B R A I N R E S E A R C H 1 4 0 6 ( 2 0 1 1 ) 1 8 – 2 9

ava i l ab l e a t www.sc i enced i r ec t . com

www.e l sev i e r . com/ loca te /b ra i n res

Research Report

Localization of brain activation by umami taste in humans

Yuko Nakamuraa, 1, Tazuko K. Gotoa,⁎, 1, Kenji Tokumoria, Takashi Yoshiurab,Koji Kobayashic, Yasuhiko Nakamurac, Hiroshi Hondab,Yuzo Ninomiyad, Kazunori Yoshiuraa

aDepartment of Oral and Maxillofacial Radiology, Faculty of Dental Science, Kyushu University, 3-1-1, Maidashi, Higashi-ku, Fukuoka, JapanbDepartment of Clinical Radiology, Graduate School of Medical Sciences, Kyushu University, 3-1-1, Maidashi, Higashi-ku, Fukuoka, JapancDepartment of Medical Technology, Kyushu University Hospital, 3-1-1, Maidashi, Higashi-ku, Fukuoka, JapandSection of Oral Neuroscience, Graduate School of Dental Science, Kyushu University, 3-1-1, Maidashi, Higashi-ku, Fukuoka, Japan

A R T I C L E I N F O

⁎ Corresponding author at: Oral Radiology, OrPhilip Dental Hospital, 34 Hospital Road, Hon

E-mail addresses: nakamura@rad.dent.ky(K. Tokumori), tyoshiu@med.kyushu-u.ac.jpyas-nkmr@r-tec.med.kyushu-u.ac.jp (Y. Naka(Y. Ninomiya), yoshiura@rad.dent.kyushu-u.1 These two authors contributed equally to

0006-8993/$ – see front matter © 2011 Elsevidoi:10.1016/j.brainres.2011.06.029

A B S T R A C T

Article history:Accepted 11 June 2011Available online 24 June 2011

There are no credible data to support the notion that individual taste qualities havededicated pathways leading from the tongue to the end of the pathway in the brain.Moreover, the insular cortex is activated not only by taste but also by non-taste informationfrom oral stimuli. These responses are invariably excitatory, and it is difficult to determinewhether they are sensory, motor, or proprioceptive in origin. Furthermore, umami is a moreunfamiliar and complex taste than other basic tastes. Considering these issues, it may beeffective to minimize somatosensory stimuli, oral movement, and psychological effects in aneuroimaging study to elicit cerebral activity by pure umami on the human tongue. For thispurpose, we developed an original taste delivery system for functional magnetic resonanceimaging (fMRI) studies for umami. Then, we compared the results produced by twoauthorized models, namely, the block design model and event-related design model, todecide the appropriate model for detecting activation by umami. Activation by the umamitaste was well localized in the insular cortex using our new system and block design modelanalysis. The peaks of the activated areas in the middle insular cortex by umami were veryclose to another prototypical taste quality (salty). Although we have to carefully interpretthe perceiving intensities and brain activations by taste from different sessions, this studydesign might be effective for detecting the accession area in the cortex of pure umami tasteon the tongue.

© 2011 Elsevier B.V. All rights reserved.

Keywords:UmamiFunctional MRITaste solution delivery deviceInsular cortex

al Diagnosis & Polyclinics, Faculty of Dentistry, The University of Hong Kong, 1B39A, Princeg Kong. Fax: +852 2858 2532.ushu-u.ac.jp (Y. Nakamura), gototk@hku.hk (T.K. Goto), toku@rad.dent.kyushu-u.ac.jp(T. Yoshiura), kokoba@med.kyushu-u.ac.jp (K. Kobayashi),mura), honda@radiol.med.kyushu-u.ac.jp (H. Honda), yuninom@dent.kyushu-u.ac.jpac.jp (K. Yoshiura).this work.

er B.V. All rights reserved.

19B R A I N R E S E A R C H 1 4 0 6 ( 2 0 1 1 ) 1 8 – 2 9

1. Introduction

Umami, discovered by Ikeda (1909), is elicited by monosodiumL-glutamate (MSG). It exists in various foods, such as soup,fish, meat, breast milk, tomatoes and some vegetables.Umami is now clearly known as the fifth basic taste (Baylisand Rolls, 1991) because it is independent from the other fourbasic tastes (sweet, salty, sour and bitter) in human tastesensation (Yamaguchi and Ninomiya, 2000) and in behavioraland electrophysiological studies in animals (Baylis and Rolls,1991; Kumazawa et al., 1991; Ninomiya and Funakoshi, 1989).In addition, the sequencing and functional expression of ataste receptor has been discovered (Chaudhari et al., 2000; Liet al., 2002; Nelson et al., 2002; Toyono et al., 2002, 2003) andthe independency of umami has been confirmed.

There are no credible data that individual taste qualitieshave dedicated pathways leading from the tongue to the endof the pathway in the brain. Especially with regard to umami,it is difficult to recognize umami itself because we do not tasteit independently in daily life. Moreover, the physiologicalsignificance of the activation of the brain by umami ispresented not only in the mouth but also in the process ofdigestion, absorption,metabolism, andother functions (Kondohet al., 2009). In addition, it is clearly known that the pathwaysfrom the tongue to the central nervous system to perceiveumami are different among species. There is evidence thatsome non-primate species (e.g., rodents) do not respondelectrophysiologically and behaviorally to umami in the sameway as humans, or they do not differentiate between MSG andNaCl (Yamamoto et al., 1988), or MSG and sweet (Heyer et al.,2003, 2004). Even anatomically, the primate taste systemmay beorganized in a different manner from that of non-primates(Beckstead et al., 1980; Norgren and Leonard, 1973). For example,inmacaques there exists an obligatory relay from the nucleus ofthe solitary tract via the taste thalamus to the taste cortex,although in rodents, there is an obligatory relay from thesolitary tract nucleus to the pontine taste area, which in turnprojects to the thalamus (Norgren and Leonard, 1973). Even inthe primary taste cortex of macaques, single neurons werefound that were tuned to respond best to glutamate (umamitaste) (Baylis and Rolls, 1991); on the other hand, in rhesusmonkeys, cluster analysis did not indicate that MSG washandled as a separate taste quality (Hellekant et al., 1997).Considering the differences across mammalian orders, we mayassume that the human taste system would differ from eventhat of monkeys.

Focusing on the primary taste cortices in the non-humanprimate, there are valuable studies which traced a singleneuron from the mouth to the central nervous system. Thesestudies showed that the primary taste cortices in the non-human primate are mainly involved in functions other thangustation. An article states that only 823 (6.5%) of the neuronstestedwere reliably responsive to one ormore of the four basictastes. The other 3014 (23.8%) responded duringmovements ofthemouth and jaw. These responses are invariably excitatory,and it was difficult to determine whether they were sensory,motor, or proprioceptive in origin (Scott and Plata-Salamán,1999). In addition, the primary taste cortex in primatescontains not only taste neurons, but also other neurons that

encode oral somatosensory stimuli including viscosity, fattexture, temperature, and capsaicin inputs (Rolls, 2006;Verhagen et al., 2004).

