Sweet Thermal Taste: Perceptual Characteristics in Water and Dependence on TAS1R2/TAS1R3

Abstract

The initial objective of this study was to determine if activation of the sweet taste receptor TAS1R2/TAS1R3 is necessary for perception of sweet thermal taste (swTT). Our approach was to inhibit the receptor with the inverse agonist lactisole using a temperature-controlled flow gustometer. Because all prior studies of thermal taste (TT) used metal thermodes to heat the tongue tip, we first investigated whether it could be generated in heated water. Experiment 1 showed that sweetness could be evoked when deionized water was heated from 20 to 35 °C, and testing with static temperatures between 20 and 35 °C demonstrated the importance of heating from a cool temperature. As in previous studies, thermal sweetness was reported by only a subset of participants, and replicate measurements found variability in reports of sweetness across trials and between sessions. Experiment 2 then showed that exposure to 8 mM lactisole blocked perception of swTT. Confirmation of the involvement of TAS1R2/TAS1R3 led to an investigation of possible sensory and cognitive interactions between thermal and chemical sweetness. Using sucrose as a sweet stimulus and quinine as a nonsweet control, we found that dynamic heating capable of producing thermal sweetness did not increase the sweetness of sucrose compared with static heating at 35 °C. However, swTT was disrupted if trials containing sucrose (but not quinine) were interspersed among heating-only trials. These findings provide new information relevant to understanding the perceptual processes and receptor mechanisms of swTT, as well as the heat sensitivity of sweet taste in general.

Introduction

Thermal taste (TT) is the illusion of taste stimulation produced by temperature alone. When it was first reported (Cruz and Green 2000), TT was proposed to result from the temperature sensitivity of the taste nerves, which had previously been demonstrated for the chorda tympani and glossopharyngeal nerves in nonhuman animals (Ogawa et al. 1968; Sato et al. 1975; Travers and Smith 1984) and the chorda tympani nerve in humans (Oakley 1985). It was hypothesized (Cruz and Green 2000) that the thermal sensitivity of a G-protein-coupled receptor (GPCR), for which there was evidence of involvement in sweet, bitter, and umami taste (Chaudhari et al. 1996; Wong et al. 1996; Hoon et al. 1999), might be responsible for sweet thermal taste (swTT), and that cold-sensitive Na⁺ and H⁺ channels in taste cells responsive to sodium salts and acids (Lindemann 1996) might account for the less frequently reported salty and sour TTs.

A pair of studies later provided qualified support for both of these hypotheses, though more strongly for sweetness than for the other tastes. In the first study, Askwith et al. (2001) reported that the epithelial sodium channel (ENaC) in rats is sensitive to cold and suggested that changes in the conductivity of this channel might explain thermal salty and/or sour taste. But subsequent studies indicated the contribution of ENaC to human salt taste is relatively minor (Ossebaard et al. 1997; Talavera et al. 2007; Roper 2015), and that its role in sour taste is uncertain (Munger 2016). In the second study, Talavera et al. (2005) found that heating increased the sensitivity of TRPM5, a calcium-gated, nonspecific cation channel that is the final step in the transduction cascade of GPCRs (Hofmann et al. 2003;Perez et al. 2003; Zhang et al. 2007; Liman 2014). This finding stands as the best evidence of a mechanism of thermal sensitivity in human taste, particularly for the most frequently reported swTT. However, because bitter and umami receptors are also mediated by GPCRs (Zhang et al. 2003), it is unclear why heating produces more reports of sweetness than bitterness (e.g., Cruz and Green 2000; Bajec and Pickering 2008b), and why umami has rarely been reported (Skinner et al. 2018). These results suggest that the heat sensitivity TRPM5 may not be the only mechanism involved in swTT.

Subsequent psychophysical studies in humans have generally not focused on possible mechanisms of TT. Most have investigated TT as an approach for identifying individuals who are sensitive to chemical tastes and flavors (e.g., Bajec and Pickering 2008b; Pickering et al. 2010; Yang et al. 2014; Pickering and Klodnicki 2016). These studies, which have compared TT and 6-n-propylthiouracil (PROP) (e.g., Bartoshuk et al. 1994; Mennella et al. 2011) as predictors of sensitivity, were motivated by evidence of a relationship between the ability to perceive swTT and the sensitivity to taste but not to chemesthetic or other somatosensory stimuli (Green and George 2004; Green et al. 2005). Although significant associations between TT and chemical taste continue to be reported (e.g., Skinner et al. 2018; Thibodeau et al. 2018), variation across studies in the frequency of TT, the range of tastes and other sensations reported, and the temperatures at which they occur, have raised questions about how much of the variation is inherent in the phenomenon and how much arises from differences in experimental methods and procedures. For example, functional magnetic resonance imaging study of TT (Hort et al. 2016) that used a novel oral appliance to hold a Peltier thermode against the tongue tip during imaging found that sweetness was reported more often during cooling than during heating, and that 2 of the most commonly reported sensations were “minty” and “metallic.” The study also found significant differences in response to temperature between thermal “tasters” and “non-tasters” in somatosensory but not gustatory cortex. A recent study (Skinner et al. 2018) that used the same oral appliance found that response variability in TT for some individuals was so high that mean taste ratings could not be calculated across trials. The results also included frequent ratings of “metallic” and “minty” as well as “astringent” and “spicy.” The variability they found led the authors to suggest that experimental procedures and response criteria used to study TT need to be revisited.

We began the present study with the primary goal of investigating the role of the sweet taste receptor TAS1R2/TAS1R3 in swTT. As noted above, the more frequent perception of swTT compared with bitter and umami tastes raised the question of whether the temperature sensitivity of TRPM5 alone can explain the thermal sensitivity of the transduction cascades of taste GPCRs. In fact, Talavera et al. (2005) proposed that TRPM5 functions as a “coincidence detector” of increases in both intracellular Ca²⁺ and heat, and suggested that additional study was needed to determine if “the combination of high temperatures and basal levels of intracellular Ca²⁺ is sufficient to activate TRPM5 and trigger the sweet sensation” (Talavera et al. 2007, p. 380).

