*Article* **Intraintestinal Delivery of Tastants Using a Naso-Duodenal-Ileal Catheter Does Not Influence Food Intake or Satiety**

**Tim Klaassen 1,2,\*, Annick M. E. Alleleyn 1, Mark van Avesaat 1, Freddy J. Troost 1,2, Daniel Keszthelyi <sup>1</sup> and Adrian A. M. Masclee <sup>1</sup>**


Received: 18 December 2018; Accepted: 20 February 2019; Published: 23 February 2019

**Abstract:** Intraduodenal activity of taste receptors reduces food intake. Taste receptors are expressed throughout the entire gastrointestinal tract. Currently, there are no data available on the effects of distal taste receptor activation. In this study, we investigate the effect of intraduodenal and/or intraileal activation of taste receptors on food intake and satiety. In a single-blind randomized crossover trial, fourteen participants were intubated with a naso-duodenal-ileal catheter and received four infusion regimens: duodenal placebo and ileal placebo (DPIP), duodenal tastants and ileal placebo (DTIP), duodenal placebo and ileal tastants (DPIT), duodenal tastants and ileal tastants (DTIT). Fifteen minutes after cessation of infusion, subjects received an *ad libitum* meal to measure food intake. Visual analog scale scores for satiety feelings were collected at regular intervals. No differences in food intake were observed between the various interventions (DPIP: 786.6 ± 79.2 Kcal, DTIP: 803.3 ± 69.0 Kcal, DPIT: 814.7 ± 77.3 Kcal, DTIT: 834.8 ± 59.2 Kcal, *p* = 0.59). No differences in satiety feelings were observed. Intestinal infusion of tastants using a naso-duodenal-ileal catheter did not influence food intake or satiety feelings. Possibly, the burden of the four-day naso-duodenal-ileal intubation masked a small effect that tastants might have on food intake and satiety.

**Keywords:** satiety; tastants; food intake; intraduodenal infusion; intraileal infusion; overweight; weight management

#### **1. Introduction**

Obesity is considered a major healthcare problem with worldwide obesity almost being tripled since 1975 [1]. Therefore, there is an increasing need for non-invasive therapies for weight management. Gastrointestinal (GI) hormones, such as cholecystokinin (CCK) and glucagon-like peptide-1 (GLP-1), have been shown to reduce food intake and hunger after intravenous administration [2–4]. Therefore, the GI-tract is an interesting target for non-invasive therapies to reduce food intake and induce satiety/satiation.

Intestinal macronutrient infusion decreases food intake and induces the release of CCK, GLP-1, and peptide YY (PYY) [5]. This mechanism is commonly referred to as intestinal- or ileal brake [6,7]. A recent review proposed a proximal to the distal gradient in the small intestine, where a more profound effect on food intake can be found after distal compared to proximal macronutrient

infusion [8]. Previous studies have demonstrated that besides macronutrients, substances referred to as tastants are able to activate certain taste receptors in the GI-tract which are coupled to enteroendocrine cells (EEC), and can trigger the release of satiety hormones (i.e., CCK, GLP-1, and PYY) [9–13]. These taste receptors can be found throughout the entire GI-tract. Expression levels for the various taste receptor differ throughout the gut. Table 1 gives a simplified visual representation of the relative expression of taste receptors throughout the human gut based on current literature [14–17].

In a recent study, van Avesaat et al. have shown that duodenal infusion of a combination of sweet, bitter, and umami tastants significantly decreased *ad libitum* meal intake, whilst increasing satiety and decreasing hunger feelings. These effects were not accompanied by changes in systemic levels of GLP-1, PYY, and CCK [18]. Up to now, no data are available on the effect of activation of taste receptors in the more distal small intestine. Since one of the functions of taste receptors in the gut is to sense food being present in the lumen, it should be investigated whether the beforementioned proximal to distal gradient found for the intestinal brake is operative for taste receptor activation.

Therefore, in the present study, we compared the effects of intraduodenal infusion versus intraileal infusion of a combination of tastants (sweet, bitter, and umami) on *ad libitum* food intake, satiation, and GI-complaints in healthy subjects. Since sweet and umami taste are sensed by various subtypes of the taste receptor family 1 (TAS1R) and bitter taste is sensed by the taste receptor family 2 (TAS2R), the combination will activate a wide range of taste receptors. We hypothesized that infusing tastants at both infusion sites (duodenum and ileum) will decrease food intake and increase satiation to the greatest extent when compared with infusion of placebo or single port infusion. Infusing in solely the duodenum or the ileum will also decrease food intake and increase satiation when compared to placebo, albeit to a lesser degree than infusing at both infusion sites simultaneously. Furthermore, we expect intraileal delivery of tastants will decrease food intake and increase satiation to a greater extent when compared with intraduodenal delivery of tastants.


**Table 1.** A simplified visual representation of the relative expression of taste receptors and gustducin throughout the human GI-tract.

Expression levels are relative to each other and a simplified visual representation with ++ indicating very high expression, + indicating high expression, +/− indicating medium expression, − indicating low expression, and −− indicating very low expression. \$ Young et al. displayed the stomach as fundus, body, and antrum. For details, please refer to Young et al. [15]. # Young et al. displayed jejunum as proximal jejunum and distal jejunum. For details, please refer to Young et al. [15]. N/A: not available. \* T2R family is expressed throughout the entire small intestine in a comparable fashion with some subtypes more abundant proximally and some distally. For details, please refer to Gu et al. [17].