The human insula represents not only taste but also thenon-taste information of oral stimuli as the results offunctional neuroimaging studies show. They are odorants(de Araujo et al., 2003c; Lombion et al., 2009), texture (DeAraujo and Rolls, 2004), and temperature (Guest et al., 2007).Umami and odor (McCabe and Rolls, 2007) in particularinfluence each other strongly. These functional pathwaysand localization in the taste cortex are more complex inhumans. However, since the single neuron experimentscannot be conducted on humans, precise brain mapping byfunctional magnetic resonance imaging (fMRI) is one of themost promisingmethods for extracting the result of activationin the brain by stimuli only from umami taste on the tongue.In most previous fMRI studies of umami, small volumes ofsolutions (0.75 or 2.0 ml) have been delivered to the partici-pant's mouth through some tubes, and then the participantwas cued by visual or tone signals to move his or her tongueand swallow the taste solution (de Araujo et al., 2003a;Grabenhorst and Rolls, 2008; Grabenhorst et al., 2008; McCabeand Rolls, 2007; Schoenfeld et al., 2004). These tasks are wellestablished for the experiments that are designed to investi-gate umamiwith odor, attention, the combination of taste andpsychological effects, and then to prove the natural andscientific features of umami in humans. In fact, it is knownthat a participant's attention and psychological effect modu-late brain activation (Bender et al., 2009; Grabenhorst andRolls, 2008, 2010; Veldhuizen et al., 2007).

From another physiological standpoint, if we performed asimple experiment that minimized the somatosensory stim-uli, oral movement, and psychological effects as much aspossible in a neuroimaging study, it might be effective indetecting the accession area in the cortex of pure umami tasteon the tongue. For this, wewould like to avoid the participant'sswallowing to reduce head movement, tongue movement,and brain activation by umami in the gastrointestinal tract(Kondoh et al., 2009; Tsurugizawa et al., 2009).

Our previous taste delivery system, which prevented theparticipant's swallowing, solved these difficulties (Kami et al.,2008); however, the system delivered the taste solution only toa small area on the tongue tip. The system was valuable forsweet; however, it can be assumed that the stimulated areawas too small for umami, and it is necessary to stimulate thelateral and posterior area of the tongue also. Therefore, it isindispensable to develop a new system to investigate the brainactivation caused by pure umami taste on the tongue.

After we develop the system, we then need to investigatewhich typeofmethodof analysis ismore adequate for acquiringmore significant results. For this, the activatedareas in the braincortex by taste should be compared using two authorizedmodels: a block design model and an event-related designmodel. The block design is efficient for long-lasting activationbased on maintaining cognitive engagement in a task bypresenting stimuli sequentially within a condition, alternatingthis with other moments (epochs) when a different condition ispresented (Amaro andBarker, 2006). The event-related design ismodeled in terms of responses to instantaneous events and isefficient for transient activation. Block or event-related design

Table 1 – Brain activations by umami and salty resulting from whole brain analysis.

Brain region x y z Z value t value PFWE-corrected k

Block designUmamiRight insula (middle) 40 −4 15 5.30 8.20 0.005 7

SaltyRight rolandic opercular 52 −2 15 5.28 8.14 0.006 2

Event-related designUmamiNo significant voxel

SaltyNo significant voxel

x, y, z (mm): the Montreal Neurological Institute (MNI) coordinates of the peak voxel in the activated clusters.k: the cluster size, corrected FWE (P<0.05).

20 B R A I N R E S E A R C H 1 4 0 6 ( 2 0 1 1 ) 1 8 – 2 9

hasbeenused ineachprevious fMRI study for taste;however,nostudy has compared the results from each design. In our study,we delivered the taste solution continuously for 6 s and theparticipants were asked to identify umami at a continuousconcentration. Since the efficiency of each model dependsprimarily on the duration of the stimulation periods chosen, wehypothesized that the block design analysis would be moreappropriate to detect brain activation by umami.

The purposes of this study were to (1) investigate thecortical responses to pure umami taste (minimizing sensa-tionsother than taste asmuchaswecould) using anewsystemfor fMRI of umami, and (2) decide the appropriate designmodelfor data analysis. The fMRI methods to detect the accessionarea of pure umami taste would be established by (1) and (2).The resultswill enable us to accumulate a database of the tasteresponses of healthy young adult subjects.

2. Results

2.1. The change of the pressure in the oral cavity duringthe solution flow

There was no change in the pressure while solution contin-uously flowed on the tongue, and the change in intraoral

Table 2 – Brain activations by umami and salty resulting from R

Brain region x y z

Block designUmamiRight insula (middle) 40 −4 15Left insula (middle) −33 −10 12

SaltyRight insula (middle) 37 −4 12Left insula (middle) −38 −10 12

Event-related designUmamiNo significant voxelSaltyRight insula (middle) 42 −4 12Left insula (middle) −36 −7 15

x, y, z (mm): the Montreal Neurological Institute (MNI) coordinates of thek: the cluster size, corrected FWE (P<0.05).

pressure occurred only when the solution flow from the tubeswas switched by solenoid valves. The change in the pressurein the oral cavity showed a mean of 212.1±43.2 Pa (mean±SD)during the 8 times the delivery solution was switched.

2.2. Rating of the intensity of each taste

The taste intensity ratings just after each session were 5.8±1.7(mean±SD) for umami and 6.1±1.2 for salty. There was nosignificant difference between them (P>0.05, paired t-test).

2.3. Head movements

The head movements of the participants were less than1.7 mm in any direction or 3.0° by rotation throughout allexperimental sessions in our study.

2.4. Brain activations by taste

2.4.1. Localized activation in the whole brainThe cortical activations by taste in the group analysis of thewhole brain analysis are shown in Table 1. The activated areasby umami were localized in the insular cortex in the blockdesign analysis. No activation by umami or salty was observedin the event-related design analysis.

OI analysis.

Z value t-value PFWE-corrected k

5.30 8.20 0.000 324.59 6.34 0.007 12

4.63 6.43 0.006 114.12 5.34 0.041 1

4.46 6.04 0.011 124.24 5.57 0.024 1

peak voxel in the maximally activated clusters.

