Introduction
Anecdotal reports suggest that blind people might develop supra-normal olfactory abilities. For example, James Mitchell, a congenitally deaf and blind boy, was allegedly able to follow the odor trail of a person for several miles (Stewart,
1815). Researchers hypothesize that such abilities could result from sensory compensation, i.e., enhanced sensitivity of functioning modalities resulting from deprivation in one or more senses (Kupers & Ptito,
2014). This compensation could emerge due to both central and peripheral reasons. For instance, in sighted individuals, the occipital cortex is activated in visual processing, and recent studies showed occipital activation in blind participants during odor-processing tasks, like odor detection, categorization and discrimination (Kupers et al.,
2011; Renier et al.,
2013). This suggest that “visual” brain regions could, as a result of functional and structural reorganization, support the processing of olfactory information in the visually impaired (Araneda, Renier, Rombaux, Cuevas, & De Volder,
2016; Kupers & Ptito,
2014). Further, an extensive use and attention to olfactory information in everyday life might promote the development of enhanced olfactory abilities (Cuevas, Plaza, Rombaux, De Volder, & Renier,
2009; Gagnon, Ismaili, Ptito, & Kupers,
2015). This possibility is further confirmed by studies showing a high subjective value of the sense of smell among blind individuals (Beaulieu-Lefebvre, Schneider, Kupers, & Ptito,
2011; Ferdenzi, Coureaud, Camos, & Schaal,
2010). A recent review on the neurobiological aspects of sensory compensation (Araneda et al.,
2016) proposed that such training may have a direct effect on olfactory bulb (OB) volume, that in turn influence olfactory function. Indeed, OB volume has been shown to be larger among the early blind individuals as compared to sighted controls (Rombaux et al.,
2010). Plasticity in the OB may underlie the enhanced olfactory perception in sighted (Buschhüter et al.,
2008; Hummel et al.,
2011; Hummel, Haehner, Hummel, Croy, & Iannilli,
2013; Mueller et al.,
2005) and among the blind (but see Mazal, Haehner, & Hummel,
2016).
Despite many hypotheses on the etiology of increased olfactory abilities of the blind people, the behavioral studies provide a mixed pattern of findings. As discussed in the following sections, inconsistent observations are reported for both sensory-driven olfactory tasks (e.g., odor detection threshold—see “
Olfactory threshold”) and higher-order olfactory functions (e.g., odor identification abilities—see “
Olfactory identification”). Although some studies presented below report enhanced olfactory abilities in blindness, there are also observations showing no reliable differences in smell function between blind and sighted individuals. Some degree of heterogeneity of findings may be expected as previous studies had small sample sizes, used different methods, or sampled from different populations (e.g., age, onset of blindness, study site location). However, to determine if there is an actual heterogeneity among the existing observations that is not simply due to chance, we conducted a meta-analysis of available studies targeting olfactory function in blind and sighted individuals.
In the current meta-analysis, we provide a comprehensive examination of olfactory function including odor threshold, odor discrimination, cued odor identification, and free odor identification in blind people as compared with sighted controls. Further, we investigated the potentially moderating roles of age and onset of blindness upon the observed olfactory differences between blind and sighted controls. Below follows a systematic review of the available scientific evidence followed by the systematic quantification of the observations through meta-analytical procedures (for a summary of standardized testing methods see Supplementary File S1).
Olfactory threshold
Olfactory threshold can be defined as the lowest concentration at which the presence of an odorant is reliably detected (Stevens,
1961). The term “olfactory detection threshold” refers to the ability to detect odorants; it is often referred to as “overall smell sensitivity”. As compared with higher-order olfactory tasks (e.g., odor identification) measurement of odor thresholds pose few demands on cognitive function (Hedner, Larsson, Arnold, Zucco, & Hummel,
2010; Sorokowska, Sorokowski, Hummel, & Huanca,
2013) and appears to draw more on peripheral functions of the olfactory system (e.g., Whitcroft, Cuevas, Haehner, & Hummel,
2016).
