Acta Psychologica 134 (2010) 105–109

Contents lists available at ScienceDirect

Acta Psychologica

journal homepage: www.elsevier .com/ locate/actpsy

Sweet odours and sweet tastes are conflated in memory

Richard J. Stevenson *, Megan J. OatenDepartment of Psychology, Macquarie University, Sydney NSW2109, Australia

a r t i c l e i n f o

Article history:Received 9 September 2009Received in revised form 6 January 2010Accepted 6 January 2010Available online 25 January 2010

PsycINFO classification:23202343

Keywords:OdourTasteInteractionSimilarity

0001-6918/$ - see front matter � 2010 Elsevier B.V. Adoi:10.1016/j.actpsy.2010.01.001

* Corresponding author. Tel.: +61 2 98508098; fax:E-mail address: richard.stevenson@psy.mq.edu.au

a b s t r a c t

Certain odours and certain tastes appear to share common perceptual properties. One example is sweet-ness, a perceptual experience that results from stimulation of taste receptors on the tongue typically bysugars. The experiment here examined for evidence of this perceptual similarity using a novel and indi-rect test. Participants were exposed six times each, to three odours (strawberry, caramel, and oregano)and three tastes (sucrose, saline, and citric acid). Following a 10-min interval, participants were givena surprise frequency estimation task, in which they had to judge how often each stimulus had occurred.If sweet-smelling strawberry and caramel odours really do share this perceptual characteristic in com-mon with sweet tasting sucrose, then frequency estimates for sucrose should be overestimated relativeto non-sweet tastes. Not only was this observed, but frequency estimates for sweet tastes were also foundto correlate with (1) evaluations from a later test of similarity between these sweet smells and sucrose,and (2) the degree to which these odours smelled sweet. These findings suggest a shared perceptual fea-ture between such odours and sucrose – sweetness – under conditions where no judgment of perceptualquality was required.

� 2010 Elsevier B.V. All rights reserved.

1. Introduction

Certain odours are commonly reported by participants to smellsweet (Dravnieks, 1985), yet sweetness is a sensation that is typi-cally associated with stimulation of a different sensory system,taste (Schiffman, 2002). The sweet sensation that is perceived whencertain odours are smelled does not appear to be produced by inad-vertent stimulation of the taste receptors in the mouth, nor by thepresence of taste receptors in the nose (Labbe, Damevin, Vaccher,Morgenegg, & Martin, 2006). Rather odour sweetness appears tobe learned, in that repeated exposure by mouth to sweet tastescombined with odourants that reach the nose via the nasopharynx(rather than the nostrils – orthonasally) results in learning of thecombination, and this combination is then retrieved when thatodour is later smelled alone (Stevenson, Boakes, & Prescott, 1998;Yeomans, Mobini, Elliman, Walker, & Stevenson, 2006). A questionof some importance, in relationship to such findings, is whether ornot the sensation generated by odours that are reported to smellsweet is actually akin to the sensation of sweetness produced byplacing certain tastants such as sucrose on the tongue. The presentstudy employed a new and indirect technique to establish whetheror not these two forms of sensation are indeed similar.

Certain odours produce a sensation that appears similar totasted sweetness. This was first noted phenomenologically, in that

ll rights reserved.

+61 2 98508062.(R.J. Stevenson).

participants spontaneously used the term sweet to describe suchsmells (e.g. strawberry; Harper, Land, Griffiths, & Bate-Smith,1968). The first empirical studies of this phenomenon testedwhether combining a sweet-smelling odour with a sweet taste,and presenting this mixture to the mouth, would result in largersweetness ratings – sweetness enhancement – relative to present-ing the sweet taste alone (Frank & Byram, 1988). Such a synergybetween a sweet taste and a sweet smell would suggest that bothgenerated similar sensations, and the finding that odour sweetnesswas able to predict, independently of other variables, the degree towhich that odour would demonstrate sweetness enhancementprovided further confirmation (e.g. Valentin, Chrea, & Nguyen,2006). A significant complication with this line of reasoning wasthe discovery that the number of rating scales that participantscompleted when evaluating an odour-taste mixture had a big im-pact upon whether or not ‘sweetness enhancement’ was observed(e.g. Clark & Lawless, 1994). Multiple relevant scales, such as get-ting participants to rate strawberry and sweetness when evaluat-ing a sucrose-strawberry mixture were found to reduce or eveneliminate the sweetness enhancement effect (van der Klaauw &Frank, 1996).