To begin to address this question in humans we set out to determine if perception of swTT requires TAS1R2/TAS1R3 to be in an active state. Our approach was to use lactisole, a sweet taste inhibitor and inverse agonist of TAS1R2/TAS1R3 (Jiang et al. 2005; Galindo-Cuspinera et al. 2006), to block receptor activation. To do so required producing swTT in water so that lactisole could be presented to the tongue tip in an aqueous solution. Although water is a far more natural context in which to perceive taste than the surface of a thermode, it was not obvious that swTT could be generated in water. First, when presented by itself, water activates gustatory regions of the human brain, including the primary taste cortex (O’Doherty et al. 2001; de Araujo et al. 2003). The mechanism and significance of this activation is unclear, but speculation that it might be a gustatory signal for water (de Araujo et al. 2003) raised the possibility that the activation could disrupt the illusory perception of sweetness. In addition, it was unknown whether the much lower thermal conductivity of water (Chol and Estman 1995) compared with the metal surface of a thermode might make it difficult to achieve the parameters of temperature change on the tongue that are necessary to induce sweetness.

However, informal testing using a specially designed temperature-controlled flow gustometer (TFG) quickly indicated that swTT could indeed be perceived, though somewhat less strongly than has been reported with metal thermodes. Experiment 1 was therefore designed to first study the incidence, intensity, and variability of swTT in water before investigating the ability of lactisole to block it. Static warm and cold temperatures were included to evaluate the thermal specificity of swTT and to serve as control conditions for the effects of heating the tongue tip from a cool temperature. After finding that swTT was perceived at a frequency similar to prior studies, experiment 2 confirmed that TAS1R2/TAS1R3 plays an essential role in swTT. Because the dependency on TAS1R2/TAS1R3 raised the possibility that interactions might occur during simultaneous chemical and thermal stimulation, a third experiment was initially designed with the goal of determining if heating that can produce swTT can also enhance the sweetness of sucrose. When preliminary testing indicated that prior exposure to sucrose interfered with subsequent perception of swTT, measurement of this potential context effect became another goal of the experiment.

Experiment 1: swTT in water

Methods

Participants

A total of 28 adults (19 females) were recruited via public postings on the Yale Medical School and Yale College campuses. All were fluent English-speakers between 18 and 45 years of age who were self-reported to be healthy, nonpregnant, nonsmokers with no known taste or smell disorders or deficiencies and had no lip, cheek, or tongue piercings. Participants were asked to refrain from eating or drinking foods or beverages for at least 1 h prior to their scheduled session. All gave written informed consent and were paid for their participation. The research was conducted in accordance with the principles expressed in the Declaration of Helsinki, and the research protocol was approved by the Human Investigations Committee of the Yale University IRB.

Stimuli

In this experiment thermal stimuli included static water temperatures of 15, 20, 30, and 35 °C, all 15 s in duration, and a dynamic heating stimulus that increased water temperature from 20 to 35 °C @ 1 °C/s, also over a duration of 15 s. These and all thermal stimuli throughout the study were delivered in deionized water to the tongue tip at a flow rate of 3 mL/s using a temperature controlled flow gustometer (TFG). The TFG was designed and built in the John B. Pierce Laboratory electronics and machine shop and is controlled by LabVIEW software that enables delivery of water and taste solutions at controlled temperatures and flow rates. Solutions are pumped from 4-L glass reservoir bottles through 2 inline Peltier heating and cooling chambers before being delivered to the base of a machined Teflon tongue bath (~4-mL volume). The solution flows up through a mixing screen and over the rim of the bath into a drain, thereby keeping solution depth constant across flow rates. A LabVIEW virtual interface enabled 1) selection from 5 different taste stimuli and deionized H₂O in the reservoir bottles via electronic pinch valves; 2) setting solution flow rates via variable speed peristaltic pumps; and 3) setting steady-state solution temperatures and/or ramp rates and target temperatures. Solution temperature is controlled via a proportional-integrative-derivative (PID) loop and monitored by a thermocouple in the delivery tube at the base of the bath. Steady-state temperatures and target temperatures were set before each trial.

Training and practice procedure

In this and subsequent experiments, participants who had not previously served in taste experiments in the laboratory attended a short practice session prior to data collection to receive training and practice in the use of the general version of the Labeled Magnitude Scale (gLMS; Green et al. 1993, 1996; Bartoshuk et al. 2004) to rate perceived taste intensity. The gLMS was displayed on a computer monitor, and subjects used a mouse to move a cursor to appropriate locations on the scale to indicate perceived intensity. After receiving instructions, participants first practiced using the scale to rate 15 remembered or imagined sensations over a wide range of intensities (e.g., the sweetness of cotton candy, the weight of a feather in your hand, the pain of biting your tongue), followed by rating the intensity of a small sample of actual taste solutions (10 taste stimuli and 5 binary mixtures).

Experimental design and procedure

The frequency of occurrence and intensity of TTs in water were measured in 2 testing sessions on separate days. At the beginning of each session participants were presented with 3 heating trials from 20 to 35 °C in which they were asked to rate the maximum intensity of any sweet, salty, sour, bitter, umami, or “other tastes” they perceived during the 15-s heating ramp. These 3 Pre-Test trials served to identify individuals who reported 1 or more tastes as greater than “barely detectable” on at least 2 of the 3 trials and whose log₁₀-mean intensity ratings across the 3 trials was greater than “barely detectable” for the purpose of assessing the intensity of swTT when it was perceived. Participants were unaware of the phenomenon of TT, were blinded to the fact that no chemical stimuli would be delivered, and were told that on a given trial they may perceive no taste, 1 taste, or multiple tastes. When the temperature reached the target temperature of 35 °C, the experimenter cued participants to lift their tongue from the solution and keep it extended from the mouth between pinched lips until they completed their intensity ratings. This practice avoided any efforts to taste the solutions inside the mouth. Participants were then instructed to rinse with 37 °C dH₂O at least 3 times before the next trial.

The Pre-Test trials were followed by a block of Test trials containing 4 replicates each of the 4 static temperatures (15, 20, 30, and 35 °C) and the dynamic heating condition (20–35 °C). Stimulus duration was 15 s for all conditions. Participants were randomly assigned to 1 of 6 pseudorandom orders to counterbalance the sequence of exposures to the temperature conditions.

