individuals who reported a history of nasal disorder or any other problem
(e.g. anosmia) that would interfere with their participation were not included
in the study group. The 12-item Brief Smell Identification Test (also termed
the Cross-Cultural Smell Identification Test or CC-SIT)
(Doty et al., 1996)
was administered to a subset of 56 of these individuals; all scored within the
normal range on this standardized test [mean (SD) = 11.26 (0.84)].
Procedures
The general experimental paradigm was straightforward. The subjects were assigned to the following experimental groups: Ascending Trial Sequence Group with Feedback (A-F) (n = 35); Ascending Trial Sequence Group without feedback (A-NF) (n = 34); Descending Trial Sequence Group with Feedback (D-F) (n = 31); and Descending Trial Sequence Group without Feedback (D-NF) (n = 35).
Fourteen two-alternative forced-choice trials were presented to each
subject at each of eight stimulus concentrations ranging in half-log steps
from 10–9.0 to –10–5.5 v/v. A trial
consisted of the bilateral presentation of two glass sniff bottles in rapid
succession to the subject. These 120 ml bottles were 8.5 cm tall with 3.8 cm
i.d. openings and 4.4 cm i.d. widths. One contained 20 ml of a given
concentration of PEA (Aldridge Chemical, Chicago, IL) dissolved in USP grade
light mineral oil; the other contained mineral oil diluent alone. The bottles
were opened and immediately placed over the subject's nose in a standardized
manner depicted elsewhere (Doty et
al., 1978). The subject's task was to report which of the two
bottles produced the stronger sensation. Even if no sensation was perceived or
if no difference was apparent between the bottles, the subject was required to
choose one or the other bottle (i.e. the task was forced-choice). The order of
the presentation of the two bottles of a trial was random, with the
stipulation that, in a block of 14 trials, the odorant was presented first on
half of the trials, and the diluent on the other half of the trials. A 15 s
interval was maintained between trials.
In the ascending trial sequence conditions, the trials began at the lowest concentration and all 14 trials at a given concentration were completed before moving to the next highest concentration. The reverse was true for the descending trial sequence condition. Under the feedback conditions, the experimenter verbally indicated to the subject whether his or her response on each trial was correct of not. Under the no-feedback conditions, no such indication was provided.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
To determine the influence of the independent measures of this study on detection performance, the percentage of correct trials was calculated for each subject at each odorant concentration level and subjected to a trial sequence (descending, ascending) x feedback condition (yes, no) x odorant concentration analysis of variance (ANOVA), with replications on the last factor. Neither of the between subjects factors—trial sequence and feedback condition— was statistically significant [respective Fs(1,131) = 0.64 and 0.00, respective Ps = 0.43 and 0.99). The trial sequence x feedback interaction was also not significant [F(1,131) = 0.05, P = 0.83). However, the within-subject factor of odorant concentration, as well as the odorant concentration x trial sequence interaction and the odorant concentration x feedback interaction, were significant [respective Fs(7,917) = 161.8, 3.87 and 2.37; respective Ps < 0.00001, 0.002 and 0.021] (Figures 1 and 2).
|
|
Assessment of differences at each odorant concentration level using one-way ANOVAs revealed significant differences only within the trial sequence group data, although, as seen in Figure 2, there was a non-significant tendency for the feedback group to be superior to the non-feedback group at the lowest two odorant concentrations. As shown in Figure 1, significantly greater performance was noted within the ascending trials group, relative to the descending trials group, at the higher odorant concentrations; the reverse was true at the lowest odorant concentration.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
The present study suggests that the direction of trial sequences—ascending or descending—has a clear influence on odor detection performance only at the extremes of the perithreshold stimulus concentration range. Specifically, better performance occurs at lower perithreshold odorant concentrations when the trials begin at the higher concentrations and subsequently descend to the lower concentrations. Conversely, better performance occurs at higher perithreshold odorant concentrations when trials begin at lower concentrations and ascend to higher concentrations. No meaningful influences of trial sequence on performance was observed for odorant concentrations midway in the concentration series—i.e. concentrations that would typically be used in the calculation of a threshold estimate.