#### **2. Materials and Methods**

This study was approved by the Medical Ethics Committee of the Maastricht University Medical Center+ (MUMC+), Maastricht, The Netherlands, and performed in full accordance with the Declaration of Helsinki (latest amendment by the World Medic Association in 2013) and Dutch Regulations on Medical Research Involving Human Subjects (WMO, 1998). This study was registered in the US National Library of Medicine (http://www.clinicaltrials.gov, ID NCT03140930). All subjects gave written informed consent before screening.

#### *2.1. Subjects*

Healthy men and women were recruited by local advertisements. Inclusion criteria were age between 18 and 65 years, a body mass index (BMI) between 18 and 25 kg/m2, with a stable weight over the past six months (<5% body weight change). Exclusion criteria were gastrointestinal complaints, history of chronic or severe disease, use of medication influencing endpoints within 14 days prior to testing, administration of investigational drugs which interfere with this study, major abdominal surgery, dieting, pregnancy or lactation, excessive alcohol consumption (>20 alcoholic consumptions per week), smoking, weight <60 kg, non-tasters of sweet, bitter or umami stimuli, evidence of monosodium glutamate (MSG)-hypersensitivity.

Prior to testing, screening was performed where abovementioned inclusion and exclusion criteria were checked, and a taste perception test was performed. Subjects tasted quinine (0.5 mmol/L), Reb A (50 mmol/L), MSG (50 mmol/L), and tap water blindly and had to indicate their sense of taste. Subjects had to identify each taste correctly in order to be eligible for the study. Furthermore, their length and weight were measured to calculate their BMI.

A sample size calculation was based on the difference in meal intake between duodenal infusion of a combination of tastants and duodenal infusion of placebo as reported by van Avesaat et al. [18]. Using a difference in means of 64 Kcal, a standard deviation of difference of 63, a power of 80%, and an alpha of 1.67%, a total number of 13 subjects were needed. An alpha of 1.67% was used to correct for multiple testing.

#### *2.2. Study Design*

In this single-blind randomized, placebo-controlled crossover study, subjects received the combination of tastants (sweet, bitter, and umami) and/or placebo (tap water) in the duodenum and/or the ileum for four consecutive test days. This results in four combinations which were infused on the various test days: duodenal placebo and ileal placebo (DPIP), duodenal tastants and ileal placebo (DTIP), duodenal placebo and ileal tastants (DPIT), duodenal tastants and ileal tastants (DTIT).

#### *2.3. Catheter Positioning*

A 305 cm long silicon 9-lumen (8-lumen, 1 balloon inflation channel, the outer diameter of 3.5 mm) custom-made naso-ileal reusable catheter (Dentsleeve International, Mui Scientific, Mississauga, Canada) was used for intubation.

One day prior to the first test day, subjects arrived at 7:40 AM at the Maastricht University Medical Center+ (MUMC+) after an overnight fast. If preferred by the subject, local anesthesia of nasal mucosa using xylocaine (10% spray, AstraZeneca, Zoetermeer, The Netherlands) was applied. After placement of the catheter in the stomach, the catheter was guided through the pylorus and into the duodenum under intermittent fluoroscopic control. Progression of the catheter from duodenum to ileum was performed as described earlier [19]. Fluoroscopy was used to check the positioning of the catheter on the first and the last test day. Radio-opaque markers were added to the infusion ports on the catheter, which accounted for the determination of the catheter position. On all test days, intestinal fluid was sampled from various infusion ports, and pH was measured using pH strips (MColorpHast™, Merck, Darmstadt, Germany) in order to estimate the catheter positioning.

#### *2.4. Preparation and Infusion of Tastants*

The combination of three tastants was infused in the duodenum, the ileum, or both the duodenum and the ileum. In order to prevent side effects from occurring, 75% of acceptable daily intake (ADI) of these tastants was infused. 540 mg Rebaudioside A (Reb A, Stevija Natuurlijk, Drachten, The Netherlands), 75 mg Quinine (Arnold Suhr, Hilversum, The Netherlands), and 2 g Monosodium Glutamate (MSG, Ajinomoto, Hamburg, Germany) were dissolved in 120 mL tap water and was used as tastant mixture for infusion, as was done by van Avesaat et al. [18]. All tastants used were non-caloric and yielded no nutritional value. The placebo infusion consisted of 120 mL of tap water. A magnetic stirrer was used to dissolve the tastants. The mixture was infused over a 60-min period with an infusion rate of 2 mL/min. This was consistent with the infusion rate of van Avesaat et al. mimicking the slow influx from the stomach to duodenum and slow transit through the gut in the ileum.

#### *2.5. Protocol*

On each test day, after an 8 h overnight fast, subjects arrived at 8:00 AM at the MUMC+. Subjects were instructed to consume the same habitual meal on the evening prior to testing. Hereafter, at t = 0 min, a standardized liquid breakfast meal (250 mL Goedemorgen drinkontbijt (Vifit); energy 145 Kcal per portion, 20.25 g carbohydrates, 8.5 g protein, and 2 g fat) was consumed. One hundred and fifty min (at t = 150 min) after breakfast consumption, a syringe containing the mixture for infusion was connected to the duodenal and ileal infusion port. The infusion was performed in 60 min at an infusion rate of 2 mL/min. Subjects received a standardized *ad libitum* lunch meal (Lasagna Bolognese (Plus supermarket); energy density per 100 g: 152 Kcal, 11 g carbohydrates, 7.1 g protein, and 8.6 g fat) fifteen min (at t = 225 min) after cessation of the infusion. The test meal was offered in excess and subjects were instructed to eat until they felt satiated.

#### *2.6. VAS for Satiation and GI-Complaints*

Feelings of satiation-/satiety feelings and GI-complaints (e.g., satiety, hunger, stomach pain, and nausea) were measured using visual analog scales (VAS, 0–100 mm) scores at various time points (t = −30, 30, 90, 150, 165, 180, 195, 210, and 240 min) during the day. Subjects were asked to indicate on a line, anchored at the low end with the lowest intensity feelings, with opposing terms at the high end, which place on the scale best reflected their feeling at that moment [20].