Such concerns led investigators to contemplate other methods toinvestigate the putative perceptual similarity between certain tastesand smells. One approach was to see whether sweet smells selec-tively facilitate the identification of sweet tastes, which they do(White & Prescott, 2007). However, the principal approach adoptedto bypass concerns about the use of rating scales has been to

106 R.J. Stevenson, M.J. Oaten / Acta Psychologica 134 (2010) 105–109

examine whether sub-threshold sweet smells might be detectedmore readily in the presence of a sweet taste relative to a non-sweettaste. At least two studies obtained evidence consistent with this(Dalton, Doolittle, Nagata, & Breslin, 2000; Delwiche & Heffelfinger,2005). The inference from these studies is that the perceptual simi-larity between the two stimuli is the key factor in promoting sub-threshold detection. However, at least two published studies havenot obtained convincing evidence for this effect (Elgart & Marks,2006; Pfeiffer, Hollowood, Hort, & Taylor, 2005), and a further ratherdifferent approach has also cast some doubt over the conclusion ofsuch findings. Rankin and Marks (2000) established whethersweet-smelling odours and sweet tastes form a common judgmentalcontext. Although similarity ratings suggested that participants didindeed judge certain odours and tastes as alike, such odours failed toprovide a common judgmental context for sweet tastes. This failureto observe a common judgmental context is perhaps surprising,especially as sweet odours and sweet tastes appear to depend uponoverlapping neural structures (i.e. both are adversely impacted bydamage to brain structures responsible for taste processing; Steven-son, Miller, & Thayer, 2007) and both sweet-smelling odours andsweet tastes selectively exert common physiological effects(increasing pain tolerance; Prescott & Wilkie, 2007).

In the study reported here we utilized a new technique to ex-plore the perceptual similarity between certain tastes and smells.It has been known for some time that similarity between items orevents can later affect judgments of the frequency with which thoseitems or events are believed to have occurred (Conrad, Brown, &Cashman, 1998; Hintzman, 2001). In particular, the greater the sim-ilarity between two or more items the more likely it is that this willresult in higher frequency estimates for those particular stimuli,relative to their actual rate of occurrence (Jones & Heit, 1993; Tus-sing & Greene, 1999). On this basis we reasoned that frequency esti-mates could serve as an indirect means of measuring perceptualsimilarity without introducing any overt judgments of odour ortaste qualities, or indeed, of similarity. Consequently, in the studydescribed below, we exposed participants to sets of tastes, includ-ing sweet tastes, and sets of smells, including sweet smells, whichwere later followed by a surprise frequency estimation task. A sur-prise task has the added advantage that it tends to draw upon moreglobal features (i.e. those likely to be most salient during percep-tion) rather than on specific local features (e.g. Freund & Hasher,1989). The frequency estimation task was then followed by a seriesof tests that included explicit judgments of the perceptual and he-donic qualities of the target odours, and the similarity betweenthese odours and sucrose. We predicted that sweet tastes, andsweet smells, would be judged, incorrectly, to have occurred morefrequently than they actually had, and more frequently than non-sweet smells and tastes, which in actuality had all occurred withequal frequency. Furthermore, we expected that the greater the de-gree of judged similarity between the sweet taste and the sweetsmells (and the degree to which these odours smelled sweet) wouldbe associated with higher frequency ratings for the sweet taste.

2. Method

2.1. Participants

Forty-four participants (11 male, 33 female) participated for coursecredit. One participant was excluded, as her English ability was inade-quate to understand the experimental instructions and tasks. Theremaining 43 participants had a mean age of 21.0 years (SD = 6.1).