Historically, olfactory thresholds were the first smell function targeted in scientific studies on smell in blindness. Griesbach (
1899), Cherubino and Salis (
1957), and Boccuzzi (
1962) found no performance differences in blind and sighted individuals across a range of custom-made olfactory threshold tests. Corroborating these findings, more recent work has reported comparable odor sensitivity in blind and sighted individuals. Here, the Sniffin’ Sticks Test (SST; Hummel, Barz, Pauli, & Kobal,
1998; Hummel, Kobal, Gudziol, & Mackay-Sim,
2007) has been the most commonly used test (Cornell Kärnekull, Arshamian, Nilsson, & Larsson,
2016; Guducu, Oniz, Ikiz, & Ozgoren,
2016; Hamáková,
2008; Luers et al.,
2014; Oniz, Erdogan, Bayazit, & Ozgoren,
2011; Schwenn, Hundorf, Moll, Pitz, & Mann,
2002; Sorokowska,
2016). Additionally, researchers using a single staircase phenyl ethyl alcohol (PEA) test (Smith, Doty, Burlingame, & McKeown,
1993), Munich Olfaction Test (MOT; Diekmann, Walger, & von Wedel,
1994; Kruggel,
1989), and air-blast olfactometric method with mint oil and fresh coffee (Zielke & Gawęcki,
2003) observed no differences between sighted and blind individuals. Two studies conducted among blind and sighted children using
n-butyl alcohol (12-item threshold test; Wakefield, Homewood, & Taylor,
2004, and 13-bottle dilution series; Rosenbluth, Grossman, & Kaitz,
2000) showed no effects of visual impairment on sensory abilities. However, other studies using the SST threshold subtest have demonstrated a superior olfactory performance in blind (Beaulieu-Lefebvre et al.,
2011; Çomoğlu et al.,
2015; Cuevas et al.,
2010). At odds with these observations, Murphy and Cain (
1986) who used a
n-butanol threshold test reported higher olfactory sensitivity in sighted relative to the blind.
Olfactory discrimination
Assessment of olfactory discrimination ability is often based on non-verbal tasks (Frijters, Kooistra, & Vereijken,
1980; Potter & Butters,
1980) where subjects are confronted with a pair or three odors where they have to find out whether the two odors are different or which of the three odors is different. However, because of many different possibilities in execution of the test results from two odor discrimination tests may not significantly correlate with each other (Weierstall & Pause,
2012). Odor discrimination abilities (as measured with the Sniffin’ Sticks test) have been shown to be associated with executive functioning (Hedner et al.,
2010; Sorokowska, Sorokowski, & Hummel,
2014).
Olfactory discrimination tasks were also applied in early works on sensory compensation. By means of a discrimination task involving presentation of 2 odorants in 20 different concentrations, Mahner (
1909) concluded that congenitally blind discriminated between odors better than sighted individuals. The effect of blindness on olfactory discrimination has also been tested with a number of different methods. In the group of studies showing comparable performance of the blind and the sighted, researchers used SST (Beaulieu-Lefebvre et al.,
2011; Cornell Kärnekull et al.,
2016; Guducu et al.,
2016; Luers et al.,
2014; Majchrzak & Eberhard,
2014; Majchrzak, Eberhard, Kalaus, & Wagner,
2017; Oniz et al.,
2011; Schwenn et al.,
2002; Sorokowska,
2016), and the Munich Olfaction Test (MOT designed by Kruggel,
1989; this method was used by Diekmann et al.,
1994). However, results of studies employing SST were not consistent—some of these studies showed superior discrimination skills of blind people (Çomoğlu et al.,
2015; Cuevas et al.,
2010), similar to works using a custom set of 30 odorants (Cuevas et al.,
2009; Renier et al.,
2013; Rombaux et al.,
2010). Again, one study demonstrated better olfactory performance in sighted than legally blind subjects in a 16-item discrimination test (Smith et al.,
1993).
Olfactory identification
Measurement of odor identification ability is the most commonly used test of olfactory function in various scientific studies. There are numerous versions of tests available, while the Sniffin’ Sticks (Hummel, Sekinger, Wolf, Pauli, & Kobal,
1997) and the University of Pennsylvania Smell Identification Test (UPSIT; Doty, Shaman, & Dann,
1984; Doty, Shaman, Kimmelman, & Dann,
1984) are most frequently used. Identification may be assessed in an uncued task where no retrieval support is provided (free identification) or by cued identification where a number of alternatives is provided of which one is the target name. Proficiency in odor identification is associated with verbal abilities (Larsson, Nilsson, Olofsson, & Nordin,
2004) and cultural context such that tests need to be specifically adapted for various countries and cultures (e.g., Oleszkiewicz et al.,
2016). Although relatively straightforward, minor manipulations in test administration may change test results significantly (e.g., reading the options prior to smelling in a cued odor identification task might significantly decrease performance in this test, Sorokowska, Albrecht, & Hummel,
2015).