#### *2.7. Statistical Analyses*

Data were analyzed using IBM SPSS statistics 24 (IBM Corporation, Armonk, NY, USA). A visual check of the normality of the data was performed. The primary outcome of this study was the amount of food intake in Kcal during an *ad libitum* lunch meal. Secondary outcomes were VAS scores for satiation-/satiety feelings and GI-complaints.

Age, BMI, and gender were calculated by descriptive statistics. Food intake in Kcal and area under the curve (AUC) for VAS scores were compared using a linear mixed model with intervention (DTIP, DPIT, and DTIT, and DPIP), test day and the interaction of intervention × test day as fixed factors. When no significant interaction was found, the interaction was removed from the model to get the best model fit.

For VAS scores, a linear mixed model that included abovementioned fixed factors with the addition of fixed factors time and time × treatment interaction was also performed.

Data are presented as mean ± standard error of the mean (SEM) (unless specified otherwise), and a *p* < 0.05 was considered statistically significant.

#### **3. Results**

#### *3.1. Subjects*

In total, 19 subjects met the inclusion and exclusion criteria. Two subjects dropped out due to discomfort induced by the naso-ileal catheter, two subjects dropped out due to incorrect position of the catheter on the first test day, and one subject was excluded after not properly following the instructions for the *ad libitum* meal on the first test day. Therefore, 14 healthy volunteers (11 female, age 25.6 ± 10.5 years, BMI 22.3 ± 1.7 kg/m2) completed the study protocol and were included in the analyses.

#### *3.2. Food Intake*

No intervention × test day interaction was found. No differences in *ad libitum* food intake in Kcal were observed after intraduodenal, intraileal or combined infusion of tastants versus placebo infusion (DPIP: 786.6 ± 79.2 Kcal, DTIP: 803.3 ± 69.0 Kcal, DPIT: 814.7 ± 77.3 Kcal, DTIT: 834.8 ± 59.2 Kcal; *p* = 0.59) (Figure 1). Furthermore, as depicted in Figure 2, no trends in individual responses were found.

**Figure 1.** The amount eaten in Kcal (mean + SEM) 15 min after cessation of the infusion of placebo both intraduodenal and intraileal (DPIP), tastants intraduodenal and placebo intraileal (DTIP), placebo intraduodenal and tastants intraileal (DPIT), and tastants both intraduodenal and intraileal (DTIT). Based on a linear mixed model, no difference in food intake was observed between the conditions (*p* = 0.59).

**Figure 2.** An individual representation per subject of amount eaten in Kcal 15 min after cessation of the infusion of placebo both intraduodenal and intraileal (DPIP), tastants intraduodenal and placebo intraileal (DTIP), placebo intraduodenal and tastants intraileal (DPIT), and tastants both intraduodenal and intraileal (DTIT). Treatment order was randomized for each subject. Each line with a unique symbol represents an individual subject. Based on a linear mixed model, no difference in food intake was observed between the conditions (*p* = 0.59).

#### *3.3. Satiation/Satiety Scores*

The mean VAS scores for the desire to eat, hunger, satiety, and fullness are depicted in Figure 3. No differences in area under the curve (AUC150–210) for these VAS scores were observed between the

various interventions. Furthermore, no intervention × timepoint interactions were found for these VAS scores.

**Figure 3.** VAS scores for desire to eat (**A**), hunger (**B**), satiety (**C**), and fullness (**D**) (mean + SEM) before, during, and after the infusion of placebo both intraduodenal and intraileal (DPIP), tastants intraduodenal and placebo intraileal (DTIP), placebo intraduodenal and tastants intraileal (DPIT), and tastants both intraduodenal and intraileal (DTIT). VAS scores were measured at t = −30, 30, 90, 150, 165, 180, 195, 210, and 240 min. No VAS scores were taken at t = 225 min. At t = 0 min, subjects received a standardized breakfast, infusion of mixtures was performed from t = 150 until t = 210 min, and *ad libitum* test meal was presented at t = 225. Based on a linear mixed model of mean scores and area under the curve (AUC150–210), no differences in desire to eat, hunger, satiety, and fullness were observed between the various conditions.

#### *3.4. GI-Complaints*

The mean VAS scores for stomach pain, bloating, and nausea are depicted in Figure 4. No differences in area under the curve (AUC150–210) for these VAS scores were observed between the various interventions. Furthermore, no intervention × timepoint interactions were found for these VAS scores.

**Figure 4.** *Cont.*

**Figure 4.** VAS scores for stomach pain (**A**), bloating (**B**), and nausea (**C**) (mean + SEM) before, during, and after the infusion of placebo both intraduodenal and intraileal (DPIP), tastants intraduodenal and placebo intraileal (DTIP), placebo intraduodenal and tastants intraileal (DPIT), and tastants both intraduodenal and intraileal (DTIT). t = −30, 30, 90, 150, 165, 180, 195, 210, and 240 min. No VAS scores were taken at t = 225 min. At t = 0 min, subjects received a standardized breakfast, infusion of mixtures was performed from t = 150 until t = 210 min, and *ad libitum* test meal was presented at t = 225 min. Based on a linear mixed model of mean scores and area under the curve (AUC150–210), no differences in stomach pain, bloating, and nausea were observed between the various conditions.

#### **4. Discussion**

Our results do not reveal any difference in satiety or food intake between duodenal administration, ileal administration or combined duodenal administration of a tastant mixture (sweet, bitter, and umami) or infusion of placebo. Moreover, no GI-complaints were caused by infusing tastants or placebo into the duodenum and/or the ileum.