2.2. Stimuli

Three odourants were employed, each dissolved in 10 ml of pro-pylene glycol and then diluted with tap water (as used throughout)

to their requisite concentration; caramel (1.0 g/L; Dragoco), straw-berry (1.2 g/L; Quest) and oregano (0.5 g/L). Odours were pre-sented in 10 ml aliquots in disposable clear plastic sample cups.All odour solutions were transparent and visually identical.

The three tastants employed were sucrose (5.8% w/v), saline(0.45% w/v) and citric acid (0.2% w/v). Tastants were presented in10 ml aliquots in disposable clear plastic sample cups (i.e. the samecups and sample size as for the odourants). Tastant solutions werealso transparent and visually indistinguishable. Finally, all stimuliwere presented at room temperature.

2.3. Procedure

Using a within-participant design, each subject completed afour-phase procedure – exposure to tastants and odourants, the fil-ler task, the frequency estimation task, and the odour/taste evalu-ations. The first phase involved exposure to two blocks ofodourants and to two blocks of tastants. An odour block consistedof 12 samples to smell, and always contained three samples of car-amel, three of strawberry, three of oregano and three of water. Ataste block consisted of 12 samples to taste, and these always in-cluded three samples of sucrose, three of saline, three of citric acidand three of water. Order of presentation within each block wasrandomized. By the end of the exposure phase participants hadsampled (smelled or tasted) 48 stimuli – 6 caramel, 6 strawberry,6 oregano, 6 sucrose, 6 saline, 6 citric acid and 12 water blanks.

On odour blocks, the experimenter instructed the participant topick up a sample, sniff it, and replace the cup on the tray and thento judge whether or not the fluid had an odour. The experimenterthen enforced a 15 s break before the next sniffing trial began. Ontaste blocks, participants were instructed to pick up the cup, pourall of the solution into their mouth, roll it around, and then expec-torate. Participants were then asked to judge whether or not thesolution had a taste. This was followed by a tap water rinse andexpectoration. After a 15 s break, the next tasting trial then began.Order of block presentation was fully counterbalanced and partic-ipants received a 2 min break between each block.

After the exposure phase was complete, participants were pre-sented with a word finder puzzle and were told that they needed toidentify as many words as possible and that this task would betimed. Participants were given 10 min in which to complete asmuch of the puzzle as possible. As these puzzles were large andcomplex, no participant identified all the words within the allottedperiod. The purpose of this filler phase was to ensure that the fre-quency estimation task drew upon long-term rather than immedi-ate memory. The word finder task was used to minimize thepossibility that participants might rehearse information aboutthe exposure phase during this interval.

The frequency estimation task then followed. Each participantcompleted two such tasks, one for the odours and another for thetastes. The test was a surprise, as no mention was made prior tothis point that participants might be asked any questions concern-ing the exposure phase. Whether the odour or taste frequency esti-mation task came first or second, was counterbalanced acrossparticipants. For the odour task the following instructions werepresented: ‘‘Earlier, you were given some fluids to smell. Some ofthese smelled fruity (strawberry odour), some like caramel (cara-mel odour), some spicy (oregano odour), and some had no smellat all. We would like you now to try and estimate (or guess) howmany times you smelled the strawberry, caramel and oreganosmelling fluids.” Three questions then followed (in a randomizedorder). One asked participants how many times they had encoun-tered strawberry odour. Another, how many times they hadencountered caramel odour, and a final one how many times theyhad encountered oregano odour. Underneath each question was ascale numbered consecutively from 0 to 24. Participants were

Sucr

ose

Salin

e

Cit

ric

acid

Car

amel

Stra

wbe

rry

Ore

gano

5

6

7

8

9

Fre

quen

cy e

stim

ate

Fig. 1. Mean frequency estimates for each stimulus.

R.J. Stevenson, M.J. Oaten / Acta Psychologica 134 (2010) 105–109 107

instructed to circle one number, reflecting their initial response,and then immediately move onto the next question. The taste taskwas constructed in a similar manner, and the instructions here ranas follows: ‘‘Earlier, you were given some fluids to taste. Some ofthese tasted sweet, some sour, some salty and some had no tasteat all. We would like you now to try and estimate (or guess) howmany times you tasted sweet, sour and salty fluids”. The questionsand response scales were as described for the odour task, with theobvious difference that taste terms (sweet, salty, and sour) weresubstituted for the odour names used above.