The basis of the differential influences of ascending and descending trial blocks at the two ends of the perithreshold odorant concentration continuum is not immediately clear. If adaptation were a pervasive phenomenon, then one would expect performance to be poorer for subjects of the descending trial sequence at all odorant concentrations, with perhaps even greater attenuation occurring at the lower concentration levels. However, poorer performance in these subjects was only observed at the higher odorant concentrations; at the lower odorant concentrations, the performance of the subjects in the descending trial sequence group actually exceeded that of the ascending series group. Although it is possible that less adaptation occurred at the lower odorant concentrations (e.g. due to the relatively low concentrations of the more immediate prior trials and reversal of adaptation due to earlier exposure to higher concentrations), this alone would not explain the better performance in the descending than in the ascending trials group at the lowest concentration levels employed in this study. Moreover, it is unlikely that a simple warm-up effect is the primary basis of this phenomenon, since more than an adequate number of trials was presented at each of the initial concentrations in both the ascending and descending groups to overcome most such effects.
Perhaps the most parsimonious explanation of the present findings is that
subjects simply improve performance over time in peri-threshold detection
tasks. Thus, on ascending trials, better relative performance would
be expected at the higher odorant concentrations, whereas on descending trials
better relative performance would be expected at the lower odorant
concentrations. Such enhancement, however, may not be the same magnitude for
ascending and descending trials. The descending group would be expected, for
example, to obtain considerable information in recognizing the difference
between the blank and the odorant on initial trials, whereas this is not the
case in the ascending group, where the odorant:blank distinction is less
clear. This is suggested by the fact that feedback tended to improve
performance at the lower, but not higher, odorant concentrations
(Figure 2). A number of
investigators have noted that improvement occurs across trials, indeed in some
cases even across days, in odor detection tasks in which perithreshold-level
stimuli are repeated presented (Pfaffman et al., 1954;
Engen, 1960;
Pangborn et al.,
1964; Doty et al.,
1981; Rabin and Cain,
1986), a phenomenon that is present even in rats
(Doty and Ferguson-Segall,
1989).
The present study employed a single chemical—phenyl ethyl
alcohol—in the detection task, begging the question as to whether
similar results would occur with other odorants and, if so, which ones.
Although, in an ultimate sense, the latter question is not testable (given the
tens of thousands of odorous chemicals available for assessment), most studies
find high correlations among thresholds for different compounds obtained from
the same subjects (Yoshida,
1984; Doty et al.,
1994). Such observations, along with recent evidence that odorant
receptor cells tuned to various physiochemical moieties are found somewhat
randomly distributed within zones of the olfactory epithelium (making them
vulnerable generally to damage of the epithelium)
(Sullivan et al.,
1994), suggest that the present findings are likely quite
general.
Feedback as a concept in psychophysical studies is complex. In descending runs, for example, a clear distinction can be made on the part of the subject between correct and incorrect trials, resulting—even when no `explicit feedback' is provided by the examiner—in the production of `implicit feedback'. No implicit feedback exists, however, in ascending runs at low odorant concentrations. In other words, under a `no feedback' paradigm, trials at higher concentrations of a descending series provide a form of task-related feedback, in that implicit information as to the correctness of the response is present (i.e. the subject has no doubt as to which stimuli are odorants and which are blanks). Such information is not available, however, at lower stimulus concentrations. Thus, ascending and descending stimulus presentation may be asymmetric in terms of their intrinsic sequential feedback properties. The data of the present study are in accord with this notion, in that at higher perithreshold odorant concentrations intrinsic feedback presumably maximized performance (hence no effects of extrinsic feedback were evident), whereas at lower odorant concentrations extrinsic feedback tended to improve detection performance. However, the effects, while present, were not marked in this study, likely because of the stable performance induced, in part, by the use of a relatively large number of trials at each concentration step and a two-alternative forced-choice response paradigm.