Van Avesaat et al. have investigated the effect of intraduodenal infusion of the same tastant mixture on food intake [18]. In that study, intraduodenal infusion of this combination of tastants, in similar study design, using the same amount of tastants significantly reduced food intake by 64 Kcal and was accompanied by changes in satiation/satiety feelings. However, it must be noted that this is a small difference, which on its own might not be clinically significant. Repeating this effect multiple times per day with each meal might result in a clinically significant decrease of caloric intake. This difference in results of food intake between the two studies may be related to differences in study design. In the study of van Avesaat et al., the subjects were intubated with a naso-duodenal catheter on every test day for the administration of tastants. The catheter was removed immediately thereafter before the subjects were presented with the *ad libitum* test meal. In the present study, subjects were intubated for several days with a naso-ileal catheter, and therefore this catheter was present while meals were offered and ingested. We hypothesize that having a naso-ileal catheter in situ for multiple days negatively influences meal ingestion to such a degree that this masks the smaller magnitude of effect that infusion of non-caloric tastants into the intestine has. On the other hand, mean caloric intake showed no major differences between the two studies.

Previous studies from our group investigating the 'intestinal brake' by infusing macronutrients in the ileum have repeatedly shown that infusion of even low doses of macronutrients results in a significant reduction of food intake, ranging between 64–188 Kcal, corresponding to a percentual decrease of 11.7%–32% of caloric intake during a single meal [5,21]. This indicates a negative feedback mechanism on food intake that arises from nutrient sensing. These data demonstrate that magnitude of the effect of macronutrient infusion on food intake is greater than the effects of infusing tastants.

Conclusively, studies investigating differences in food intake should be aware that naso-ileal intubation might mask a small effect. Therefore, other delivery options, such as encapsulation, should be considered in the future.

Results of studies investigating the effects of single tastants on food intake, satiation/satiety, and GI peptides are not consistent. An initial strong decrease of hunger with a steep increase thereafter has been observed after administration of a non-caloric sweetener [22]. Ingestion of low caloric sweeteners did not influence energy intake compared with a control condition (intake of water) [23]. Adding an umami tastant to a meal did not affect appetite sensations, but has been shown to result in an increase of subsequent food intake [24]. Recently, increased attention has been given to the effects of bitter substances on satiety and food intake. Intake or infusion of bitter substances (quinine, denatonium benzoate) not only reduced antral motility [25,26] but also increased satiety scores and resulted in a significant decrease in food intake [27]. A possible mechanism explaining the strong aversive effects of bitter tastants is that bitter taste is evolutionarily linked to toxic substances, as has been showed by presenting newborn infants with bitter substances [28].

Alleleyn et al. have shown that the inhibition of food intake shows a proximal to the distal gradient, with higher effects observed after distal versus proximal administration of nutrients [8]. Based on our data, such a gradient was not observed for intestinally administered tastants. Intestinal taste receptor expression varies for various taste receptors, where some taste receptors are more profound proximally in the GI-tract, while expression of other taste receptors is higher in the more distal intestine [14–17].

We thought the proximal to distal gradient found for macronutrient infusion might be operable for taste receptor activation, which was clearly not the case. It is possible that taste receptors inhibit food intake in a different fashion than macronutrients. For instance, it has been speculated that taste receptors function by sensing the type of food (i.e., sweet for carbohydrates, umami for amino acids, and bitter for toxic substances) [29]. Since bitter tastants are linked to toxic substances, another working mechanism for bitter tastants could be through an aversive reaction of subsequent food intake.

From an evolutionary perspective, a more pronounced inhibitory or aversive effect for toxic substances could be expected to occur in the most proximal parts of the GI tract. However, there are no data available with respect to activation of oral (bitter) taste receptors on subsequent food intake. It is therefore unclear, whether activation of more proximal taste receptors will reveal more pronounced effects on food intake and satiation/satiety. Consequently, further studies are needed to investigate whether more proximal activation of taste receptors results in a stronger decrease in food intake.

Published data on the role of GI peptides in the regulation of food intake after administration of tastants are not in line. Van Avesaat et al. found a clear effect of intraduodenal administration of tastants on food intake that was not accompanied by changes in GLP-1- or PYY level [18]. Other studies, however, did show a decrease in systemic ghrelin- and motilin levels [25,26] and an increase in systemic CCK levels [27] after administration of a bitter tastant.

A limitation of our study is that the wash-out period consisted of only one day. Prolonging the wash-out period over one day would have resulted in a longer period of naso-ileal catheter intubation increasing the discomfort to our volunteers. No interaction effect between intervention and test day was found on food intake, satiety scores or GI-complaints, indicating that no carry-over effect was present.

Another limitation of the present study was the absence of systemic GI hormone measurements. This would have provided a complete analysis of the effects of intestinal tastant administration on eating behavior. However, van Avesaat et al. showed a decrease in food intake and an increase in satiety scores, which was not accompanied by changes in systemic GI hormone levels [18]. Therefore, no systemic GI hormone measurement was conducted in the present study.

It has to be noted that the ideal duration of administration of the intervention and of the timing between intervention and serving the *ad libitum* meal is unknown. We employed a design similar to that of van Avesaat et al. based on their positive results [18]. Future research protocols should consider these factors.

Studies investigating the effects of tastants on food intake up to now focus on only acute effects in a single *ad libitum* meal. It is not known whether repetitive or chronic administration of tastants will lead to other results. More data are needed on the long-term effects of tastants, especially on daily energy intake.