Data analysis

Because intensity ratings on the gLMS tend to be log-normally distributed across individuals (Green et al. 1993, 1996), the data in this and subsequent experiments were normalized by transformation to log₁₀ prior to statistical analysis. The primary analyses of taste intensity in this and subsequent experiments were repeated-measures and mixed-effect analysis of variances (ANOVAs) conducted on the data for separate taste qualities. Tukey Honestly Significant Difference (HSD) tests were used to investigate sources of significant effects and interactions.

Results

Figure 1 displays the frequency with which the participants reported sweet, bitter, sour, salty, umami, and “other tastes” for the 5 temperature conditions tested. The results confirmed that swTT can be perceived at the tongue tip in flowing water, particularly during heating from 20 to 35 °C. Sweetness was evoked much less frequently at static temperatures of 30 and 35 °C, and was almost never at temperatures of 15 and 20 °C.

Figure 1.

The percentage of trials in which sweet, bitter, sour, salty, umami, and “other” tastes were reported for the 5 temperature conditions and 2 testing sessions of experiment 1. Note that “20–35 °C Pre” dynamic heating data were collected over the 3 trials of the Pre-Test block, whereas the “20–35 °C” dynamic heating data were collected over 4 trials that were intermixed in the test block with 4 presentations each of the other 4 static temperature conditions.

The results for the 3 tastes (sweet, bitter, and umami) that are sensed by GPCRs differed greatly. Bitterness was the second most often reported taste during heating, but far less often than sweetness, and reports of umami were very rare, never exceeding 6% of trials at any temperature. For all tastes the frequencies of report were generally consistent between Pre-Test and Test trials and between sessions. The extremely low incidence of “other” tastes indicates that under conditions of aqueous stimulation, temperature rarely evoked sensations that are not solely of gustatory in origin.

An ANOVA conducted on the intensity ratings for sweetness showed there was a significant main effect of Temperature Condition (F_4,108 = 38.9, P < 0.00001) and an interaction between the effects of Temperature Condition and Session (F_4,108 = 4.99, P < 0.001) that was attributable primarily to consistently higher sweetness ratings during heating from 20 to 35 °C (Tukey HSD, P < 0.0005). The latter interaction is consistent with more frequent reports of sweetness during dynamic heating in session 2 together with fewer reports of sweetness during exposure to static temperatures of 30 and 35 °C. There were no significant effects involving Replicate.

An ANOVA on the data for bitter taste intensity also confirmed there was a main effect of Temperature Condition (F_4,108 = 3.40, P < 0.05), but there was no interaction with Session. There were again no effects involving Replicate. Analysis of the intensity data for salty and sour found only a Temperature Condition × Session interaction for saltiness (F_4,108 = 3.16, P < 0.05), which appeared to result from a reversal between sessions in the log₁₀-mean intensity of saltiness in the dynamic heating and 15 °C static cooling conditions. Slight tendencies for more frequent responses at colder temperatures for saltiness did not result in a significant effect of Temperature Condition. Analyses of the data for umami and “other tastes” found no significant effects, which was not surprising given the very low frequencies of reports in these response categories (no more than 7% and 5% of trials, respectively). The frequent initial reports of swTT allowed us to investigate the consistency of reports within and between testing sessions. Table 1 displays the number of individuals who reported swTT in the Pre-Test and Test blocks within (top) and between (bottom) testing sessions. Within sessions, most individuals who met the response criterion during the Pre-Test block also did so in the Test block, with 72.7% (8/11) “Repeats” in session 1 and 87.5% (14/16) in session 2. The same level of consistency was found between sessions, with 72.7% (8/11) “Repeats” across Pre-Test blocks and 90.0% (9/10) across Test blocks. The data also show that the appearance of “New” swTT perceivers contributed more to variability across sessions than did “Dropouts”: 8 and 5 individuals qualified as “New” perceivers in the Pre-Test and Test blocks of session 2, respectively. Within sessions only 2 and 0 individuals qualified as “New” perceivers in the Test blocks of sessions 1 and 2, respectively. In contrast, “Dropouts” were less frequent both within and between sessions, ranging from just 10% (1/10) to only 27.2% (3/11) of initial swTT perceivers.

Table 1.

Variability in perceivers of swTT between test blocks (top) and test sessions (bottom)

	Pre-Test	Test	Repeats	Dropouts	New
Session 1	11 →	10	8	3	2
Session 2	16 →	14	14	2	0

	Session 1	Session 2	Repeats	Dropouts	New
Pre-Test	11 →	16	8	3	8
Test	10 →	14	9	1	5

To evaluate the perceived intensity of swTT we calculated log₁₀-mean ratings of sweetness and all other tastes during dynamic heating in the 4 Test block trials for the 9 individuals who met the swTT response criteria in both sessions (“Repeats” in Table 1). Figure 2 compares these ratings (Perceivers) with ratings from the 19 participants who did not meet the criteria (Non-Perceivers). The data indicate that during heating from 20 to 35 °C, swTT remained at approximately the same intensity across trials, while ratings of all other tastes were well below “barely detectable” for both Perceivers and Non-Perceivers. A mixed effects ANOVA with swTT Status (Perceivers vs. Non-Perceivers) as the grouping factor confirmed the expected main effects of swTT Status (F_1,26 = 9.75, P < 0.005), Taste (F_5,130 = 15.37, P < 0.00001), and the interaction between Status and Taste (F_5,130 = 5.34, P < 0.0005).

Figure 2.

Log₁₀-mean ratings of taste intensity for Perceivers and Non-Perceivers of swTT are shown for all 5 taste response categories during the 4 dynamic heating (20–35 °C) test trials (numbered 1–4 on the x axes) of experiment 1. Letters on the right y axis denote category labels on the gLMS: BD = barely detectable; W = weak; M = moderate; S = strong. Error bars indicate standard errors of the means.

ANOVAs were not conducted on the data for bitter, umami, salty, sour, or “other” tastes because reliable statistical analyses were not possible with so few participants meeting the response criteria at 1 or more temperatures in both Test blocks (i.e., bitter = 3, salty = 2, sour = 2, umami = 1, other = 0).