|
Acknowledgments |
|---|
Supported by grant PO1 DC 00161 from the National Institute on Deafness and Other Communication Disorders, National Institutes of Health.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Dorries, K.M., Adkins-Regan, E. and Halpern, B.P. (1991) Sex difference in olfactory sensitivity to the boar chemosignal, androstenone, in the domestic pig. Anim. Behav., 42,403 -411.[CrossRef]
Doty, R.L. (1991) Olfactory system. In Getchell, T.V., Doty, R.L., Bartoshuk, L.M. et al. (eds),Smell and Taste in Health and Disease . Raven Press, New York, pp. 175-203.
Doty, R.L. and Ferguson-Segall, M. (1989) Influence of adult castration on the olfactory sensitivity of the male rat: a signal detection analysis. Behav Neurosci 103,691 -694.[Medline]
Doty, R.L. and Laing, D. (2003) Psychophysical measurement of human olfactory function, including odorant mixture assessment. In Doty, R.L. (ed.), Handbook of Olfaction and Gustation, 2nd edn. Marcel Dekker, New York, pp.203 -228.
Doty, R.L., Brugger, W.E., Jurs, P.C., Orndorff, M.A., Snyder, P.J. and Lowry, L.D. (1978) Intranasal trigeminal stimulation from odorous volatiles: psychometric responses from anosmic and normal humans. Physiol. Behav., 20,175 -185.[CrossRef][Medline]
Doty, R.L., Snyder, P.J., Huggins, G.R. and Lowry, L.D. (1981) Endocrine, cardiovascular, and psychological correlates of olfactory sensitivity changes during the human menstrual cycle. J. Comp. Physiol. Psychol.95 , 45-60.[CrossRef][Web of Science][Medline]
Doty, R.L., Smith, R., McKeown, D.A. and Raj, J. (1994) Tests of human olfactory function: principal components analysis suggests that most measure a common source of variance. Percept. Psychophys., 56,701 -707.[Web of Science][Medline]
Doty, R.L., Marcus, A. and Lee, W.W. (1996) Development of the 12-item Cross-cultural Smell Identification Test (CC-SIT). Laryngoscope,106 , 353-356.[CrossRef][Web of Science][Medline]
Engen, T. (1960) Effect of practice and instruction on olfactory thresholds. Percept. Mot. Skills,10 , 195-198.
Pangborn, R.M., Berg, H.W., Roessler, E.B. and Webb, A.D. (1964) Influence of methodology on olfactory response. Percept. Mot. Skills, 18,91 -103.
Pfaffmann, C., Schlosberg, H. and Cornsweet, J. (1954) Variables affecting difference tests. In Peryam, D.R., Pilgrim, F.J. and Peterson, M.S. (eds), Food Acceptance Testing Methodology. Washington, DC: National Academy of Science, Nat Res Council, pp. 4-17.
Rabin, M.D. and Cain, W.S. (1986) Determinants of measured olfactory sensitivity. Percept. Psychophys. 39,281 -286.[Web of Science][Medline]
Sullivan, S.L., Ressler, K.J. and Buck, L.B. (1994) Odorant receptor diversity and patterned gene expression in the mammalian olfactory epithelium. Prog. Clin. Biol. Res., 390,75 -84.[Medline]
Yoshida, M. (1984) Correlation analysis of detection threshold data for `standard test' odors. Bull. Fac. Sci. Eng. Cho. Univ.,27,343 -353.
Accepted June 2, 2003
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. H. Gaffney and D. J. Paustenbach A Proposed Approach for Setting Occupational Exposure Limits for Sensory Irritants Based on Chemosensory Models Ann. Hyg., June 30, 2007; (2007) mem019v1. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||