**Author Contributions:** The authors' responsibilities were as follows: Conceptualization, D.K. and A.A.M.M.; methodology, T.K., D.K., F.J.T. and A.A.M.M.; formal analysis, T.K.; investigation, T.K., A.M.E.A. and M.v.A.; resources, T.K. and F.J.T.; writing-original draft preparation, T.K.; writing-review and editing, T.K., A.M.E.A., D.K. and A.A.M.M.; supervision, D.K. and A.A.M.M.; project administration, T.K.; funding acquisition, D.K. and A.A.M.M.

**Funding:** This study was an investigator-initiated study. Will Pharma B.V. received funding from 'Subsidie MKB Innovatiestimulering Topsectoren' (MIT), grant number MTHLA16192, and covered all relevant costs related to the execution of the study.

**Acknowledgments:** We thank all the volunteers for participating in this study.

**Conflicts of Interest:** The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript or in the decision to publish the results. No specific grant was received for open access publication. T.K. received a salary from Will Pharma BV as part of the 'Subsidie MKB Innovatiestimulering Topsectoren' (MIT) for the period related to the execution of the present study. D.K. and A.A.M.M. have received an unrestricted grant from Will Pharma B.V. for execution of a study unrelated to the present study. A.M.E.A., M.v.A., F.J.T. reported no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Taste Perception and Caffeine Consumption: An fMRI Study**

#### **Laura Gramling 1,†, Eleni Kapoulea <sup>1</sup> and Claire Murphy 1,2,3,\***


Received: 9 October 2018; Accepted: 18 December 2018; Published: 24 December 2018

**Abstract:** Caffeine is ubiquitous, yet its impact on central taste processing is not well understood. Although there has been considerable research on caffeine's physiological and cognitive effects, there is a paucity of research investigating the effects of caffeine on taste. Here we used functional magnetic resonance imaging (fMRI) to investigate group differences between caffeine consumers and non-consumers in blood-oxygenation-level-dependent (BOLD) activation during hedonic evaluation of taste. We scanned 14 caffeine consumers and 14 caffeine non-consumers at 3 Tesla, while they rated three tastes: caffeine (bitter), sucrose (sweet), and saccharin (sweet with bitter after taste), in aqueous solutions. Differences in BOLD activation were analyzed using voxel wise independent samples *t*-tests within Analysis of Functional Neuroimage (AFNI). Results indicated that during the hedonic evaluation of caffeine or sucrose, caffeine non-consumers had significantly greater activation in neuronal areas associated with memory and reward. During the hedonic evaluation of saccharin, caffeine consumers had significantly greater activation in areas associated with memory and information processing. The findings suggest caffeine consumption is associated with differential activation in neuronal areas involved in reward, memory, and information processing. Further research on intensity and hedonics of bitter and sweet stimuli in caffeine consumers and non-consumers will be of great interest to better understand the nature of differences in taste perception between caffeine consumers and non-consumers.

**Keywords:** fMRI; caffeine; taste; memory

#### **1. Introduction**

Caffeine consumption is ubiquitous. It currently ranks as the most popular psychostimulant in the world [1]. Eighty-five percent of the United States' population consumes at least one caffeinated beverage daily [2]. Many beverages contain caffeine, including coffee, the most widely consumed beverage after water [3]. Other widely consumed caffeinated beverages are tea and energy drinks, which typically contain a high caffeine content, as well as a high glucose content [2,4]. Despite caffeine's bitter taste and the fact that bitter tastes often discourage intake, coffee and tea remain two of the most widely ingested beverages [5]. Caffeine's widespread consumption warrants a better understanding of its effects.

Evidence supporting caffeine's ability to exert beneficial effects is abundant [6]. When consumed in moderate amounts, caffeine has been reported to decrease fatigue and increase energy [6]. Caffeine has also been reported to increase motor performance on sustained response tasks. For example, participants randomly assigned in a double-blind study to either consume a drink containing 40 mg

of caffeine or placebo, showed enhanced performance on a selective attention task when exposed to the experimental condition [7]. Further, caffeine produces mild autonomic nervous system arousal and improved mood when compared to a non-caffeinated placebo [8]. During a visuomotor task, participants demonstrated increased blood-oxygenation-level-dependent (BOLD) activation in the putamen and insula after consuming 200 mg of caffeine [9]. The putamen is part of the basal ganglia, an area that has been shown to modulate the top-down influence of the prefrontal cortex on sensory processing in humans [10]. Increased activation in the striatum following caffeine consumption suggests that caffeine can act as a cognitive enhancer by modulating these attentional areas [9].

While caffeine consumed at moderate doses may provide consumers with a number of favorable effects, research suggests negative consequences as a result of caffeine consumed at higher doses [8,11]. Increasing caffeine consumption can exert dose-dependent effects on a number of acute autonomic responses, including increased blood pressure [8]. Caffeine consumed at 300–800 mg can induce anxiety, nervousness, and insomnia [11]. Further, withdrawal from caffeine is detectable overnight, and causes fatigue, stress, as well as decreased alertness and clear-headedness in heavy caffeine consumers [12–14].

The motivational desire to ingest a certain food incorporates a combination of flavor, learned associations, and physiological state that integrate to produce a food reward [15,16]. Since bitter taste is typically avoided by many species and may be an adaptation to protect them from adverse physiological effects, repeated consumption of caffeine may be a learned process [5]. The choice to consume caffeine may occur as a result of altered activation in brain areas related to reward pathways, particularly in areas associated with processing food rewards. Previous studies have reported that altered neuronal processing can occur as a consequence of repeated ingestion of a substance [17,18]. For example, habitual consumption of non-nutritive sweeteners has been associated with altered processing of sweet taste in individuals who regularly consume diet soda [17]. When compared to non-diet soda drinkers, diet soda drinkers demonstrated greater activation in areas related to reward processing, such as the dopaminergic midbrain, in response to sweet taste. Diet soda drinkers also exhibited greater activation in orbitofrontal cortex (OFC) Brodmann Area (BA) 47, an area related to pleasantness evaluation, when rating saccharin. Therefore, food consumption choices may be associated with altered neuronal activation.