21B R A I N R E S E A R C H 1 4 0 6 ( 2 0 1 1 ) 1 8 – 2 9

2.4.2. Region-of-interest (ROI) analysisThe peaks of the activated areas by umami and salty werelocated in the middle insula, and they were very close in

Umami

Salty

Block design analysis

RightLeft

RightLeft

[40, -4, 15][-33, -10, 12]

]21,4-,73[

z = 5.30z = 4.59

36.4=z

Scan

% B

old

sign

al c

hang

e

1 2 3 4 5 6 7 1 2 3 4 5 6 7

[-38, -10, 12] z = 4.12

0

0.05

0.10

0.15

0.20

-0.05

-0.10

0

0.05

0.10

0.15

0.20

-0.05

-0.10

0

0.05

0.10

0.15

0.20

-0.05

-0.10

0

0.05

0.10

0.15

0.20

-0.05

-0.101 2 3 4 5 6 7 1 2 3 4 5 6 7

Fig. 1 – Brain activations in the insular cortex by umami and saltactivations in the insular cortex which are significant at P<0.05were localized in themiddle insular cortex. In the event-related dactivated by salty were located in the middle insula both in the bNeurological Institute (MNI) coordinates of the peak voxel in the msignal for tastes within regions of interest are shown as graphs.participants are shown in every 3-seconds bin.

proximity in the 3D coordinates (Table 2, Fig. 1). Thesignificantly activated areas by umami were localized in themiddle insula in the results from the block design analysis, but

Event-related design analysis

No significant activation

RightLeft]21,4-,24[[-36, -7, 15] 64.4=zz = 4.24

0

0.05

0.10

0.15

0.20

-0.05

-0.10

0

0.05

0.10

0.15

0.20

-0.05

-0.101 2 3 4 5 6 7 1 2 3 4 5 6 7

y. Group analysis: random effect analysis, n=20. The(FWE corrected) were shown. The activated areas by umamiesign, the activation by umami was not significant. The areaslock and event-related design analyses. [x, y, z]: the Montrealaximally activated clusters. The percent change in the BOLD

The means and the standard error of the mean for 20

22 B R A I N R E S E A R C H 1 4 0 6 ( 2 0 1 1 ) 1 8 – 2 9

no significant activation was seen in the event-related designanalysis. The areas activated by salty were located in themiddle insula in both the block and event-related designanalyses (Table 2, Fig. 1).

2.4.3. HandednessIn this study, nineteen participants (9 males and 10 females)were right handed, and one male writes with his right handand eats with his left hand. His brain activations did not showany different activation patterns to those of other participants.

2.5. Preliminary studies of the system

2.5.1. The duration of the stimulation periodIn sensory evaluations with each participant wearing his orher own intra-oral device, participants perceived umami onand over 6 s and the sensation began to plateau. Shorter ON-durations (e.g. 2 or 3 s) were inappropriate to induce maxi-mum intensity of taste. Next, we compared the brainactivations by 6 s and 9 s of ON-duration. The fMRI resultsshowed that there was no apparent difference in theactivations in the insular cortex in two participants. Since itis desirable that total experimental time is short to avoidparticipants' fatigue, we selected the ON-duration of 6 s. Onthe other hand, for the tasteless solution, the duration of 15 sfor tasteless solution was deemed optimal in washing out theprevious taste sensation and to avoid the effect of habituationand preparatory actions. Considering these, 6 s for taste and15 s for tasteless solution were taken as optimal durations fordelivery of solution.

2.5.2. Responses to different stimulus concentrationsThe brain responses to different concentrations were seen inthe same anatomical cluster in the insula. The result inducedby the higher concentration showed a greater Z statistic value(z value from SPM analysis for the peak voxel) than thatinduced by the lower concentration: Z values of activation byhigh and low concentrations were 4.32 and 2.90, respectively,in the left insula and 4.68 and 2.82, respectively, in the rightinsula in one participant. The corresponding Z values were2.72 and 2.41 in the left insula and no significant activation inthe right insula of another participant. Although there was aninter-individual difference in the absolute Z value, the brainactivationswere stronger in the same anatomical region in thesame individual for the higher concentration.

2.5.3. Test–retest reliabilityThe results on two different days, 2 months apart showed thatthe activated areas by umami in the insular cortex were veryclose, that is, within a 3-mm range on x, y and z axes (p<0.005uncorrected), although there were some differences between Zvalues on the two days. Activationswere seen in the left insularcortex with peak coordinates at [−40, −7, 5] (Z=4.64) on the firstday and [−38, −10, 8] (Z=2.84) on the second day; in the rightinsular cortex, the peak coordinates were [40, −4, 8] (Z=3.65) onthe first day and [37, −4, 10] (Z=2.80) on the second day. Theactivatedareaby saltyhadexactly thesamecoordinatesonboththe first and second days in the left insular cortex [−38, −10, 5](Z=3.99 and 4.02), and there was no significant activation in theright insular cortex on the second day.

3. Discussion

3.1. New taste delivery system for umami

The new taste delivery system for umami proved to besuccessful in localizing the taste areas in the insular corticesin humans. Participants could recognize the umami tastewhile wearing the intra-oral device during fMRI experiments,as the ratings was 5.8±1.7 (mean±SD), and there were nosignificant differences in the rating value between umami andsalty.

The device used in this study had several features. First, itconstantly provided taste on the dorsal surface and lateralsides of the tongue as widely as possible, so that not onlyfungiform but also the anterior half of the foliate papillae,which were important for detecting umami, were stimulated.In addition, the targeted area was standardized on the wholetongue and the laterality or anterior-posterior differences ofthe sensation of the tongue did not affect the activation in thebrain (Fig. 2B). Second, the advantage of the design of theintra-oral device was that it greatly reduced the tactilestimulus of the surface skin of the face, lip and tongue.There is an interaction of gustatory and lingual somatosen-sory perceptions at the cortical level in humans (Cerf-Ducastelet al., 2001) and tactile stimulation of the mouth region inmonkeys (Scott and Plata-Salamán, 1999). The inlet and outlettubes in our device were fixed to the mouthpiece so that theydid not touch the face or lip. In addition, the intra-oral devicein this study did not move during the fMRI session because ahard removable mouthpiece was fabricated for each partici-pant and fitted on the maxillary and mandibular dentalarches. We also checked that participants could perform theexperiment comfortably without gagging. No participantsreported gagging during fMRI scanning; hence, mechanical,tactile or thermal stimuli by friction could not occur. Third, oursystem delivered a constant amount of solution of taste ortasteless saliva on the whole tongue throughout the wholesession without any rest period and with the flow rate andsuction rate strictly monitored and controlled during fMRIscanning. Therefore, participants did not feel unstable stimulidue to sudden changes in the solution flow. Indeed, in thesupplementary study (Section 2.1 and Section 4.3.2), there wasno change in pressure while solution continuously flowed onthe tongue, and any changes in pressure occurred only whenthe solutions in the tubes were switched by solenoid valves.The change in the pressure in the oral cavity of the participantcorresponded to 1.75% of the intraoral pressure (12,100 Pa)that is produced between palate and tongue while humansswallow saliva (Tamine et al., 2010). Finally, our systemreduced head and tongue movements by preventing swallow-ing. Head movement in fMRI causes movement artifacts. Themovement itself can impair the alignment of the images. Inaddition, even after realignment during the analysis by SPM8,there is residual movement related to the variance present inthe fMRI time-series, causing a loss of sensitivity and,potentially, also specificity (Andersson et al., 2001). It hasthus been suggested that it is best to avoid head movementitself. Furthermore, brain activation when umami is swal-lowed through the pharynx to the gastrointestinal tract