Immediately after the frequency task was completed, partici-pants were given the evaluation task. This was composed of twoparts, which were presented in a fixed order. The first was to sniffeach of the three odourants (caramel, oregano, and strawberry),and then, for each one, judge how sweet, how caramel-like, howstrawberry-like, how oregano-like and how strong it smelled (allusing seven-point category scales, anchors ‘‘Not at all” [1] to ‘‘Very”[7]) and how much they liked or disliked the smell (anchors, ‘‘Dis-like” [1], ‘‘Indifferent” [4], to ‘‘Like” [7]). Every attribute was ratedfor each sample. A 30 s interval separated each of these three trials,and order of presentation was counterbalanced across participants.The second task involved judging the similarity of a sucrose solu-tion to each of the three odours, caramel, strawberry and oregano.Participants sampled and expectorated the sucrose solution,rinsed, and then smelled each odour in a counterbalanced order.After sniffing each odour participants completed a similarity scale(seven-point category scale from 1 ‘‘Not at all similar” to 7 ‘‘Verysimilar”). This test was self-paced.

3. Results

3.1. Manipulation check

To ensure that the odours used in the experiment had theappropriate characteristics, participants were asked both to evalu-ate their properties (including how sweet they smelled) and tojudge their similarity to sucrose. The two odours that were desig-nated as the sweet-smelling stimuli were both judged to smell sig-nificantly sweeter than the non-sweet-smelling odour (cherry[M = 6.1/7] vs. oregano [M = 1.4/7], t(42) = 25.15, p < 0.001; cara-mel [M = 5.6/7] vs. oregano [M = 1.4/7], t(42) = 12.90, p < 0.001).Similarly, both sweet-smelling odours were rated as smelling morelike the taste of sucrose than the non-sweet-smelling odour (cher-ry [M = 3.4/7] vs. oregano [M = 1.3/7], t(42) = 7.22, p < 0.001; cara-mel [M = 3.9/7] vs. oregano [M = 1.3/7], t(42) = 7.92, p < 0.001). Thetwo sweet-smelling odours did not significantly differ from eachother on either sweetness or similarity ratings.

3.2. Frequency ratings

Estimated frequency ratings for the three odours and the threetastes are illustrated in Fig. 1. Participants actually experiencedeach odour and taste six times. To test whether the frequency rat-ings obtained exceeded the actual frequency with which the stim-uli were presented, six one-tailed one-sample t-tests (mu = 6) wereconducted, with alpha set at 0.017 following Bonferroni correction.Participants significantly overestimated the occurrence of sweettasting stimuli (t(42) = 3.25, p = 0.002) and sweet-smelling caramelodour (t(42) = 2.65, p = 0.011), with sweet-smelling cherry odourapproaching significance (t = 2.32). Participants’ frequency esti-mates for the spicy smelling oregano odour and the salty and sourtastes, did not significantly differ from their actual frequency ofoccurrence.

We then established whether the frequency estimates for thetwo sweet-smelling odours differed from each other, and whether