Interestingly, no studies have reported superiority of the blind individuals in cued identification. This has been shown with the use of various standardized tests, e.g., the SST (Beaulieu-Lefebvre et al.,
2011; Çomoğlu et al.,
2015; Cuevas et al.,
2010; Guducu et al.,
2016; Hamáková,
2008; Luers et al.,
2014; Majchrzak & Eberhard,
2014; Majchrzak et al.,
2017; Oniz et al.,
2011; Schwenn et al.,
2002; Sorokowska,
2016; Sorokowska & Karwowski,
2017), UPSIT (Smith et al.,
1993), Munich Olfaction Test (MOT, Kruggel,
1989) in Diekmann et al. (
1994), Monex 40 Sniffin’ Sticks battery (Freiherr et al.,
2012) in a study by Iversen, Ptito, Møller and Kupers (
2015). Null effect of blindness was also reported in studies using a variety of custom-made tools, like a set of 30 odors (Cuevas et al.,
2009), a set of 38 odorants (Gagnon et al.,
2015), and by a test comprising 25 common items identified by blind children (Rosenbluth et al.,
2000).
In contrast, a different pattern of findings is obtained for free identification. Although two studies observed comparable performance in free identification using the SST identification test (Sorokowska,
2016; Sorokowska & Karwowski,
2017), others report superior performance among blind compared to sighted. As is true for the other assessed olfactory domains, these methodologies used varied techniques including a set of 30 odorants (Cuevas et al.,
2009; Renier et al.,
2013; Rombaux et al.,
2010), 80 everyday substances (Murphy & Cain,
1986), and 12 odors taken from SST battery (Cornell Kärnekull et al.,
2016). Free identification abilities were found to be superior also among blind children; here, researchers applied 16 (Wakefield et al.,
2004), or 25 common odors (Rosenbluth et al.,
2000).
Other olfactory abilities
Olfactory abilities encompass various skills, not only these tested by typical smell tests. Such abilities were also compared between blind and sighted participants. Again, some studies demonstrated comparable performance of blind and sighted participants—for example, in retronasal identification test, i.e., in a task involving identification of flavors delivered through participant’s mouth. Null effect of blindness was shown for a test consisting of 38 odorants (Gagnon et al.,
2015) and retronasal smell test designed by Heilmann, Strehle, Rosenheim, Damm and Hummel (
2002) in the study of Cuevas et al. (
2010). Also episodic odor recognition performance (Cornell Kärnekull et al.,
2016; Sorokowska & Karwowski,
2017) was similar among blind and sighted individuals. Finally, event-related potentials (ERPs) were analyzed in response to both olfactory and trigeminal stimuli. Trigeminal nerve cells respond to tactile, thermal, or nociceptive stimulation, and trigeminal sensations include stinging, burning, tickling etc. (Hummel & Livermore,
2002; Kleemann et al.,
2009). Observed ERPs pattern did not differentiate blind and sighted subjects for neither olfactory nor trigeminal stimuli (Cuevas et al.,
2011; Guducu et al.,
2016; Schwenn et al.,
2002).
Other studies showed that olfactory abilities of blind people were better than those of sighted individuals. The visual deprivation effect was observed, for example, in an odor categorization task involving a set of 30 odorants (Cuevas et al.,
2009; Renier et al.,
2013) and for free identification time for 38 odorants (Gagnon et al.,
2015), or 25 common items tested among blind children (Rosenbluth et al.,
2000). The two existing questionnaire studies demonstrated higher olfactory awareness of blind adults (Beaulieu-Lefebvre et al.,
2011) and more olfactory-related behaviors (self-assessed reactions to odors in different situations) among blind children compared to their sighted peers (Ferdenzi et al.,
2010).