Fluid bottles

Suction

apparatus

MRI scanner room

Solenoid valves

Flow meters

Flow rate = 1.83 ml/s

Subject

Intra-oral device

Personal

computer

AnteriorPosterior

Lateral view

A Intra-oral device B The area of the tongue where the taste solution delivered

C Extra-oral device

Small holes

Small holes

tongue

Outlet tubesOutlet tubes

Inlet tubesInlet tubes

MouthpiecesMouthpieces

Fig. 2 – The taste solution delivery system. (A) The intra-oral device of the taste solution delivery system. Hard removablemouthpieces were fabricated for the maxillary and mandibular dental arches of each participant. The four solution-inlet tubeswere connected to the anterior side of themaxillary andmandibularmouthpieces. The outlet tubewas attached at the posteriorside of themouthpieces andwas connected to the suction apparatus so that subjects did not have to swallow the administeredfluids. (B) The area of the tongue where the taste solution was delivered. Distilled water that contained green food dye(KYORITSU FOODS Co., Inc., Tokyo, Japan) was provided by the intra and extra-oral devices to a participant's tongue. Thedelivered areawas standardized on thewhole tongue, and the fields of the fungiform and the anterior half of the foliate papillaewere stained. (C) The diagram of the extra-oral device of the taste solution delivery system. Four fluid bottles were held at aheight of 170 cm, and taste solutions and a tasteless control solutionwere delivered to the intra-oral device. This systemdid notuse positive pressure but a siphon effect for delivering the taste solution. Solenoid valves were driven by an originally writtenprogram on a personal computer and delivered each fluid alternately. The flow rate was controlled at 1.83 ml/s. A constantamount of solution of taste or tasteless saliva on the whole tongue has existed throughout the whole session. This extra-oraldevice was placed outside the scanner room.

23B R A I N R E S E A R C H 1 4 0 6 ( 2 0 1 1 ) 1 8 – 2 9

24 B R A I N R E S E A R C H 1 4 0 6 ( 2 0 1 1 ) 1 8 – 2 9

(Kondoh et al., 2009; Tsurugizawa et al., 2009) should beavoided. Participants also used an eye-mask and received novisual signs or sound cues in advance of the administration ofthe taste solution. The participants were lying down andrelaxed, and instructed just to receive the taste solution andconcentrate on tasting.

The extra-oral system was also originally developed andthe system delivered solutions from the fluid bottles to theintra-oral device below by gravity. As a special note, thepersonal computer of our system was connected to the MRIsystem and received the signal at the start of each volume ofthe EPI scanning to synchronize the timing of the introductionof the stimulus onto themouth with the time-series of the MRimages. If there had been a discrepancy in the timing betweenthe stimuli and MRI time-series, it would have been a seriousproblem (Van de Moortele et al., 1997). These methods wereeffective in providing the liquid stimulus to the intra-oraldevice in a constant manner.

In summary, our new taste delivery system administeredclear and constant taste stimuli to the tongue as widely aspossible and minimized most non-taste stimuli. This systemallowed us to detect the cortical responses of umami assimply, strongly and clearly as the salty taste in the insularcortices in humans.

3.2. Activated areas by umami in the brain

In the present study, activations by umami were seen in themiddle insula both in whole brain and ROI analyses. In theprevious human fMRI studies of umami, activations were seenin the anterior insula (de Araujo et al., 2003a; Grabenhorst etal., 2008; McCabe and Rolls, 2007; Schoenfeld et al., 2004) andanterior and middle insula (Grabenhorst and Rolls, 2008). Ithas been reported that the subjective ratings of the intensityof umami were correlated with the activations in the anteriorinsula (Grabenhorst et al., 2008). Furthermore, when attentionwas being paid to the taste intensity, the intensity ratingsshowed a positive correlation with the activations in theanterior and middle insular cortices (Grabenhorst and Rolls,2008). As for the non-taste oral sensation, temperature iscorrelated with the anterior insula (Guest et al., 2007), fat withthe anterior andmiddle insula (De Araujo and Rolls, 2004), andviscosity with the middle insular cortex (De Araujo and Rolls,2004). There is considerable variability in the location of thepeak response to taste across taste studies (Bender et al., 2009;Kobayakawa et al., 1999; Ogawa et al., 2005; Rolls, 2009; Smallet al., 1999; Verhagen and Engelen, 2006). Because differentstudies use different tasks, some variability may arise as afunction of the task that the subject is asked to perform(Bender et al., 2009). In addition, considering the aboveevidence in the previous studies and the experimental designin this study, perceiving taste intensity unconsciously aftertaste delivery onto the tongue might have an added effect onactivation in the middle insular cortex.

3.3. Comparison of umami and salty

The peaks of the activated areas by umami were very close inproximity in the 3D coordinates to other prototypical tastestimuli (salty) in the insular taste cortex, suggesting that the

basic perception system of umami is very similar to that ofprototypical tastant. In previous neuroimaging studies oftaste, activation by taste was reported over the insular/opercular cortex. There have been only two studies thathave compared the activated areas by umami and basictastants (de Araujo et al., 2003a; Schoenfeld et al., 2004). Theresults of Schoenfeld et al. could suggest that the humangustatory cortex is chemotopically organized; although theresponses were very stable on an intra-individual level, sixsubjects showed a high inter-individual variability of theactivations. In contrast, in de Araujo et al.'s, (2003a)study,cortical areas that were shown to be activated by sweet werealso activated by the umami. In our study, to solve the problemof the high inter-individual variability of the subjects in fMRI fortaste, the number of subjects was set at 20, and the randomeffect analysis model was employed so that results accountedfor both scan-to-scan and participant-to-participant variabilityand allowed a population inference to be made.

There might be small differences in the location ofactivated areas between umami and salty. However, we arevery careful to interpret these differences as differences oftastes, for two reasons. First, in the previous neurophysiolog-ical studies of single neurons of monkeys, there was no cleartopographic organization of taste sensitivity in some areas inthe primary taste cortex (Scott and Plata-Salamán, 1999;Yaxley et al., 1990). In addition, some neurons respond tonot only one but also two to four tastes (Yaxley et al., 1990),and cells that respond best to umami have been found in theprimary taste cortex intermingled with cells with the bestresponses to other stimuli (Baylis and Rolls, 1991). Second,although fMRI has the best spatial resolution among themapping methods of the human brain, the spatial resolutionof the results is shown in the unit of millimeters because theactivated areas are analyzed statistically, and the resolution ofthe images is produced in fMRI. Therefore, we cannot assertwhether or not the small differences are because of thedifferences of the taste. When we attempted to directlycompare umami vs. salty by analyzing both (umami — salty)and (salty — umami), there were no statistical differencesbetween activations by umami and by salty (P>0.05, FWE-corrected).