the estimates for the two non-sweet tastes differed from eachother as well. Paired sample t-tests (two-tailed, alpha = 0.05) re-vealed no significant difference in frequency estimates in eithercase. These ratings were, respectively, collapsed to form a sweetodour estimate, and a non-sweet taste estimate, so that these couldbe compared to participants of non-sweet odour estimate, andsweet taste estimate, within the same analysis of variance. In addi-tion, we also examined here whether the order in which the fre-quency tests had been completed exerted any effect onparticipants ratings. Using a three-way mixed-design ANOVA, withPerceptual similarity (sweet quality vs. non-sweet quality) andModality (odour vs. taste) as within-participant factors and Testorder (odour estimates first vs. taste estimates first), revealed justone significant effect – that of Perceptual similarity (F(1,41) = 5.04,MSE = 6.22, p < 0.05, partial eta-squared = 0.11). Here, sweet tast-ing and smelling stimuli were judged to have occurred significantlymore frequently (M = 7.4) than non-sweet-smelling and tastingstimuli (M = 6.6). Finally, we tested whether frequency estimatesfor the sweet taste significantly exceeded that for non-sweettastes, and whether the same held true for sweet odours relativeto the non-sweet odour. For tastes, a paired sample t-test (one-tailed, alpha = 0.05) revealed that participants judged the sweettasting stimulus to have occurred significantly more frequentlythan the non-sweet tasting stimuli (t(42) = 2.03, p < 0.025). How-ever, this difference was not significant for the sweet-smelling ver-sus the non-sweet-smelling odours (t = 1.51).

3.3. Correlations between measures

All the frequency estimates for odours and tastes were posi-tively correlated (median r(43) = 0.41, p < 0.01, range = 0.58), thushigher frequency estimates for any one stimulus tended to predicthigher frequency estimates for the others. If the perceptual similar-ity between the two sweet-smelling odours and the sweet tastewas responsible for higher frequency estimates for the sweet taste,then we might expect that participants who judged these odours asparticularly similar to sucrose and who rated them as particularlysweet-smelling, would also tend to judge the sucrose solution tohave occurred more frequently. This was indeed the case. Greaterfrequency estimates for sucrose were significantly associated with:(1) higher judgments of similarity between sucrose and the sweetodours combined (r(43) = 0.33, p < 0.025, one-tailed); and (2) high-er sweetness ratings for the sweet odours combined (r(43) = 0.25,p = 0.05, one-tailed). These relationships were not significant fornon-sweet tastes (collapsing across salty and sour). However, thesecorrelations were positively signed and this may not be surprisinggiven the intercorrelations between frequency estimates for allstimuli noted above.

108 R.J. Stevenson, M.J. Oaten / Acta Psychologica 134 (2010) 105–109

Although similarity and sweetness ratings were expected to beassociated with greater frequency judgments for sweet tastes, it isequally plausible that hedonic similarity might have driven this ef-fect, or perhaps even similarity in perceived intensity. However,the data were not consistent with these possibilities. Hedonic rat-ings of the sweet-smelling odours combined were not associatedwith frequency estimates for sucrose. A Williams (1959) test indi-cated that the magnitude of the association between sweet tastefrequency estimates and similarity judgments (see above) signifi-cantly exceeded the magnitude of this latter correlation(t(42) = 1.90, p < 0.05, one-tailed), confirming that perceptual sim-ilarity between sweet tastes and sweet smells was a better predic-tor of frequency estimates for sweet tastes (this difference onlyapproached significance when odour sweetness ratings were usedinstead of similarity ratings). Intensity ratings for the sweet odourscombined were not significantly associated with frequency esti-mates for sweet tastes, and hedonic and intensity ratings werenot significantly associated with frequency estimates for non-sweet tastes combined, either.

4. Discussion

Consistent with expectations, participants appeared to conflatememories of sweet tastes with memories of sweet smells. That iswhen they were asked to judge the frequency with which theseand other events occurred in a surprise test later in the experiment,they tended to overestimate frequencies in the most commonlyoccurring perceptual category – sweet things. This effect was mostpronounced for sweet tastes, as these stimuli were judged to haveoccurred 28% more frequently than they actually occurred duringthe exposure phase of the experiment. Sweet tastes were also re-ported as occurring significantly more frequently than the non-sweet tastes. Similarly, frequency estimates for sweet smells alsotended to be inflated, although the effects here were not asmarked. Turning to the correlational data, we observed significantassociations between participants’ frequency estimates for sweettastes and their odour evaluations. These evaluations suggestedperceptual similarity between sweet tastes and the sweet smells.The more similar participants reported the sweet smells and su-crose to be, and the sweeter these odours were judged to smell,the more likely they were to report that sucrose solutions had oc-curred with a high frequency. We also noted that participants fre-quency estimates were positively intercorrelated, a point wereturn to later in the discussion.