We concluded from these data that swTT could be generated in water at a frequency, intensity, and stability across trials that was sufficient to determine whether pharmacological blockade of TAS1R2/TAS1R3 by lactisole prevents perception of swTT.

Experiment 2: the role of TAS1R2/TAS1R3 in swTT

Methods

Participants

Forty participants (24 females) between 18 and 45 years of age were paid to participate in this experiment. Recruitment, eligibility criteria, and instructed restraints on consumption prior to testing were the same as in experiment 1. The research was conducted in accordance with the principles expressed in the Declaration of Helsinki, and the research protocol was approved by the Human Investigations Committee of the Yale University IRB.

Stimuli

Temperature stimulation was limited to dynamic heating from 20 to 37 °C @ 1 °C/s. Two chemical stimuli were also used: an 8.0 mM solution of the sweet taste receptor blocking agent lactisole (Jiang et al. 2005) [sodium 2-(4-methoxyphenoxy) propanoic acid; Chem-Impex International Inc.], which was prepared in dH₂O and buffered to pH 7.0 with NaOH; 0.5 M sucrose (Sigma-Aldrich), and a mixture of 0.5 M sucrose and with 8.0 mM lactisole, which was used to verify blockade of the sweet taste receptor. All thermal and chemical stimuli were delivered to the tongue tip using the TFG described in experiment 1.

Experimental design and procedure

This experiment comprised 2 sessions on separate days. The first session served to identify perceivers of swTT for participation in a second session in which the effect of lactisole on swTT was measured. The screening session included 2 dynamic heating conditions that were expected to produce reports of swTT based on prior results: warming dH₂O from 20 to 37 °C and from 30 to 37 °C, both @ 1 °C/s. The stimuli were again delivered using the TFG and each was presented 5 times in pseudorandom order. To provide a stringent test of the ability of lactisole to block swTT, only the top quartile of participants (n = 10; 6 females) who reported swTT in the screening session were invited to participate in the lactisole session. These individuals all reported more than “barely detectable” sweetness during at least 3 of the 5 dynamic heating trials, and as a group had a mean swTT taste rating above “weak.”

Testing in the lactisole session began with both of the dynamic heating stimuli used in the screening session followed by a repeat of the heating trials with 8.0 mM lactisole. Because water rinses after lactisole exposure can produce a sweet “water taste” (Galindo-Cuspinera et al. 2006; Alvarado et al. 2017), participants rinsed the whole mouth multiple times between trials to eliminate residual sweetness. To confirm blockade of the sweet taste receptor, this sequence of testing was repeated with the 0.5 M sucrose stimulus presented alone and in mixture with 8.0 mM lactisole.

Results

The data in Figure 3 show that lactisole effectively eliminated swTT perceived by the most sensitive individuals (i.e., the top quartile of participants, n = 10) identified in the first session. Expressed as a percentage of initial sweetness intensity, lactisole reduced swTT by approximately 95%. As expected, lactisole also blocked the sweetness of 0.5 M sucrose in the same individuals. Separate ANOVAs were conducted on the sweetness data for the dynamic heating and sucrose conditions, with Temperature Ramp (20–37 and 30–37 °C) and Treatment (lactisole vs. no-lactisole) as factors. The analyses confirmed there were significant main effects of Treatment on the intensity of swTT (F_1,9 = 116.52, P < 0.00001) and on sucrose sweetness (F_1,9 = 153.04, P < 0.00001). An effect of Temperature Ramp on swTT fell just short of statistical significance (F_1,9 = 4.75, P = 0.057). This trend is consistent with prior evidence that swTT was perceived more strongly when the starting temperature was 20 rather than 30 °C (Cruz and Green 2000).

Figure 3.

The blocking effect of lactisole on swTT and sucrose sweetness at the tongue tip. In both conditions the aqueous stimulus (dH₂O or 0.5 M sucrose) was heated from 20 to 37 °C and from 30 to 37 °C on separate trials, with and without 8.0 mM lactisole added. Letters on the right y axis denote semantic categories of the gLMS: BD = barely detectable; W = weak; M = moderate; S = strong. Error bars indicate SEMs.

Figure 4 contains the log-mean intensity ratings for all 4 taste qualities for the 20–37 °C dynamic heating condition with or without lactisole present. Heating did not produce clearly perceptible tastes other than sweetness. These data also rule out the possibility that swTT was blocked by a weak taste of lactisole or the NaOH buffer.

Figure 4.

Log₁₀-mean intensity ratings of sweetness (Sw), saltiness (Sa), sourness (So), and bitterness (Bi) at the tongue tip during heating of dH₂O from 20 to 37 °C, with and without 8.0 mM lactisole. Letters on the right y axis denote semantic categories of the gLMS: BD = barely detectable; W = weak; M = moderate; S = strong. Error bars indicate SEMs.

Overall, the evidence that swTT is dependent on TAS1R2/TAS1R3 enabled us to proceed with the study of possible sensory or cognitive interactions between chemical and thermal stimulation.

Experiment 3: interactions between thermal and chemical sweetness

Methods

Participants

Thirty-six adults (20 females) between the ages of 18 and 45 served in the experiment. Recruitment, eligibility criteria, and instructed restraints on consumption prior to testing were the same as in experiment 1. All gave informed consent and were paid for their participation. The research was conducted in accordance with the principles expressed in the Declaration of Helsinki, and the research protocol was approved by the Human Investigations Committee of the Yale University IRB.

Stimuli

The TFG was used again in this experiment to deliver all stimuli to the tongue tip. The TT stimulus was heating from 20 to 35 °C @ 1 °C/s, and taste solutions of 56 and 180 mM sucrose and 0.018 and 0.056 mM quinine (Sigma-Aldrich) prepared in dH₂O. To provide a more sensitive measure of possible enhancement of sucrose sweetness in the 20–35 °C dynamic heating condition, both sucrose concentrations were substantially below the 0.5 M solution of experiment 2. The sucrose and quinine stimuli were also presented at static temperatures of 20 and 35 °C for 3 and 15 s.