3.4. Comparison between the block and event-relateddesign

Results by umami and salty were compared by using bothstatistical models. The efficiency of each model dependsprimarily on the duration of the stimulation periods chosen. Ifthe underlying hemodynamic response does not conform tothe employed model, estimated error variance increases andthe sensitivity with which statistical inferences are mademaybe compromised (Friston et al., 1995). In our study, theactivations in the insula by umami were significant accordingto the block design analysis but not significant according tothe event-related design analysis (Tables 1 and 2, Fig. 1). Thesefindings suggest that the activation induced by the tastestimulus was sustained sufficiently for 6 s of ‘ON’ duration.

For the tasteless solution, the duration of 15 s for tastelesssolution was optimal in washing out the previous tastesensation and to avoid the effect of habituation and

25B R A I N R E S E A R C H 1 4 0 6 ( 2 0 1 1 ) 1 8 – 2 9

preparatory actions. During 15 s, oromotor signals were notintroduced because participants did not move or swallow inthis study; instead, the suction apparatus continuouslyaspirated any solution in the mouth. Considering these, 6 sfor taste and 15 s for tasteless solution were taken as optimaldurations for delivery of solution and our experimental designwhich ensured that constant conditions were repeated wasappropriate for the block design analysis.

In data analysis, we modeled the tasteless condition as thebaseline rather than a separate condition (Worsley, 2001).Therefore, we did not attempt to model the control conditionusing a shorter duration in block type of design. Ourexperimental paradigm was very simple that avoided atten-tion to signaled cues, movements of the tongue or swallowing,and so on. Hence, this approach allowed the most basic andsimple analysis model to be applied.

3.5. The reliability of the systemIn a preliminary study about the responses to differentstimulus concentrations, the brain responses to the higherconcentration of solution were stronger than those to thelower concentration in the same anatomical region, whereasthere was some inter-individual difference in the increasingratio of Z value. In another preliminary study for test–retestreliability, the peak coordinates of activated areas by umamiin the insular cortex were within a 3-mm range on x, y and zaxes, although there were some differences between Z valueson the two different days. Hence, the locations of activationsand responses were very stable even in different conditions(concentration of stimuli and different days) on an intra-individual level, while the stability of Z values may not beguaranteed. As potential limitations of this fMRI study, it isassumed that differences in sensory perception occur betweensessions. It must be acknowledged that the possibility thatsubjects do not perceive umami at the same intensity duringthe different sessions of the experiment. The habituationeffects and effect by combined taste quality in the experi-ments also have to be considered. However, statistical modelsdesigned to fit habituation effects associated with repeatedumami stimulation had not been attempted in this study.Such models could have provided a better account for whyinsula activity changes throughout the experiment. In thegroup analysis, it is possible to make a population inference ofactivated areas and Z values using the random effect analysismodel to account for both scan-to-scan and participant-to-participant variability during group analysis. However, wehave to carefully interpret the perceiving intensities of tasteand Z values from different sessions, such as different datesand/or different subjects and, in particular, in investigations ofbrain activation by taste during clinical follow-up studies inpatients with a taste disorder.

3.6. Conclusion

Our study has shown that the pathways from the tongue tothe central nervous system for perceiving pure and passiveumami reached the middle insula in healthy humans. Thepeaks of the activated areas by umami were very close toanother prototypical taste (salty). The appropriate model foranalyzing umami in this study designwas shown to be a block

design. Using these methods, a database of the activations bytwo basic tastes experienced by healthy young adults can beassembled. Those data could be useful for diagnoses ofpatients who have conditions such as taste disorders, nervepalsy, or tongue cancer.

4. Experimental procedures

4.1. Participants

Twenty healthy young volunteers (10 men and 10 women; agerange 19–29 years, mean 24.2±2.7 years) participated in thecurrent fMRI study. Nineteen participants (9 men and 10women) were right-handed, and one male writes with hisright hand and eats with his left hand. Participants were non-smokers and reported no oral or nasal complaints, allergies,tongue deformities or obstruction. They did not have anyhistory of neurological illness and were not under medication.The participants had not eaten for 3 h before the investigation.The Human Experimentation Committee of Kyushu Universi-ty approved all experimental procedures, and written in-formed consent was obtained from all participants.

4.2. Taste solution

The taste solutions were umami taste produced by 0.1 Mmonosodium glutamate (MSG) and 0.005 M inosine monopho-sphate (IMP), and salty taste (0.1 M sodium chloride). Bothsolutions presented were dissolved in distilled water. Atasteless solution containing the main ionic components ofsaliva (25 mM KCl plus 2.5 mM NaHCO3) was also used as thecontrol to remove the effects of the other stimuli produced byintroducing a fluid into the mouth (de Araujo et al., 2003b;O'Doherty et al., 2001). This solution was used becauseprevious fMRI study suggested that water elicits a tastesense (de Araujo et al., 2003b), and tasteless saliva has beenused as a control condition in other fMRI studies of umami inhumans (de Araujo et al., 2003b; Grabenhorst et al., 2008;McCabe and Rolls, 2007; O'Doherty et al., 2001; Small et al.,2003; Veldhuizen et al., 2007). Sodium chloride was used as aprototypical taste quality and taste similar to MSG becauseMSG contains sodium ions, which evoke the salty taste assodium chloride. Solutions were left for at least 5 h in the MRIscanner room so that the mean temperature of solutions atthe time of the experiment was 25.2±0.9 °C.

4.3. Taste solution delivery system

A novel system for umami that consisted of an intra-oral andan extra-oral device was developed.

4.3.1. Intra-oral deviceThe intra-oral device was designed to provide taste solutionon the dorsal and lateral surfaces of the tongue as widely aspossible so that participants could sense the unique taste ofumami as strongly as possible. In addition, this system wasdesigned to prevent swallowing during fMRI. First, hardremovable mouthpieces were fabricated to fit the maxillaryand mandibular dental arches of each participant by using a

ON (6 sec)

Control (15 sec)

ON (6 sec)

Control (15 sec)

Time

The predicted hemodynamic response function

Event-related

design

Block design

Fig. 3 – The fMRI paradigm. An experiment paradigmconsisted of 32 pairs of ON and OFF durations. The durationsof the ON (32 times for each taste) and OFF (artificial saliva)were 6 and 15 s, respectively.