One interpretation of these results is that when participants areasked to estimate the frequency with which the perceptual eventshere had occurred, those that were perceptually more alike wereconflated. Furthermore, this conflation occurred because of oneparticular perceptual feature, namely sweetness, which was sharedby two of the three odours and one of the three tastes. This inter-pretation may be subject to a number of criticisms. First, partici-pants presumably encode a variety of information about thestimuli during the exposure phase. This could include semantictype information (this is a sour taste; this is cherry odour), as wellas perceptual type information (having an experience of ‘sour’, orof a ‘cherry-like’ smell). To what extent would drawing uponsemantic type information invalidate the conclusion that percep-tual similarity is responsible for frequency estimate conflation?Arguably it would not if their descriptions of the stimuli werebased upon the stimuli’s perceptual characteristics. Indeed, if par-ticipants were labeling the odours, something which is usuallyquite difficult to do, then this should tend to weaken any confla-tion, as the odour labels (i.e. caramel, strawberry, and oregano),would be quite distinctive from those applied to the tastes (i.e.sweet, sour, and salty). On this basis then, where the semantic levelinformation is drawn from what participants perceive, then

semantic and perceptual information should concur and it maynot matter greatly which type of information is utilized duringthe frequency estimation task.

A second concern is whether some other shared characteristic,rather than sweetness, might have generated the conflation be-tween the sweet tasting and smelling stimuli. One probable candi-date common to all the sweet stimuli is that they are liked morerelative to salty or sour tastes, and the smell of oregano. Not onlydo the data here argue against liking as the basis for conflation,but so do the extant literature (e.g. Yeomans et al., 2006). First,we observed here that perceptual similarity between sweet tastesand sweet smells was positively correlated with higher frequencyestimates for sweet tastes. However, these similarity ratings couldhave been at least partially based upon liking, more liked thingsbeing judged more similar. To eliminate this possibility we con-ducted a further test, namely to examine the correlation betweenthe similarity ratings of sweet smells (combined) and sucrose, withfrequency ratings for sweet tastes, controlling for hedonic ratingsof the sweet smells. This partial correlation was still significant(r(40) = 0.34, p < 0.025, one-tailed), suggesting that even if a hedo-nic component was involved in forming similarity judgments, theshared perceptual characteristics were still predictive (the sameholds true for odour sweetness ratings too). Second, it is importantto note that there was no indication of any association betweenparticipants’ hedonic ratings for the sweet smells and sweet tastefrequency estimates. Third, if hedonic similarity were a key driverof the effects reported in this manuscript, then one might expectthat the unpleasant stimuli – saline, citric acid and oregano –would also show a similar effect, yet this was not observed. Finally,the literature also suggests that hedonic and sweetness effects canbe independent. This has emerged from studies examining the con-sequences of pairing sweet tastes with unfamiliar odours. A keyconcern in such studies is whether participants conflate sweetnessand hedonic ratings in their evaluations (i.e. when participants re-port an increase in ‘odour sweetness’ following odour-sucrose pair-ings does this really mean they now just like the odour more?). Atleast two experiments suggest this is unlikely. One demonstratedthat participants who do not like sweet solutions still show reliablechanges in odour sweetness following odour-sucrose pairings(Yeomans et al., 2006). Another found that testing participantswhen sated, following odour-sucrose pairings, revealed changesin sweetness, but no changes in hedonic ratings, in contrast to afurther group who were tested when hungry, who showed changesin sweetness and liking (Yeomans & Mobini, 2006).

A further issue concerns the positive correlations observed be-tween the frequency estimates for all the taste and odour stimuli.Clearly, participants must have known that all the stimuli oc-curred, broadly, at a similar rate, indeed this is reflected in the fre-quency ratings themselves, all of which are around the samemagnitude. So while we were able to observe changes in frequencyestimates for sweet tastes, and sweet smells, these occurred in thecontext of this broader knowledge. We suggest that one conse-quence of this is that where participants were especially prone toconflate sweet tastes and smells, this probably had the additionaleffect of inflating their frequency estimates for the other non-sweet stimuli. We suggest that this tendency may have led to theobserved positive correlations between all of the frequency esti-mates, and the general tendency for frequency estimates to exceedthe actual frequency of occurrence.