Experimental design and procedure

The possibility that interactions can occur between swTT and sweet or bitter taste was assessed in 2 sessions on separate days. Of interest was whether 1) intermittent exposure to a prototypical sweet taste stimulus would interfere with perception of swTT, and 2) whether dynamic heating from 20 to 35 °C that is capable of producing swTT can enhance the sweetness produced by an agonist of TAS1R2/TAS1R3. Two concentrations of sucrose were included to investigate whether enhancement of sweet taste by dynamic heating might be detectable only with weaker stimuli. Two concentrations of quinine that in preliminary testing produced bitterness similar in intensity to the sweetness of the 2 sucrose concentrations were also included as control stimuli.

As in experiment 1, testing sessions began with 3 Pre-Test exposures to the dynamic heating stimulus in dH₂O alone, followed by a Test block containing 4 presentations of the dynamic heating stimulus in dH₂O and the 2 concentrations of sucrose or quinine in the specified temperature conditions, including dynamic heating. Subjects were randomly assigned to 1 of 6 pseudorandom stimulus presentation orders in the Test block, and there was a 1-min intertrial interval during which subjects rinsed at least 3 times with 37 °C dH₂O to avoid carryover effects of the chemical stimuli to the water-only trials.

Results

As in experiment 1, measurements of swTT were derived from individuals who reported at least “barely detectable” sweetness on at least 2 of the 3 trials in the Pre-Test block and whose log₁₀-mean sweetness intensity rating across the 3 trials was greater than “barely detectable.” Figure 5 displays the taste intensity data for individuals who met these criteria in the sucrose intermixed (top graph; n = 17) and quinine intermixed (bottom graph; n = 17) sessions.

Figure 5.

Perception of swTT in experiment 3 for the Pre-Test block of trials that contained only dH₂O vs. the Test block that contained trials with sucrose (top) or quinine (bottom) intermixed with dH₂O trials. Asterisk indicates a significant reduction in swTT when swTT trials were intermixed with sucrose trials. Letters on the x axis indicate ratings of sweetness (Sw), saltiness (Sa), sourness (So), bitterness (Bi), and umami (Um). Letters on the right y axis denote semantic categories of the gLMS: BD = barely detectable; W = weak; M = moderate; S = strong. Error bars indicate SEMs.

Perception of swTT during the Test block was markedly reduced in the context of sucrose but not quinine. An ANOVA conducted on the data from the 3 Pre-Test trials and first 3 Test trials with Replicate and Taste Quality as factors confirmed the reduction of swTT in the context of sucrose was significant (main effect of Condition; F_1,16 = 5.37, P < 0.05). There was also a main effect of Taste Quality (F_4,64 = 25.12, P < 0.00001) and a significant Condition × Taste Quality interaction (F_4,64 = 6.83, P < 0.00001), which post hoc tests confirmed was due to the significant reduction in sweetness (only) in the Test block (Tukey HSD, P < 0.05). The consistently low ratings of other tastes show that the sucrose context selectively affected swTT without increasing reports of other tastes. A separate analysis of the quinine data showed that while there was the expected main effect of Taste Quality (F_4,64 = 51.12, P < 0.00001), the Taste Quality × Condition interaction was not significant.

To evaluate the perception of swTT over individual trials, Figure 6 displays the sweetness data for all Pre-Test and Test trials for the sucrose and quinine sessions. These data show that interference of swTT in the context of sucrose was significant on the first dynamic heating Test trial and persisted over the remaining 3 trials. An ANOVA comparing sweetness ratings on Pre-Test trials to sweetness ratings on Test trials 2–4 found a main effect of Condition (F_1,16 = 7.83, P < 0.02) but no interaction with Trial, thus confirming the persistence of suppression across trials. Post hoc tests also showed there were no significant differences in swTT intensity among the Test trials. The same analysis performed on the quinine data found no effect of Condition.

Figure 6.

The data of Figure 5 broken down by the 3 trials in the Pre-Test block (open bars) and the 4 trials in the Test block (hatched bars) in the sucrose (top) and quinine (bottom) sessions. Asterisks indicate significant reductions in sweetness compared with the third Pre-Test trial (Tukey HSD, P < 0.05).

While there was a slight trend toward reduced swTT across Test trials, there was not a significant Condition × Trial interaction. In terms of the magnitude of change in swTT in the context of sucrose or quinine, sucrose reduced swTT over the 4 Test trials by an average of 78.1% compared with 31.5% for quinine.

The significant interference of swTT in the context of sucrose but not quinine could potentially arise from adaptation of sweet taste during repeated exposures to sucrose. However, this possibility is inconsistent with the occurrence of significant interference on the first swTT test trial, and the absence of significant reductions in swTT following additional exposures to sucrose in later trials. Possible adaptation of sucrose sweetness was also investigated in a mixed-effects ANOVA with Replicate, Concentration (56 and 180 mM), and Duration (3 and 15 s) as within-subject factors, and swTT Status (i.e., Perceivers vs. Non-Perceivers) as the grouping factor. The analysis found only the expected main effect of Concentration (F_1,34 = 96.8, P < 0.000001) and a Replicate × Concentration × Duration interaction (F_1,34 = 4.46, P < 0.05). No consistent trend was found toward adaptation across replicates or stimulus durations, and post hoc tests found no significant differences in the sweetness of the 56- or 180-mM sucrose stimuli across replicates for either the 3- or 15-s stimulus durations (Tukey HSD, all Ps > 0.05). The absence of a significant difference for the 56 mM stimulus is particularly relevant because its sweetness was similar in intensity to swTT.

To investigate whether thermal conditions capable of evoking swTT might also enhance the sweetness of sucrose, we compared sucrose sweetness during dynamic heating from 20 to 35 °C with sweetness reported in the static heating and cooling conditions. Of particular interest was the comparison with exposure to 35 °C for 15 s, which matched both the duration and target temperature of the dynamic heating stimulus. The data from participants who met the swTT criterion in the Pre-Test block of the sucrose session (n = 17) are shown in Figure 7. Analysis of the sucrose data showed that while there was a significant main effect of Temperature Condition (F_4,64 = 16.17, P < 0.000001) and a Temperature Condition × Concentration interaction (F_4,64 = 8.73, P < 0.00001), both effects were due to strong suppression of the sweetness of 56 mM sucrose at 20 °C (Figure 7a). Tukey HSD tests showed there were no significant differences in sweetness between dynamic heating and exposure to 35 °C for 3- or 15-s for either sucrose concentration.