26 B R A I N R E S E A R C H 1 4 0 6 ( 2 0 1 1 ) 1 8 – 2 9

plastic disc (polyethylene terephthalate G copolymer; ERKO-DENT, Pfalzgrafenweiler, Germany). Second, the four solution-inlet tubeswere connected to the anterior side of themaxillaryand mandibular mouthpieces by resin (Ortho Crystal; Nissindental products Inc., Kyoto, Japan). Third, the mandibularmouthpiece was fixed at the comfortable occlusal position byresin. Fourth, an outlet tube was attached at the posterior sideof the mouthpieces (Fig. 2A). This design allowed the tastesolutions to stimulate the dorsal surface and lateral sides ofthe tonguewithout any sense of discomfort. The area to whichour taste delivery system delivered solution to each of thepapillae fields on the tongue is shown in Fig. 2B. The end of theoutlet tube was connected to a continuous suction apparatus(Fig. 2C), and the participants were instructed not to swallowthe administered fluids. All participants were pre-trainedoutside the MRI scanner to taste the solution with theindividual intra-oral devices fixed on their dental arches.Any discomfort or problems were addressed and we con-firmed that participants could perform the experimentcomfortably without gagging. All participants reported thatthey could distinguish umami from salty taste wearing theintra-oral devices.

4.3.2. Extra-oral deviceThe extra-oral devicewas placed outside theMR scanner roombecause they contained metal. Four fluid bottles deliveredfluid from a height of 170 cm to the intra-oral device throughfour Teflon tubes, each connected to a solenoid valve. Tosynchronize the timing of the introduction of the solutionsinto the mouth with the time-series of the MR images, apersonal computer was connected to the MRI system andreceived the signal at the start of each volume of the EPIscanning. The timing and period to deliver the taste solutionsand artificial saliva in the tubes were switched by solenoidvalves that were controlled by a specially written computerprogram.

The flow rate was controlled using flowmeters andwas setat 1.83 ml/s, while the suction rate of the solution in themouth was strictly monitored and controlled during fMRIscanning (Fig. 2C). Hence, a constant amount of solution oftaste or tasteless saliva existed on the whole tongue through-out the whole fMRI session.

To confirm the change of pressure in the oral cavity of theparticipant during the experiment, a supplementary studywas performed using a U-shaped tube containing distilledwater as a manometer. One of the inlet tubes of the intra-oraldevicewas connected to the U-shaped tube and changes of theliquid level were measured while the solutions were deliveredon the tongue by our taste delivery system. The difference in1 cm of the liquid level represented a pressure change of98.1 Pa in the oral cavity.

4.4. Experimental design

4.4.1. The preliminary study of the duration of the stimulationperiodWe performed preliminary studies to determine an appropri-ate duration of taste. In sensory evaluations with eachparticipant wearing his or her own intra-oral device, partici-pants rated intensity of umami. Next, we compared the brain

activations by different ON-duration in the fMRI results. Forthe tasteless solution, the duration as optimal in washing outthe previous taste sensation and to avoid the effect ofhabituation and preparatory actions were considered.

4.4.2. Experimental designThe session included umami and salty tastes. The design isshown schematically in Fig. 3. A session consisted of 64 pairsof durations of ON (32 times for each taste) and OFF (artificialsaliva) in alternation. In the ON state, the two tastes weredelivered in a random permuted sequence. The duration ofthe ON and OFF blocks was 6 and 15 s, respectively, consid-ering the results from the preliminary studies above(Section 2.5.1).

Before the experiments, participants were instructed toidentify only taste qualities such as ‘umami’ and ‘salty’ aftertaste liquids were released onto their tongue. Participantsreceived no signals before administration of the taste solution.They were not instructed to rate the intensity or pleasantnessof the taste, and used no equipment for evaluating tasteduring the acquisition of functional imaging.

4.5. Rating of the intensity of each taste

Immediately after a session finished, the participants ratedthe perceived intensity of each taste using a visual analogscale ranging from 0 for very weak (no taste) to 10 for veryintense (‘strongest imaginable taste’). Data were then ana-lyzed with a paired t-test (SPSS 19.0 software, SPSS Inc,Chicago, USA), with a significance level of P<0.05.

4.6. fMRI data acquisition

Imageswere acquiredwith a 3.0-Tesla whole-bodyMRI scanner(Philips). Functional scans were obtained using a T2*-weightedecho-planar imaging (EPI) sequence (TR=3000ms, TE=35ms,flip angle=90°, the field of view=200 mm, matrix size=80×80pixels, voxel size=2.5 mm×2.5 mm×2.5 mm, slice thick-ness=2.5 mm, gap=1 mm). Each EPI volume consisted of 36axial slices in ascending order.

For anatomical reference, T1-weighted 3D TFE images(TR=8.2 ms, TE=3.7 ms, flip angle=8°, FOV=240 mm, voxel

27B R A I N R E S E A R C H 1 4 0 6 ( 2 0 1 1 ) 1 8 – 2 9

size=1.0 mm×1.0 mm×1.0 mm) were obtained for eachsubject.

4.7. fMRI data analysis

Image processing and data analysis were performed using theStatistical Parametric Mapping 8 (SPM8) software package (theWellcome Trust Centre for Neuroimaging, London, UK)running in the MATLAB 7.4 (The MathWorks Inc., Sherborn,MA, USA) environment. The first three volumes of functionalimages were discarded in each session owing to unsteadymagnetization, and the remaining 445 volumes per sessionwere used for the analysis.

4.7.1. Data analysis by block design modelThe following preprocessing steps to the serial imagining datawere applied: realignment, coregistration, segmentation,application of the normalization parameters to the realignedfunctional scans so that each scan was matched to theMontreal Neurological Institute (MNI) template, spatialsmoothing with a 6 mm full width at half maximum of theGaussian kernel, and high-pass filtering with a cut-off periodof 128 s. We calculated the range of head movements of theparticipants during fMRI in the step of realignment.

In a within-participant analysis, General Linear Modelswere applied to the fMRI data. The taste and control weredesigned as 6 s and 15 s box-car function, respectively,convolved with a canonical hemodynamic response function(Fig. 3). The data analysis was performed by setting the controlsolution as the baseline and measuring the taste-baselinedifference (Worsley, 2001).

4.7.2. Data analysis by event-related design modelSlice timing correction (centered at TR/2) was applied to thepreprocessing steps in the block design analysis. In a within-participant analysis, General Linear Models were applied, anda stick function at the timing of the input of the taste fluid wasconstructed (Fig. 3).

4.7.3. AnalysisIn both designs, the voxels with stimulus-related MR signalchanges were identified. Significant correlations between theobserved response and the modeled response were estimated,thus yielding t-valuemaps. The activations in the insula of eachsubjectwere investigatedandanatomically identified accordingto a brain atlas (Mai et al., 2004). Statistical parametric mapsfrom each individual data set were then entered into a groupanalysis, which was a random effects analysis that accountedfor both scan-to-scan and participant-to-participant variabilityand which allowed us to make a population inference.