The effects of caffeine consumption on central aspects of taste perception are not well understood. In addition to caffeine's bitter taste [5], there is some suggestion from psychophysical studies that caffeine, which is an adenosine-receptor antagonist, may influence perception of some sweeteners through its action on adenosine receptors in sweet-sensitive taste cells [5,19,20]. The current study investigates differences between habitual caffeine consumers and non-consumers on brain activation during hedonic evaluation of taste, rather than the acute effects of caffeine consumption or withdrawal from caffeine consumption [21,22].

The purpose of the current study was to test the hypothesis that caffeine consumers and non-consumers may show differential brain activation, assessed with functional magnetic resonance imaging (fMRI), during hedonic evaluation of a bitter taste (caffeine), a sweet taste (sucrose), and a sweet taste with bitter after taste (saccharin). Results suggesting differential brain activation in association with caffeine consumption and different taste stimuli adds to preceding literature regarding caffeine's influences on taste perception. Since caffeine consumption was a defining factor in group membership, it was chosen as the representation for bitter taste. Sweet taste was also chosen as a taste stimulus in response to preceding literature suggesting that caffeine may influence perception of sweet taste [5,19,20]. Saccharin was chosen as the third taste stimulus since it evokes a combination of bitter and sweet taste and may result in differential activation during taste processing in comparison to caffeine and/or sucrose. We aimed to investigate differential brain activation during the hedonic evaluation of taste to determine (1) whether caffeine consumers have greater activation than non-consumers in areas related to reward processing (e.g., nucleus accumbens, OFC BA 10); (2) whether caffeine non-consumers have greater neuronal activation than consumers in memory pathways, such as areas in the medial temporal lobe (MTL); and (3) whether caffeine non-consumers may rely upon activation of a larger network than consumers in order to perform the task.

#### **2. Methods**

#### *2.1. Participants*

The current sample (*n* = 28) consisted of 12 males and 16 females. Participants were divided into one of two groups: caffeine non-consumer (*n* = 14) and caffeine consumer (*n* = 14). Participants were divided into these groups based on answers to a survey that was administered after study completion. Participants who reported they not drink caffeinated beverages were labeled as caffeine non-consumers. Participants who responded that they did consume caffeinated beverages constituted the consumers group. Groups were matched on age, body mass index (BMI), and gender. Participants were part of a larger study investigating fMRI and taste processing. The Institutional Review Boards at San Diego State University and University of California, San Diego approved the study. All participants gave informed consent and were given monetary compensation for their participation.

#### *2.2. Screening Session*

The current study used the methodology described in detail in Haase, Cerf-Ducastel, Buracas, and Murphy (2007) [23]. All participants completed one screening session and one event-related fMRI session. At the initial screening, participant information, height, and weight were recorded. Participants were screened for metal in their body for the fMRI scan, as well as ageusia and anosmia with forced choice taste and odor threshold measures [24]. Being left-handed was an exclusionary criterion to avoid differential lateral activation in hemispheres due to handedness [25]. Participants who met the study criteria returned to complete one fMRI scan.

#### *2.3. Odor and Taste Threshold Measures*

In order to screen for anosmia, odor thresholds for the odor n-butyl alcohol (butanol) were assessed for each nostril monorhinically using a forced choice, ascending methods of limits test [24]. The solutions were in a series of 10; each dilution was one-third the concentration of the solution preceding it. On each trial the participant was presented with two bottles: one containing distilled water and the other containing the odor stimulus. The participant was asked to decide which bottle contained an odor. There was a 45 s inter-stimulus interval between each stimulus delivery to avoid adaptation [26]. If the participant chose the incorrect bottle, a higher concentration was given on the next trial. Once the participant met the criterion of choosing correctly on five successive trials the odor threshold was determined.

In order to screen for ageusia, taste thresholds for sucrose were assessed using a sip and spit, forced choice staircase procedure [24]. Stimuli were presented in 14 concentrations of sucrose, ranging from 0.0032 to 0.36 M in geometrical progression. All stimuli were presented at room temperature in distilled water [24]. The experimenter presented the participants with two cups, one containing distilled water and the other containing sucrose solution. The stimulus was sipped, held in the mouth for 10 s, and expectorated. After the participant sampled 10 ml of water and solution, he (she) was asked to select the stimulus with the sweet taste. The experimenter increased the concentration until the participant consistently (twice in a row) chose the stronger stimulus. This procedure was then reversed to a descending series until the participant failed to choose the correct stimulus. Participants were required to rinse with distilled water before each stimulus to avoid adaptation and waited a minimum of 30 s between each stimulus. Testing continued for five reversals with the mean of the last four reversals taken as the threshold.

#### *2.4. Neuroimaging Procedure*

Functional MRI data were collected in order to investigate brain response of caffeine consumer and caffeine non-consumer groups to stimuli during the physiological state of hunger. All scanning sessions occurred in the morning, and participants were instructed to fast 12 h prior to the scan. When stimuli were presented, participants used a joystick to rate pleasantness on a modified general Labeled Magnitude Scale. The scale was projected on a screen visible to the participant through a mirror attached to the head coil [23,27].