Figure 7.

Shown are log₁₀-mean ratings of perceived sweetness (a) and bitterness (b) for the 5 temperature conditions of experiment 3 and the 2 concentrations of sucrose and quinine that were tested. Bar values with different letters are significantly different (Tukey HSD, P < 0.05). Letters on the right y axis denote semantic categories of the gLMS: BD = barely detectable; W = weak; M = moderate; S = strong. Error bars indicate SEMs.

Analysis of the quinine data found different effects of temperature, particularly for the lower concentration (Figure 7b). A main effect of Temperature Condition (F_4,64 = 4.64, P < 0.005) and a Temperature Condition × Concentration (F_4,64 = 5.62, P < 0.001) interaction were found for quinine, but the differences in bitterness intensity across conditions at the lower concentration were opposite to those for sucrose. Compared with the dynamic heating condition, the bitterness of 0.018 mM quinine was not reduced by exposure to 20 °C, but was reduced by exposures to 35 °C (Tukey HSD tests; P < 0.05). Thus for the lower quinine concentration, dynamic heating to 35 °C appeared to prevent the reduction in bitterness that occurred during 3- and 15-s static exposures to 35 °C. In contrast, no significant differences in bitterness among temperature conditions were found for the higher quinine concentration.

Discussion

Dependence of swTT on TAS1R2/TAS1R3

While neurophysiological studies in nonhuman animals (e.g., Ogawa et al. 1968, 1972; Sato et al. 1975; Nakamura and Kurihara 1988; Breza et al. 2006; Wilson and Lemon 2014; Lemon 2017) and in humans (Oakley 1985) have provided clear evidence that heating can modulate chemical stimulation in the sweet taste pathway, experiment 2 provided the first direct evidence of the involvement of TAS1R2/TAS1R3. Because lactisole has been proposed to block sweet taste by mechanically locking the TAS1R2/TAS1R3 dimer in an inactive state, causing intracellular calcium to fall below baseline levels (Jiang et al. 2005), the present finding is consistent with the hypothesis of Talavera et al. (2005) that TRPM5 acts as a “coincidence detector” of heat and intracellular calcium. However, the integral relationship between heat and calcium in sweet taste transduction also implies that the heat sensitivity of TRPM5 alone cannot explain the differences in TT among the tastes mediated by GPCRs. Similarly, differences in the effects of heating on sweet (Green and Nachtigal 2015) and bitter (Green and Andrew 2017) stimuli, and the effects of exposure to 15 and 20 °C on sucrose and quinine in the present study (Figure 7), are also difficult to explain solely in terms of the heat sensitivity of TRPM5. Instead these effects suggest there may be an additional heat-sensitive mechanism upstream of TRPM5 that is capable of modulating intracellular calcium.

One possibility is that heating from a cool temperature may change the conformation of TAS1R2/TAS1R3 (Green and Nachtigal 2012; Wilson and Lemon 2013; Green and Nachtigal 2015). This hypothesis receives indirect support from evidence that mechanical changes take place in the transmembrane proteins of other receptors at temperatures in the range in which swTT occurs (Hoffstaetter et al. 2018), and it has been proposed that conformational change is a common mechanism of channel gating in heat-sensitive ion channels (Liu and Montell 2015). The hypothesis is also supported by evidence that the N-terminal domain of GPCRs can regulate basal signaling activity in the absence of ligands (Meye et al. 2014; Müller et al. 2016; Coleman et al. 2017), which implies that any thermally evoked change in receptor conformation would have the potential to modulate intracellular calcium. Effects of dynamic heating on receptor conformation might also explain the differences in occurrence of sweet, bitter, and umami TTs, since structural differences in the N-terminal domains of the respective GPCRs (Kinnamon 2012; Liman et al. 2014) could affect their thermal sensitivity.

Although the thermal sensitivity of TRPM5 and other receptor proteins makes it reasonable to focus on TAS1R2/TAS1R3 as the locus of swTT, it has been proposed that sweet taste in humans may also be mediated by a pathway involving glucose transporters (Yee et al. 2011; Sukumaran et al. 2016). Nothing is yet known about the temperature sensitivity of this proposed taste pathway, however, and we could find no published evidence that lactisole can block it. In addition, further study of taste signaling within and between TRCs (Roper 2007; Chaudhari and Roper 2010) will be necessary to determine if other temperature-sensitive mechanisms exist downstream of the TAS1R2/TAS1R3 transduction cascade that may also contribute to the thermal sensitivity of taste fibers, and thus swTT.

Interactions with chemical taste

The dependence of swTT on TAS1R2/TAS1R3 suggested that perceptual interactions might occur between chemical and thermal stimulation of sweet taste. While it may not be surprising that enhancement of sweetness did not occur for 0.5 M sucrose in experiment 2 or 180 mM sucrose in experiment 3, the lack of enhancement for 56 mM sucrose during heating from 20 to 35 °C is more telling. This result indicates that thermal activation of the transduction cascade during heating is not additive with chemical activation, which is consistent with the receptor conformation hypothesis of swTT: state changes in conformation of the N-terminal domain during agonist binding would be likely to affect the response of the protein to heating.

In addition, the greater suppression of sweetness for 56 mM sucrose compared with 180 mM sucrose during static cooling at 20 °C (Figure 7a) is unlikely to be related to the mechanism of swTT, which is strongly dependent on warming the tongue from a cold temperature over several seconds (Figure 1). Instead, the stronger effect of cold on the sweetness of 56 mM sucrose is consistent with prior evidence that cooling affects perceived sweetness less at higher concentrations (Bartoshuk et al. 1982; Frankmann and Green 1987), and evidence from a study in mice that temperature has a smaller effect on central taste neurons at high sucrose concentrations (Wilson and Lemon 2014).