In the group analysis, unpredicted peaks were consideredsignificant at P<0.05, after family wise error (FWE) correctionacross the entire brain at the voxel level. For predicted peaks,our ROI approach used a WFU PickAtlas (Maldjian et al., 2003,2004; Tzourio-Mazoyer et al., 2002) to create masks ofpredicted ROIs, including the taste-related area insula. ROIswere selected in the insular cortex based on previous studiesshowing that insular regions form the core of the gustatoryperception both in humans (Small et al., 1999) and themacaque monkey (Yaxley et al., 1990). Peaks within these

masks were considered significant at P<0.05, after FWEcorrection across the voxels of the ROI.

Finally, the localization of activated areas induced by eachtaste in the insula was evaluated. The results by block andevent-related designs were compared in order to choose areliable design for the analysis of the umami taste.

4.8. Preliminary studies of the reliability of the system

4.8.1. Responses to different stimulus concentrationsIn a preliminary study to investigate the reliability of thesystem, we performed fMRI experiments using taste solutionswith different concentrations. The brain activations inducedby the solutions of 0.05 M (Low) and 0.25 M (High) of MSG wereinvestigated in the same session for two participants (a 32-year-old man and a 23-year-old woman, both right-handed).The same protocol as used in this study was performed.

4.8.2. Test–retest reliabilityTo determine the reproducibility of the system, a subject (a 25-year-old woman, right-handed) participated in the experi-ment twice on two different days, 2 months apart. The sameprotocol as used in this study was performed on each day.

Funding

This work was supported by a Grant-in-Aid for ScientificResearch from the Ministry of Education, Japan (19390479 toTK.G.) and Society for Research on Umami Taste (to TK.G.).

Acknowledgments

We are grateful to Mr. Michael Guinn for critically reading theEnglish manuscript. We also thank Dr. Trevor Lane forproofreading the manuscript.

R E F E R E N C E S

Amaro Jr., E., Barker, G.J., 2006. Study design in fMRI: basicprinciples. Brain Cogn. 60, 220–232.

Andersson, J.L., Hutton, C., Ashburner, J., Turner, R., Friston, K.,2001. Modeling geometric deformations in EPI time series.NeuroImage 13, 903–919.

Baylis, L.L., Rolls, E.T., 1991. Responses of neurons in the primatetaste cortex to glutamate. Physiol. Behav. 49, 973–979.

Beckstead, R.M., Morse, J.R., Norgren, R., 1980. The nucleus of thesolitary tract in the monkey: projections to the thalamus andbrain stem nuclei. J. Comp. Neurol. 190, 259–282.

Bender, G., Veldhuizen,M.G., Meltzer, J.A., Gitelman, D.R., Small, D.M.,2009. Neural correlates of evaluative compared with passivetasting. Eur. J. Neurosci. 30, 327–338.

Cerf-Ducastel, B., Van de Moortele, P.F., MacLeod, P., Le Bihan, D.,Faurion, A., 2001. Interaction of gustatory and lingualsomatosensory perceptions at the cortical level in the human:a functional magnetic resonance imaging study. Chem. Senses26, 371–383.

Chaudhari, N., Landin, A.M., Roper, S.D., 2000. A metabotropicglutamate receptor variant functions as a taste receptor. Nat.Neurosci. 3, 113–119.

28 B R A I N R E S E A R C H 1 4 0 6 ( 2 0 1 1 ) 1 8 – 2 9

de Araujo, I.E., Kringelbach, M.L., Rolls, E.T., Hobden, P., 2003a.Representation of umami taste in the human brain.J. Neurophysiol. 90, 313–319.

de Araujo, I.E., Kringelbach, M.L., Rolls, E.T., McGlone, F., 2003b.Human cortical responses to water in the mouth, and theeffects of thirst. J. Neurophysiol. 90, 1865–1876.

de Araujo, I.E., Rolls, E.T., Kringelbach, M.L., McGlone, F., Phillips,N., 2003c. Taste-olfactory convergence, and the representationof the pleasantness of flavour, in the human brain. Eur. J.Neurosci. 18, 2059–2068.

De Araujo, I.E., Rolls, E.T., 2004. Representation in the human brainof food texture and oral fat. J. Neurosci. 24, 3086–3093.

Friston, K.J., Holmes, A.P., Poline, J.-B., Grasby, P.J., Williams, S.C.,Frackowiak, R.S., Turner, R., 1995. Analysis of fMRI time-seriesrevisited. NeuroImage 2, 45–53.

Grabenhorst, F., Rolls, E.T., 2010. Attentional modulation ofaffective versus sensory processing: functional connectivityand a top-down biased activation theory of selective attention.J. Neurophysiol. 104, 1649–1660.

Grabenhorst, F., Rolls, E.T., 2008. Selective attention to affectivevalue alters how the brain processes taste stimuli. Eur. J.Neurosci. 27, 723–729.

Grabenhorst, F., Rolls, E.T., Bilderbeck, A., 2008. How cognitionmodulates affective responses to taste and flavor: top-downinfluences on the orbitofrontal and pregenual cingulatecortices. Cereb. Cortex 18, 1549–1559.

Guest, S., Grabenhorst, F., Essick, G., Chen, Y., Young, M., McGlone,F., de Araujo, I., Rolls, E.T., 2007. Human cortical representationof oral temperature. Physiol. Behav. 92, 975–984.

Hellekant, G., Ninomiya, Y., Danilova, V., 1997. Taste inchimpanzees II: single chorda tympani fibers. Physiol. Behav.61, 829–841.

Heyer, B.R., Taylor-Burds, C.C., Mitzelfelt, J.D., Delay, E.R., 2004.Monosodium glutamate and sweet taste: discriminationbetween the tastes of sweet stimuli and glutamate in rats.Chem. Senses 29, 721–729.

Heyer, B.R., Taylor-Burds, C.C., Tran, L.H., Delay, E.R., 2003.Monosodium glutamate and sweet taste: generalization ofconditioned taste aversion between glutamate and sweetstimuli in rats. Chem. Senses 28, 631–641.

Ikeda, K., 1909. New seasonings. J. Tokyo Chem. Soc. 30, 820–836[in Japanese].

Kami, Y.N., Goto, T.K., Tokumori, K., Yoshiura, T., Kobayashi, K.,Nakamura, Y., Honda, H., Ninomiya, Y., Yoshiura, K., 2008. Thedevelopment of a novel automated taste stimulus deliverysystem for fMRI studies on the human cortical segregation oftaste. J. Neurosci. Methods 172, 48–53.

Kobayakawa, T., Ogawa, H., Kaneda, H., Ayabe-Kanamura, S.,Endo, H., Saito, S., 1999. Spatio-temporal analysis of corticalactivity evoked by gustatory stimulation in humans. Chem.Senses 24, 201–209.

Kondoh, T., Tsurugizawa, T., Torii, K., 2009. Brain functionalchanges in rats administered with monosodium L-glutamatein the stomach. Ann. N. Y. Acad. Sci. 1170, 77–81.

Kumazawa, T., Nakamura, M., Kurihara, K., 1991. Canine tastenerve responses to umami substances. Physiol. Behav. 49,875–881.