In conclusion, our data provide further support for the argu-ment that certain odours and certain tastes can share common per-ceptual features, sweetness in this particular case. This unusualsituation (typically referred to as synesthesia when dealing withthe rare instances of shared perceptual experience across differentmodalities) is apparently routine in the chemical senses. The labelsynesthesia may at first seem an unreasonable one, especially, as

R.J. Stevenson, M.J. Oaten / Acta Psychologica 134 (2010) 105–109 109

noted in the introduction that sweet smells are learned. However,recent evidence suggests that learning too plays a part in the acqui-sition of the rare synesthesias. While these clearly have a geneticbasis, as Marks and Osgood (2005) note, many synesthetic inducershave to be learned (e.g. musical notation, letters, and words) and socannot have been present since birth. In addition, a number ofstudies have obtained convincing evidence that exposure to partic-ular stimulus relationships in childhood (e.g. coloured fridge mag-nets) accounts for the nature of the inducer-concurrent mappingsin adulthood (e.g. Hancock, 2006; Witthoft & Winawer, 2006). Amore detailed discussion of the similarities and differences be-tween odour-induced tastes and the rare synesthesias can be foundin Stevenson and Tomiczek (2007). Finally, there may be goodfunctional grounds for this universal form of synesthesia (if this la-bel is correct), namely the evolutionary advantage of detecting, byorthonasal olfaction, sugar-rich (i.e. energy dense) foods. This func-tional explanation is given further credence by the observationthat rats too appear to be able to experience ‘odour sweetness’(Harris & Thein, 2005). Although these interactions between tasteand smell may be especially entwined (perhaps because of a com-mon function in generating flavour and its role in food selection,see Stevenson, 2009), they reflect a broader trend across all percep-tual modalities towards multisensory integration, aimed at suc-cessfully accomplishing particular functional goals (see Calvert,Spence, & Stein, 2004, for numerous examples). So while sensorymodalities may be a useful way of categorizing our experiences,common functional goals may be more important still.

Acknowledgements

The authors would like to thank Liz Sburlati for help with test-ing participants and the Australian Research Council for theirsupport.

References

Calvert, G., Spence, C., & Stein, B. (2004). Handbook of multisensory processes.Cambridge, MA: MIT Press.

Clark, C. C., & Lawless, H. T. (1994). Limiting response alternatives in time-intensityscaling: an examination of the halo-dumping effect. Chemical Senses, 19,583–594.

Conrad, F. G., Brown, N. B., & Cashman, M. (1998). Estimating the frequency ofevents from unnatural categories. Memory and Cognition, 31, 552–562.

Dalton, P., Doolittle, N., Nagata, H., & Breslin, P. A. S. (2000). The merging of thesenses: integration of subthreshold taste and smell. Nature Neuroscience, 3,431–432.

Delwiche, J., & Heffelfinger, A. L. (2005). Cross-modal additivity of taste and smell.Journal of Sensory Studies, 20, 512–525.

Dravnieks, A. (1985). Atlas of odor character profiles. PA: American Society for Testingand Materials.

Elgart, B. Z., & Marks, L. E. (2006). Detection of weak gustatory-olfactory flavormixtures. Chemical Senses, 31, A49.

Frank, R. A., & Byram, J. (1988). Taste-smell interactions are tastant and odorantdependent. Chemical Senses, 13, 445–455.

Freund, S. J., & Hasher, L. (1989). Judgments of category size: Now you have them,now you don’t. American Journal of Psychology, 102, 333–353.

Hancock, P. (2006). Monozygotic twins’ colour-number association: A case study.Cortex, 42, 147–150.

Harper, R., Land, D. G., Griffiths, N. M., & Bate-Smith, E. C. (1968). Odour qualities: Aglossary of usage. British Journal of Psychology, 59, 231–252.