#### *2.5. Stimulus Delivery*

The stimuli used in this study were pure tastes delivered in aqueous solutions: 0.04 M caffeine, 0.64 M sucrose, and 0.028 M saccharin. These concentrations were chosen based on a previous study from our laboratory reporting how stimulus delivery method impacted the slopes of taste intensity functions for these stimuli [28]. The simulated stimulus delivery system was shown to produce psychophysical functions with slopes that were generally lower than experiments conducted with the sip and spit technique and that were similar to slopes of intensity functions associated with the dorsal flow procedure [28]. The concentrations chosen for the present study reflect the highest concentrations of each stimulus tested in Reference [28].

Stimuli were presented orally and presentations were randomized during functional data acquisition through the use of a computer-controlled delivery system (Figure 1). All taste stimuli were presented while the participant was inside the scanner, where the participant lay supine with a bite bar, which was positioned comfortably between the lips so that the tubes delivered stimuli to the tip of the tongue. Immediately before, during, and after the scan, participants rated the pleasantness and intensity of each stimulus. The taste stimuli and water were delivered at room temperature each through a unique 25-ft long plastic tube, which was connected to a different computer-programmable syringe pump. The pumps were programmed to present 0.3 mL of solution in 1 s.

**Figure 1.** Stimulus delivery paradigm. Reprinted with permission from Haase et al. (2007) [23].

The imaging session consisted of two functional runs. During the functional runs, each stimulus was presented in 0.3 mL of solution for a total of 16 times with a 10 s inter-stimulus interval. Participants were presented with water twice; first as a rinse, and then as a baseline to be used in data analysis. A complete outline of the stimulus delivery protocol used in the fMRI sessions is described in the Journal of Neuroscience Methods [23].

#### *2.6. Imaging Acquisition*

Functional MRI sessions took place at the Center for Functional Magnetic Resonance Imaging at the University of California, San Diego. All data were collected using a 3T General Electric Signa Excite short-bore scanner (GE Healthcare, Chicago, IL, USA). Structural data were acquired for anatomical localization of the functional images. Parameters used to acquire structural images were as follows: T1—weighted whole-brain fast spoiled gradient echo (FSPGR) sequences, field of view (FOV) = 25.6 cm, slice thickness = 1 mm, resolution = 1 × <sup>1</sup> × 1 mm3, echo time (TE) = 30 ms, Locs per slab = 190, flip angle = 15◦. Parameters used to acquire functional images were as follows: T2\*—weighted images, 32 axial slices, FOV = 19.2 cm, matrix size = 64 × 64, resolution = 3 × <sup>3</sup> × 3 mm3, flip angle = 90◦, echo time (TE) = 30 ms, repetition time (TR) = 2000 ms.

#### *2.7. Imaging Analysis*

Imaging data were processed using FMRIB Software Library (FSL, Analysis Group, FMRIB, Oxford, UK) and Analysis of Functional NeuroImage (AFNI, open source software) [29,30]. Data were preprocessed to correct head movement and alignment as well as to concatenate the runs. Temporal and spatial smoothing of the brain images were also applied. Images were spatially smoothed to four full widths at half maximum (FWHM), auto-masked to remove voxels located outside of the brain, and normalized into Talairach space to control for individual variation in structural differences.

We conducted the analyses within AFNI, using 3dDeconvolve, on each participant's concatenated runs based on the specified contrast (e.g., activation during evaluation of caffeine minus activation during evaluation of water) that accounted for the timing of delivery of the stimulus and the water baseline, which served as a control for identifying non-gustatory intra-oral stimulation [30,31]. Deconvolution estimates the hemodynamic response per voxel in a participant's concatenated runs given the experimental paradigm (i.e., stimulus onset timing) using ordinary least squares regression. The output from 3dDeconvolve contains fit coefficients (i.e., beta weights) for each voxel, indicating the amplitude of the signal model for each contrast, and corresponding *t*-statistics.

Several thresholding steps were taken in an attempt to control for Type I error in all group analyses. Individual voxels were thresholded at *p* ≤ 0.015. To protect a whole-brain probability of false positives at an overall alpha of 0.05, group statistical maps were corrected for multiple comparisons at the cluster level using the AFNI program ClustSim [31]. ClustSim uses Monte Carlo simulations to compute the probability of generating a random "significant" cluster of noise (i.e., a false positive) given the individual voxel threshold, the voxel connection radius, the amount of blurring, and the search volume (i.e., overall dataset size). For an overall alpha level of 0.05, a cluster threshold of 21 contiguous voxels was applied. Neuronal activation in the caffeine consumers group during hedonic evaluation of the individual taste stimuli was subtracted from activation in the caffeine non-consumers group.

#### *2.8. Demographic Data Analysis*

To examine potential demographic differences between caffeine consumers and caffeine non-consumers, multivariate analyses of variance (MANOVA) were performed using caffeine status as an independent variable. Age, gender group, body mass index (BMI), taste threshold, right odor threshold, and left odor threshold were dependent variables. The results can be found in Table 1.


**Table 1.** Participant characteristics for caffeine non-consumers and matched caffeine consumers.

BMI: body mass index.

#### *2.9. Psychophysical Data Analysis*

The general Labeled Magnitude Scale (gLMS) was used to collect intensity ratings and a modified version of the gLMS was used to collect hedonic ratings for caffeine, sucrose, and saccharin taste before and after each scan [27]. To examine between group differences in psychophysical ratings, a MANOVA was performed using caffeine status as an independent variable. Results are shown in Tables 2 and 3. Repeated measures analyses of variance (RM-ANOVA) were performed to examine possible differences between hedonic and intensity ratings of each taste before and after stimuli were presented during the scan.


**Table 2.** Hedonic ratings for caffeine non-consumers and matched caffeine consumers.

**Table 3.** Intensity ratings for caffeine non-consumers and matched caffeine consumers.


\* Significant difference between caffeine consumers and caffeine non-consumers.