Another aim of experiment 3 was to investigate whether prior exposure to a prototypical sweet stimulus interrupts swTT, a possibility that was noticed during preliminary testing for the experiment. The significant reduction in swTT in blocks of trials that contained sucrose but not quinine confirmed the existence of a context effect. A possible explanation for this difference is that exposure to the sucrose stimuli across trials resulted in sweet taste adaptation. However, the absence of significant reductions in the sweetness of 56 mM sucrose, which produced a sweetness similar in intensity to swTT, is inconsistent with this explanation, as is the occurrence of a significant drop in sweetness on the first swTT trial. However, the results do not rule out the possibility that exposure to sucrose may produce a more rapid adaptation of the effect of heating on TAS1R2/TAS1R3. Finally, a perceptual contrast effect could potentially be induced by the stronger sweetness of the 180 mM stimulus (Rankin and Marks 1991), but such an effect would also be expected to reduce the perceived sweetness of 56 mM sucrose.

Alternatively, the selective suppression of swTT by sucrose may have been caused by changes in the way heat activation of the sweet taste pathway was encoded following exposure to a prototypical sweet taste stimulus. As noted earlier, electrophysiological studies have shown across-fiber correlations in the response of chorda tympani fibers to heating and sucrose, but the associations were imperfect: heating produced a lower-fidelity version of chorda tympani stimulation compared with sucrose (Ogawa et al. 1968; Sato et al. 1975). In addition, parallel activation of heat-sensitive thermoreceptors of the trigeminal nerve (Poulos and Lende 1970; Lemon et al. 2016) produces a spatially and temporally correlated pattern of bimodal stimulation that could make discrimination of the sweet taste signal more difficult. Indeed, von Bekesy (1964) described a qualitative similarity between sensations of warmth and sweetness that led us in the present and prior studies (Cruz and Green 2000; Green and George 2004; Green et al. 2005) to avoid asking participants to rate taste and temperature sensations on the same trials. Under such challenging perceptual conditions, exposure to a prototypical sweet stimulus may decrease the likelihood that a subsequent bimodal pattern of stimulation will be decoded as both sweetness and warmth. This interpretation is compatible with models of hierarchical perceptual processing (e.g., Bayesian inference: Lee and Mumford 2003) that have been applied in studies of perceptual illusions in other senses (Geisler and Kersten 2002; Weiss et al. 2002; Goldreich 2007). In this view, perceptual processing relies on inferences about a stimulus or object based on both the sensory signal and past experience. Relevant to interference of swTT by sucrose may be the concept of “opportunistic learning,” which proposes that the perceptual system “learns” a particular stimulus or object when it is presented “under conditions of low ambiguity” (Kersten et al. 2004; p. 285). In contrast to the low-fidelity and ambiguous pattern of sweet taste stimulation produced by heating, presentation of sucrose interspersed with TT trials may provide such an opportunity. This possibility could be investigated by measuring the effect on swTT of pre-exposure to sweeteners alone vs. sweeteners in mixture with other tastes that increase the perceptual ambiguity of the sweet stimulus.

Variability in swTT between individuals and over time

Variability is an inherent characteristic of many perceptual illusions, particularly those like TT that are produced by unusual and ambiguous stimuli that create perceptual conflicts (e.g., Aafjes et al. 1966; Hamalainen et al. 1982; Green 2002; Nath and Beauchamp 2012). The present study provided quantitative evidence of the variability in swTT under the relatively natural conditions of tasting an aqueous stimulus with the tongue tip. But consistent with prior studies with Peltier thermodes (Cruz and Green 2000; Bajec and Pickering 2008b; Yang et al. 2014; Skinner et al. 2018), half of participants or fewer were identified as perceivers of swTT. The data of Table 1 further indicate that perception of swTT was not always consistent for these individuals across blocks of trials, as was recently shown (Skinner et al. 2018). Inconsistency in reports of tastes across pairs of replicate trials has led some investigators to use 3 categories of “taster status”: individuals who reported the same taste as at least “weak” on both replicates were considered “tasters,” those who reported no taste on either trial were considered “non-tasters,” and those who reported a “weak” taste on just one were categorized as neither “tasters” nor “non-tasters” (e.g., Bajec and Pickering 2008b; Thibodeau et al. 2018). While the variability of swTT makes it necessary to develop criteria for identifying perceivers, it also means that even sensitive individuals may not perceive the illusion on 2 consecutive trials. Accordingly, in experiment 1 we included 3 Pre-Test trials and identified swTT perceivers as those who reported sweetness above “barely detectable” on at least 2 of the 3 trials, and in experiment 2 perceivers were selected based on reporting swTT on at least 3 of 5 trials. These criteria yielded highly significant group differences in ratings of sweet taste intensity. On the other hand, the percentage of individuals classified as perceivers was no higher than in prior studies that utilized contact thermodes and set a more stringent intensity criterion of rating swTT above “weak” (Cruz and Green 2000; Green and George 2004; Green et al. 2005; Bajec and Pickering 2008b). This suggests that the naturalness of gustatory stimulation is not a major factor in perception of swTT, and/or that the lower thermal conductivity of water compared with the metal surface of a thermode produced less robust activation of sweet taste receptors.

The only hypothesis so far proposed to explain individual differences in swTT is that perceivers have a higher “central gain” for taste (and perhaps olfactory) stimulation, resulting in stronger sensations of taste for equivalent levels of receptor activation (Green and George 2004). However, support for this hypothesis has been uneven. Though a subsequent study from the same laboratory replicated the results of the first study by finding greater responsiveness to tastes but not to somatosensory/chemesthetic stimuli (Green et al. 2005), 2 later studies found higher responsiveness to oral somatosensory stimuli as well as taste stimuli (Bajec and Pickering 2008b; Yang et al. 2014), and the results of experiment 3 lend only partial support to the hypothesis. Analysis of the intensity ratings for 56 mM sucrose showed that perceivers of swTT rated it twice as sweet as non-perceivers (2-tailed t-test, P < 0.05), but no difference was found for the stronger 180 mM sucrose stimulus or for either concentration of quinine.