Li, X., Staszewski, L., Xu, H., Durick, K., Zoller, M., Adler, E., 2002.Human receptors for sweet and umami taste. Proc. Natl. Acad.Sci. U. S. A. 99, 4692–4696.

Lombion, S., Comte, A., Tatu, L., Brand, G., Moulin, T., Millot, J.L.,2009. Patterns of cerebral activation during olfactory andtrigeminal stimulations. Hum. Brain Mapp. 30, 821–828.

Mai, J.K., Assheuer, J., Paxinos, G. (Eds.), 2004. Atlas of the humanbrain, second ed. Academic Press, San Diego, pp. 24–29, 54–64,80–81.

Maldjian, J.A., Laurienti, P.J., Burdette, J.H., 2004. Precentral gyrusdiscrepancy in electronic versions of the Talairach atlas.NeuroImage 21, 450–455.

Maldjian, J.A., Laurienti, P.J., Kraft, R.A., Burdette, J.H., 2003. Anautomated method for neuroanatomic and cytoarchitectonicatlas-based interrogation of fMRI data sets. NeuroImage 19,1233–1239.

McCabe, C., Rolls, E.T., 2007. Umami: a delicious flavor formed byconvergence of taste and olfactory pathways in the humanbrain. Eur. J. Neurosci. 25, 1855–1864.

Nelson, G., Chandrashekar, J., Hoon, M.A., Feng, L., Zhao, G., Ryba,N.J., Zuker, C.S., 2002. An amino-acid taste receptor. Nature416, 199–202.

Ninomiya, Y., Funakoshi, M., 1989. Peripheral neural basis forbehavioural discrimination between glutamate and the fourbasic taste substances in mice. Comp. Biochem. Physiol. AComp. Physiol. 92, 371–376.

Norgren, R., Leonard, C.M., 1973. Ascending central gustatorypathways. J. Comp. Neurol. 150, 217–237.

O'Doherty, J., Rolls, E.T., Francis, S., Bowtell, R., McGlone, F., 2001.Representation of pleasant and aversive taste in the humanbrain. J. Neurophysiol. 85, 1315–1321.

Ogawa, H., Wakita, M., Hasegawa, K., Kobayakawa, T., Sakai, N.,Hirai, T., Yamashita, Y., Saito, S., 2005. Functional MRIdetection of activation in the primary gustatory cortices inhumans. Chem. Senses 30, 583–592.

Rolls, E.T., 2009. Functional neuroimaging of umami taste:what makes umami pleasant? Am J Clin. Nutr. 90,804S–813S.

Rolls, E.T., 2006. Brain mechanisms underlying flavour andappetite. Philos. Trans. R Soc. Lond. B 361, 1123–1136.

Schoenfeld, M.A., Neuer, G., Tempelmann, C., Schüssler, K.,Noesselt, T., Hopf, J.M., Heinze, H.J., 2004. Functionalmagnetic resonance tomography correlates of tasteperception in the human primary taste cortex.Neuroscience 127, 347–353.

Scott, T.R., Plata-Salamán, C.R., 1999. Taste in the monkey cortex.Physiol. Behav. 67, 489–511.

Small, D.M., Gregory, M.D., Mak, Y.E., Gitelman, D., Mesulam,M.M.,Parrish, T., 2003. Dissociation of neural representation ofintensity and affective valuation in human gustation. Neuron39, 701–711.

Small, D.M., Zald, D.H., Jones-Gotman, M., Zatorre, R.J., Pardo, J.V.,Frey, S., Petrides, M., 1999. Human cortical gustatory areas: areview of functional neuroimaging data. Neuroreport 10, 7–14.

Tamine, K., Ono, T., Hori, K., Kondoh, J., Hamanaka, S., Maeda, Y.,2010. Age-related changes in tongue pressure during swallowing.J. Dent. Res. 89, 1097–1101.

Toyono, T., Seta, Y., Kataoka, S., Harada, H., Morotomi, T., Kawano,S., Shigemoto, R., Toyoshima, K., 2002. Expression of themetabotropic glutamate receptor, mGluR4a, in the taste hairsof taste buds in rat gustatory papillae. Arch. Histol. Cytol. 65,91–96.

Toyono, T., Seta, Y., Kataoka, S., Kawano, S., Shigemoto, R.,Toyoshima, K., 2003. Expression of metabotropic glutamatereceptor group I in rat gustatory papillae. Cell Tissue Res. 313,29–35.

Tsurugizawa, T., Uematsu, A., Nakamura, E., Hasumura, M.,Hirota, M., Kondoh, T., Uneyama, H., Torii, K., 2009.Mechanisms of neural response to gastrointestinal nutritivestimuli: the gut-brain axis. Gastroenterology 137, 262–273.

Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Crivello, F.,Etard, O., Delcroix, N., Mazoyer, B., Joliot, M., 2002. Automatedanatomical labeling of activations in SPM using a macroscopicanatomical parcellation of the MNI MRI single-subject brain.NeuroImage 15, 273–289.

Van de Moortele, P.F., Cerf, B., Lobel, E., Paradis, A.L., Faurion, A.,Le Bihan, D., 1997. Latencies in fMRI time-series: effect ofslice acquisition order and perception. NMR Biomed. 10,230–236.

Veldhuizen, M.G., Bender, G., Constable, R.T., Small, D.M., 2007.Trying to detect taste in a tasteless solution: modulation of

29B R A I N R E S E A R C H 1 4 0 6 ( 2 0 1 1 ) 1 8 – 2 9

early gustatory cortex by attention to taste. Chem. Senses 32,569–581.

Verhagen, J.V., Engelen, L., 2006. The neurocognitive bases ofhuman multimodal food perception: sensory integration.Neurosci. Biobehav. Rev. 30, 613–650.

Verhagen, J.V., Kadohisa, M., Rolls, E.T., 2004. Primate insular/opercular taste cortex: neuronal representations of theviscosity, fat texture, grittiness, temperature, and taste offoods. J. Neurophysiol. 92, 1685–1699.

Worsley, K.J., 2001. Statistical analysis of activation images. In:Jezzard, P., Matthews, P.M., Smith, S.M. (Eds.), Functional MRI:

An Introduction to Methods. Oxford University Press, Oxford,pp. 251–270.

Yamaguchi, S., Ninomiya, K., 2000. Umami and food palatability.J. Nutr. 130, 921S–926S.

Yamamoto, T., Matsuo, R., Kiyomitsu, Y., Kitamura, R., 1988. Tasteeffects of ‘umami’ substances in hamsters as studied byelectrophysiological and conditioned taste aversiontechniques. Brain Res. 451, 147–162.

Yaxley, S., Rolls, E.T., Sienkiewicz, Z.J., 1990. Gustatory responsesof single neurons in the insula of the macaque monkey.J. Neurophysiol. 63, 689–700.