Interestingly, all magnetic resonance studies discussed in the current review have reported olfactory-related superiority in blind individuals. First, they were found to have higher OB volumetric measurements assessed by an MRI scan (Rombaux et al.,
2010). Further, fMRI activation patterns differed between the blind and the sighted participants. Researchers observed stronger occipital activation in blind subjects during odor-processing tasks (discrimination or categorization of fruit and flower odors in Renier et al.,
2013, and odor detection in Kupers et al.,
2011) and stronger response to olfactory stimuli in primary (right amygdala) and secondary (right orbitofrontal cortex and bilateral hippocampus) olfactory areas (Kupers et al.,
2011). Finally, in the only existing study on a social aspect of olfaction, i.e., identification of fear from samples of male odor (Iversen et al.,
2015), blind people performed better than the sighted.
Discussion
The potential presence of olfactory compensation in blindness has interested scientists for decades. Many studies conducted since the beginnings of the twentieth century explored this topic, although findings regarding olfactory compensation have been inconclusive. The results from the present meta-analysis show that the olfactory abilities of blind and sighted people are not much different overall. No positive effects from visual impairment were observed for all aspects analyzed in the current research: odor detection threshold, olfactory discrimination, free and cued odor identification abilities. In addition, age, proportion of women and blindness onset did not moderate the observed, null findings.
Consistent with what has been suggested by experts in the area of sensory compensation (Kupers & Ptito,
2014), we found that compensatory effects in smell function are not straightforward. Notably, the obtained effect sizes for all odor functions were highly heterogeneous and the observed differences between blind and sighted individuals in single studies were mostly observed in small studies. Potential explanations for this heterogeneity are further discussed below.
Most previous studies did not show significant differences between congenital and late blind participants (e.g., Çomoğlu et al.,
2015; Sorokowska,
2016). However, some small, single studies indicated that the olfactory abilities of early blind groups differed from the performance of sighted people. Early blind participants performed better than sighted people in free identification (the effects were particularly strong in Cuevas et al.,
2009 and in Renier et al.,
2013 and Rombaux et al.,
2010 studies) and discrimination tests (Cuevas et al.,
2009; Mahner,
1909; Renier et al.,
2013; Rombaux et al.,
2010). Additionally, in the current meta-analysis we observed a slight, albeit non-significant trend indicating that early blind participants tend to perform better than sighted people in the threshold task. The findings on early blind subjects are particularly interesting, given the existing hypotheses regarding their superior olfactory performance. Probably, the observed magnitude and direction of effects in the case of early blind people results from cerebral reorganization that could support their olfactory processing. Degree of such a reorganization could change, depending on a moment of sensory loss. Although in a study involving a mixed sample of early and late-blind people (visual acuity below 0.1), Luers et al. (
2014) showed that the duration of blindness does not correlate with olfactory function (
r between 0.01 and 0.17 for SST subtests), in Majchrzak et al. (
2017), the correlations reported for olfactory discrimination and identification in a sample of blind and visually impaired people were mostly positive and significant (
r = 0.234,
p < 0.05 for odor discrimination and
r between − 0.48 and 0.19 for SST identification subtest, depending on the reason of visual impairment). Nevertheless, it is possible that complete loss of sight before visual development is a different case. The functional reorganization in the occipital cortex (Leclerc, Saint-Amour, Lavoie, Lassonde, & Lepore,
2000) could aid some unisensory processes, which was shown, for example, for auditory skills (Gougoux, Zatorre, Lassonde, Voss, & Lepore,
2005). However, some aspects of sensory abilities and performance can be also impeded in blindness. Absence of a calibrating visual reference frame in the congenitally blind can, for example, negatively influence multisensory spatial integration between hearing and touch (Hötting, Rösler, & Röder,
2004) or ability to localize sound sources in the vertical spatial plane (Lewald,
2002; Zwiers, Van Opstal, Cruysberg, Opstal, & Cruysberg, 2001). This illustrates how blindness could underlie both enhanced and decreased sensory skills—and might explain why the overall pattern of results in the case of olfaction is not very simple. Although calibration problems seem not to be the case for the sense of smell, there might be some additional issues, like development of specific experience-based associations that differ in sighted and blind and which result in differences in olfactory processes. Further, available neural resources could be used more extensively for modalities other than the sense of smell, not allowing for development of olfactory superiority.