Harris, J. A., & Thein, T. (2005). Interactions between conditioned and unconditionedflavor preferences. Journal of Experimental Psychology: Animal Behavior Processes,31, 407–417.

Hintzman, D. L. (2001). Similarity, global matching, and judgments of frequency.Memory and Cognition, 29, 547–556.

Jones, C. M., & Heit, E. (1993). An evaluation of the total similarity principle – Effectsof similarity on frequency judgments. Journal of Experimental Psychology:Learning, Memory and Cognition, 19, 799–812.

Labbe, D., Damevin, L., Vaccher, C., Morgenegg, C., & Martin, N. (2006). Modulationof perceived taste by olfaction in familiar and unfamiliar beverages. Food Qualityand Preference, 17, 582–589.

Marks, L. E., & Osgood, E. C. (2005). Developmental constraints on theories ofsynesthesia. In L. C. Robertson & N. Sagiv (Eds.), Synesthesia: Perspectives fromcognitive neurosciences (pp. 215–236). NY: OUP.

Pfeiffer, J. C., Hollowood, T. A., Hort, J., & Taylor, A. J. (2005). Temporal synchronyand integration of sub-threshold taste and smell signals. Chemical Senses, 30,539–545.

Prescott, J., & Wilkie, J. (2007). Pain tolerance selectively increased by a sweet-smelling odor. Psychological Science, 18, 308–311.

Rankin, K. M., & Marks, L. E. (2000). Chemosensory context effects: role of perceivedsimilarity and neural commonality. Chemical Senses, 25, 747–759.

Schiffman, S. S. (2002). Taste quality and neural coding: Implications frompsychophysics and neurophysiology. Physiology and Behavior, 69, 147–159.

Stevenson, R. J. (2009). The psychology of flavour. Oxford: OUP.Stevenson, R. J., Boakes, R. A., & Prescott, J. (1998). Changes in odor sweetness

resulting from implicit learning of a simultaneous odor-sweetness association:An example of learned synesthesia. Learning and Motivation, 29, 113–132.

Stevenson, R. J., Miller, L. A., & Thayer, Z. C. (2007). Impairments in the perception ofodor-induced tastes and their relationship to impairments in taste perception.Journal of Experimental Psychology: Human Perception and Performance, 34,1183–1197.

Stevenson, R. J., & Tomiczek, C. (2007). Olfactory-induced synesthesias: A reviewand model. Psychological Bulletin, 133, 294–309.

Tussing, A. A., & Greene, R. L. (1999). The effects of stimulus similarity on frequencyjudgments of presented and nonpresented words. American Journal ofPsychology, 112, 259–275.

Valentin, D., Chrea, C., & Nguyen, D. H. (2006). Taste-odour interactions in sweettaste perception. In W. Spillane (Ed.), Optimising sweet taste in foods (pp. 66–84).Cambridge: Woodhead Publishing Limited.

Van der Klaauw, N. J., & Frank, R. A. (1996). Scaling component intensities ofcomplex stimuli: The influence of response alternatives. EnvironmentInternational, 22, 21–31.

White, T. L., & Prescott, J. (2007). Chemosensory cross-modal stroop effects:Congruent odors facilitate taste identification. Chemical Senses, 32, 337–341.

Williams, E. J. (1959). The comparison of regression variables. Journal of the RoyalStatistical Society (Series B), 21, 396–399.

Witthoft, N., & Winawer, J. (2006). Synesthetic colors determined by having coloredrefrigerator magnets in childhood. Cortex, 42, 175–183.

Yeomans, M. R., Mobini, S., Elliman, T. D., Walker, H. C., & Stevenson, R. J. (2006).Hedonic and sensory characteristics of odors conditioned by pairing withtastants in humans. Journal of Experimental Psychology: Animal BehaviorProcesses, 32, 215–228.

Yeomans, M. R., & Mobini, S. (2006). Hunger alters the expression of acquiredhedonic but not sensory qualities of food-paired odors in humans. Journal ofExperimental Psychology: Animal Behavior Processes, 32, 460–466.