On the other hand, the concept of a stable swTT phenotype is supported by the data in Table 1 that indicate the repeatability of swTT for individuals identified as perceivers was relatively high. Averaged over both sessions of experiment 1, 80.1% of perceivers in the Pre-Test block also met the criteria in the Test block. In addition, the number of “Dropouts” between sessions was no higher than between Pre-Test and Test blocks within a session, and many more “New” perceivers appeared between sessions than within sessions, resulting in a net increase in perceivers over time. Although the sample is small, this outcome raises the possibility that perceptual learning occurs during repeated exposures to the dynamic heating stimulus. In any case, steady or increasing numbers of perceivers across sessions could be explained if stable sources of individual differences, such as sweet receptor expression or central gain, interact with factors that vary over briefer timeframes, such as trial-by-trial differences in activation of TAS1R2/TAS1R3 receptors and fluctuations in selective attention. Unfortunately a scarcity of data on trial-to-trial or day-to-day variation in perception of chemical taste (Mattes 1988; Stevens et al. 1995) makes it difficult to know how much of the variability observed here is specific to swTT.

Differences in sensations reported as TTs

Another source of variability in the study of TT is the disparity in sensations that are reported as TTs. The data of experiment 1 differ from some prior studies in the paucity of reports of “other tastes” and the absence of reports of swTT at cool and cold temperatures. With respect to disparate sensations, metallic (Bajec and Pickering 2008b; Yang et al. 2014; Hort et al. 2016; Skinner et al. 2018), astringent (Bajec and Pickering 2008b; Skinner et al. 2018) and minty and spicy (Hort et al. 2016; Skinner et al. 2018) have all been reported at frequencies that rival or exceed those for sweetness and other wholly gustatory sensations. However, the somatosensory and/or olfactory components of metallic, astringent, minty, and spicy sensations makes their designation as TTs questionable (Skinner et al. 2018). Mintiness derives primarily from the odors of mint oils and menthol stimulation of cold receptors and nociceptors (Peier et al. 2002; Namer et al. 2005), and spiciness derives chiefly from stimulation of heat-sensitive receptors and nociceptors by chemesthetic agents in peppers and other spices (Green 1996; Bandell et al. 2007; Roper 2014). Astringent agents often have bitter and/or sour tastes, but their characterizing perceptual qualities are tactile (Bate-Smith 1954; Breslin et al. 1993; Green 1993; Lawless et al. 1994; Bajec and Pickering 2008a; Lee et al. 2012; Ployon et al. 2018). Similarly, some trivalent and divalent salts produce a metallic quality, but the majority of evidence points to olfaction as its primary source (Hettinger et al. 1990; Lim and Lawless 2005).

This evidence makes it likely that some sensations that have been reported as TT are methodological artifacts. For example, prolonged exposure of the tongue to ambient air is a potential source of astringent and other less well-defined sensations. In studies where such sensations were most frequently reported, participants were required to extend the tongue for at least 40 s during cooling trials and 45–60 s during heating trials (Hort et al. 2016; Skinner et al. 2018). Exposure to ambient air cools the tongue’s surface as saliva evaporates, and casual observation demonstrates that as exposure continues, coolness, dryness, and other nondescript sensations can begin to appear which, under the demand characteristics of an experiment, might be reported as astringent or metallic. This is particularly likely with respect to astringency, for which “drying” is a salient perceptual quality (Lawless et al. 1994; Thomas and Lawless 1995; Lawless et al. 1996). In studies that use metal thermodes the mere sight of the thermode might increase reports of metallic taste, particularly when metallic is offered as a taste response (Yang et al. 2014).

Designating sensations that are reported equally often during warming and cooling as TTs is also subject to question. Sensations perceived independently of temperature, as has been reported for sweet and metallic sensations (Yang et al. 2014; Skinner et al. 2018), cannot be explained in terms of the temperature sensitivity of thermoreceptors (Patapoutian et al. 2003; Venkatachalam and Montell 2007) or peripheral (Ogawa et al. 1968; Sato et al. 1975; Breza et al. 2006) or central gustatory neurons (Lemon 2017), none of which respond equally to heating and cooling. In the case of swTT, the data of the present study show not only that reports of sweetness were extremely infrequent at cool and cold temperatures (Figure 1), but also that at 20 °C, the addition of a low concentration of sucrose failed to evoke even “barely detectable” sweetness (Figure 7a).

However, the lack of clear relationships to temperature in the bitter, sour, and salty data of experiment 1 raises the possibility that the present data also include reports of tastes that are unrelated to TT. For example, in experiment 1 the more frequent reports of bitterness, sourness, and saltiness compared with umami and “other tastes” may reflect a bias toward reporting familiar tastes (O’Mahony et al. 1979). This bias is particularly likely in experiments like the present one in which perceptible sensations might not be experienced on every trial, and when participants must base their responses on a list of possible sensations. In the present study, simply dipping the tongue tip in flowing water might have been a source of tastes that are unrelated to thermal stimulation. The similar frequency of reports of bitter and sour tastes across temperatures may have been due to rinsing saliva from the tongue, which can evoke “water tastes,” the most common being bitter and sour (Bartoshuk et al. 1964; Bartoshuk 1968). Some variability in reports of TT may also arise from quality confusions that also occur between chemically stimulated tastes (Meiselman and Dzendolet 1967; McAuliffe and Meiselman 1974; Hettinger et al. 1999; Crouzet et al. 2015), particularly when stimulation is weak (Meiselman and Dzendolet 1967). These multiple factors clearly mean that not all sensations reported during thermal stimulation should be accepted uncritically as TTs, particularly if they have no systematic relationship to temperature, are reported only infrequently (e.g., <10% of trials), or are characterized by sensations that are predominantly somatosensory or olfactory in origin.

Conclusions

Despite complications posed by the variability of TT and differences in psychophysical methods and procedures across studies, evidence from research in other senses (Eagleman 2001; Jazayeri and Movshon 2007) has shown that studying sensory illusions can lead to better understanding of the mechanisms of veridical perception as well as the illusions themselves. The present study contributes to both of these goals. First, studying swTT under a more natural condition of tasting adds to our understanding of the thermal and perceptual characteristics of the illusion. Second, the evidence that swTT requires activation of TAS1R2/TAS1R3 suggests the thermal sensitivity of sweet taste may depend on an interaction between the temperature sensitivity of TRPM5 and a second, upstream mechanism, which we propose may involve temperature-dependent changes in receptor conformation. Such an interaction could potentially explain the striking differences in TT among tastes sensed via GPCRs.

Funding

This research was supported by a grant from the National Institutes of Health [RO1DC005002].

Conflicts of interest

The authors report no conflicts of interest.