Nevertheless, it needs to be highlighted that our meta-analysis shows rather minimal compensatory plasticity for olfaction which is not in line with most findings on in the unisensory tactile (Van Boven, Hamilton, Kauffman, Keenan, & Pascual-Leone,
2000) and auditory domain processing (Lessard, Paré, Lepore, & Lassonde,
1998). Based on our research, some new hypotheses might be presented as to why blind individuals do not develop very high olfactory capacities in some domains to compensate for their lack of vision.
First, the aim of sensory compensation processes is to alleviate the incapacitating consequences of sensory deficit or loss (Bäckman & Dixon,
1992). Both blind and sighted people could be equally proficient in some skills, and in this case it would be not possible to develop some olfactory abilities any better (for example, studies on cued olfactory identification typically demonstrate a strong ceiling effect; Hummel, Kobal, Gudziol, & Mackay-Sim,
2007). Second, compensatory processes could be more pronounced for other sensory modalities because the olfactory and visual data are not necessarily redundant—for example, for assessments of attractiveness, visual and olfactory cues are not consistent (Sorokowska,
2013); lack of vision would not necessarily enhance contradictory or complementary signals. Third, it is still possible that superior abilities of the blind people would be observed during processing of olfactory information outside laboratory context, in more ecologically valid studies. For example, blind people could be compared to the sighted in detection of odors in an environment containing also other smells, or in recognition of smells they would be exposed to on the way to the testing facility. Such studies would be more appropriate to test the hypothesis on the increase in olfactory performance due to more extensive, daily olfactory training of blind people and higher olfactory awareness. Further, in natural experiments, sighted people would probably not be equally focused on olfactory stimuli like during laboratory testing (e.g., in many analyzed studies the eyes of sighted people were closed or covered, thus limiting the regular sensory input). Perhaps, the olfactory superiority of the blind people would only be observed in conditions where participants’ attention would not be specifically driven to olfactory processing. Finally, in our meta-analysis, the studies in which significant differences between sighted and blind people were observed were mostly based on few observations. This might suggest a large individual variation, especially among the early blind where olfactory expertise may result from a more active attention towards olfactory information that can ultimately yield a keener sense of smell. Similar to trained subjects (employed by the Philadelphia Water Department), blind people could become better in odor detection (Smith et al.,
1993), or other olfactory abilities. Such acquisition of olfactory function has also been noticed in a number of studies on “olfactory training” which suggest that olfactory function can be improved by regular, short-term exposure to odors (Sorokowska, Drechsler, Karwowski, & Hummel,
2017). However, it also needs to be noted that in studies observing the highest differences between blind and sighted, olfactory abilities in sighted were relatively low. In future studies, a more balanced sample selection is required to conduct reliable comparisons.
Another interesting issue are the effect differences across the four different olfactory tasks. For example, the highest number of negative effects (indicating slight advantage of sighted subjects over the blind participants) was observed for cued identification. As discussed in the introduction, cultural context is important for the identification test execution (e.g., Oleszkiewicz et al.,
2016; Sorokowska & Hummel,
2014). It is possible that due to different way of life, blind people are exposed to different odors than the sighted people, and thus some tests which are theoretically based on common smells can be more difficult for the blind people than for the sighted. Further, the meta-analysis indicated a non-significant difference between blind and sighted individuals in olfactory threshold, although we observed a slight tendency for early blind individuals to perform better than the sighted people in this subtest, which was discussed earlier in this section. The effect in the threshold task was not moderated by publication bias or influenced by small studies. Importantly, ceiling effect is not very often observed in olfactory threshold studies, and therefore, changes in this olfactory function can be seen relatively easily. It is also crucial that threshold tests are rather independent from verbal abilities (Hedner et al.,
2010; Sorokowska, Sorokowski, Hummel, & Huanca,
2013). The absent difference between sighted and blind individuals suggests that one of potential moderators of the effects observed in single studies testing other olfactory abilities could be verbal skills.
In contrast, the higher olfactory discrimination skills among blind relative sighted were driven by three small studies that yielded especially large effects. The effect disappeared when estimated with the control of possible influences of underpowered studies. Nevertheless, it needs to be remembered that in the case of olfactory discrimination, the results highly depend on the way the odors are presented (Weierstall & Pause,
2012). The outlier studies observed on our funnel plot used the same methodology, i.e., discriminating between two odors (Cuevas et al.,
2009; Mahner,
1909; Renier et al.,
2013; Rombaux et al.,
2010), unlike, e.g., Sniffin’ Sticks, where a participant needs to identify which of the three odors is different (for details of the method see: Hummel, Kobal, Gudziol, & Mackay-Sim,
2007). This finding warrants further investigations.
Overall, blind individuals achieved 1.20 standard deviation higher scores in free identification. However, the large effect was primarily driven by two very small groups of blind people for whom enormously high effect sizes were observed; when the bias was corrected for, the effect proved unreliable. It needs to be noted that two groups participating in these studies (first in Cuevas et al.,
2009; and second in Renier et al.,
2013 and Rombaux et al.,
2010 studies) were asked to identify the same set of 30 odorants. It is possible that for some reasons, these odorants were easier to recognize for blind people relative to the sighted. Further, higher scores in free olfactory identification might be associated with better memory and retrieval of smell descriptors in the group of early blind subjects. Although recent studies (Cornell Kärnekull et al.,
2016; Sorokowska & Karwowski,
2017) showed that olfactory memory in the blind people is not better than this of sighted individuals, the retrieval of certain odor labels might be more effective among blind people, facilitating free identification of odorants. For example, Gagnon et al. (
2015) showed that blind people were faster in recognizing orthonasally presented odors which could indicate that this task was easier for them than for the sighted subjects.
One interesting finding of this meta-analysis was that age did not moderate the observed effects. This is surprising, as age in general is an important factor when evaluating olfactory abilities. (Larsson, Finkel, & Pedersen,
2000; Sorokowska, Schriever, et al.,
2015). However, detrimental effects of aging on the sense of smell could be partially due to an age-related decrease in cognitive abilities; that may affect some olfactory abilities (Hedner et al.,
2010). If the olfactory superiority of blind people indeed results from daily smell training (Gagnon et al.,
2015), this training might alleviate some of the aging effects. This is a topic that needs further investigation in future work.
Based on the review of studies presented in the introduction, we might suggest various future directions for research on olfactory performance of the blind people. First, an interesting factor that could influence performance of blind and sighted people regards the individual differences. Many studies on visual impairment have shown that some factors related to blind individuals’ daily life (e.g., physical activity) might influence several properties of perception (e.g., Seemungal, Glasauer, Gresty, & Bronstein,
2007). Likewise, it could be interesting to analyze individual differences related to olfactory perception, e.g., attention paid to olfactory stimuli, also directly in relation to olfactory functions, as these characteristics might moderate possible olfactory superiority. Second, future studies could test some elements not addressed in our general review and meta-analysis, like the odorants applied in previous research—for example, certain olfactory abilities depend also on trigeminal qualities of applied substances (Kleemann et al.,
2009). However, no effect of blindness was observed in Boccuzzi (
1962) for vanillin—one of a few odorants that do not produce trigeminal sensations (Doty et al.,
1978). Further, as discussed in the sections on olfactory identification test results, certain adaptations of identification tasks might be necessary, so that these tests would be equally difficult for both participating groups. Third, it would be very interesting to study performance of blind people in more complex olfactory tasks, e.g., in spatial orientation (e.g., Welge-Lussen, Looser, Westermann, & Hummel,
2014), or changes-detection tasks (e.g., Croy, Krone, Walker, & Hummel,
2015). Finally, a more detailed analysis of aspects related to the performance of the late-blind groups (including effects of various reasons of visual impairment, or relationship of olfactory abilities and age of blindness onset / blindness duration like in Majchrzak et al.,
2017) would be a very interesting idea for future meta-analytic studies.
A certain limitation to our study was the dependence on secondary sources for the analyzed data. Authors use different definitions of blindness, and the works analyzed in the current review and meta-analysis (all theoretically on blindness and olfaction) sometimes referred to “legally blind”, “early blind” or “congenitally blind” subjects. We recommend that future works on olfaction and blindness use more detailed and specific definitions. Other guidelines for future research on the topic of visual deprivation and olfaction include thorough analysis and control of the blindness status (late vs. congenital vs. early blind participants), and attention paid to participants with residual vision whose performance might differ from that of completely blind participants.