#### **3. Results**

#### *3.1. Demographic*

There were no significant differences in age (F (1, 26) = 3.193, *p* = 0.086), BMI (F (1, 26) = 0.001, *p* = 0.972) or gender (F (1, 26) < 0.001, *p* = 1.000). There were also no significant differences in taste threshold (F (1, 26) = 0.169, *p* = 0.684) or in the odor threshold for the right nostril (F (1, 26) = 2.935, *p* = 0.099) or for the odor threshold for the left nostril (F (1, 26) = 1.252, *p* = 0.273).

#### *3.2. Psychophysical Data*

A MANOVA was performed to examine between group differences of hedonic and intensity ratings (Tables 2 and 3). Caffeine non-consumers demonstrated significantly higher ratings for post-scan intensity ratings for sucrose (F (1, 26) = 4.390, *p* = 0.046) and saccharin (F (1, 26) = 7.312, *p* = 0.012) when compared to the post-scan intensity ratings for caffeine consumers.

There were no significant differences between caffeine consumers and non-consumers in pleasantness ratings of caffeine (F (1, 26) = 1.3686, *p* = 0.253), saccharin (F (1, 26) = 0.094, *p* = 0.762), or sucrose (F (1, 26) = 0.392 *p* = 0.537). There were also no significant differences between sucrose intensity ratings before and after stimuli were presented during the scan (F (1, 26) = 0.442, *p* = 0.512). There were significant differences between intensity ratings before and after the scan for caffeine (F (1, 26) = 10.173, *p* = 0.004) and saccharin (F (1, 26) = 6.558, *p* = 0.016). For caffeine consumers, neither saccharin intensity ratings (F (1, 13) = 0.568, *p* = 0.464) nor caffeine intensity ratings

(F (1, 13) = 0.077, *p* = 0.786) were significantly different before and after taste was presented during the scan. For caffeine non-consumers, caffeine intensity ratings (F (1, 13) = 11.833, *p* = 0.004) and saccharin intensity ratings (F (1, 13) = 6.551, *p* = 0.024), were significantly different before and after the taste was presented. Intensity ratings for caffeine non-consumers were significantly higher after the stimuli were presented during the scan.

#### *3.3. Functional Neuroimaging*

During the hedonic evaluation of caffeine, caffeine non-consumers had significantly greater neuronal activation in the right cuneus, right precuneus, left anterior cingulate, medial frontal gyrus, and left superior frontal gyrus (See Table 4 and Figure 2).

**Table 4.** Regions of significantly greater activity in caffeine non-consumers compared to caffeine consumers while judging the pleasantness of caffeine.


Hem.: Hemisphere; R: right; L: left; Regr. Coef.: Regression coefficient; Minimum cluster = 21 voxels, *p* = 0.015.

**Figure 2.** Brain activation during the hedonic evaluation of caffeine. Orange indicates areas where caffeine non-consumers had significantly greater activation in comparison to caffeine consumers.

During the hedonic evaluation of saccharin, caffeine non-consumers had significantly lower neuronal activation than caffeine consumers in the middle temporal gyrus, inferior temporal gyrus, middle occipital gyrus, right fusiform gyrus, right lingual gyrus, and right cuneus (See Table 5 and Figure 3).

**Table 5.** Regions of significantly greater activity in caffeine consumers compared to caffeine non-consumers while judging the pleasantness of saccharin.


Hem.: Hemisphere; R: right; L: left; Regr. Coef.: Regression coefficient; Minimum cluster = 21 voxels, *p* = 0.015.

**Figure 3.** Brain activation during the hedonic evaluation of saccharin. Blue indicates areas where caffeine consumers had significantly greater activation in comparison to caffeine non-consumers.

During the hedonic evaluation of sucrose, caffeine non-consumers had significantly greater neuronal activation in the anterior cingulate, medial frontal gyrus, right superior frontal gyrus, OFC BA 10, posterior cingulate, cingulate gyrus, and precuneus (See Table 6 and Figure 4).


**Table 6.** Regions of significantly greater activity in caffeine non-consumers compared to caffeine consumers while judging the pleasantness of sucrose.

Hem.: Hemisphere; R: right; L: left; Regr. Coef.: Regression coefficient; Minimum cluster = 21 voxels, *p* = 0.015.

**Figure 4.** Brain activation during the hedonic evaluation of sucrose. Orange indicates areas where caffeine non-consumers had significantly greater activation in comparison to caffeine consumers.

#### **4. Discussion**

How one perceives taste stimuli has been shown to influence food choice and repeated consumption of a tastant may lead to altered taste preferences [32–35]. There are neuroimaging data to suggest that the human brain responds differently as a result of habitual consumption [17,18]. However, to our knowledge, there is no human research investigating brain response during hedonic evaluation of taste in caffeine consumers and non-consumers. In this study, we examined brain response in self-reported caffeine consumers and caffeine non-consumers during an fMRI scan to

investigate whether regular consumption of caffeine is associated with differential activation of areas related to memory, reward, and information processing. Imaging data from the present study indicate that caffeine consumers and caffeine non-consumers have significantly different neuronal activations in areas related to memory, reward, and information processing when processing individual taste stimuli. Each participant was exposed to 0.3 mL/sec of each tastant for 16 repetitions resulting in a total consumption of <5 mL, suggesting that these differences in activation occurred as a result from processing the taste alone, rather than the possible physiological effects of ingestion. When rating caffeine and sucrose, caffeine non-consumers had significantly greater activation in areas related to memory, reward, and information processing. During hedonic evaluation of saccharin, caffeine consumers had significantly greater activation in areas related to information processing. Overall, our results indicate differential neuronal activations between both groups during the processing of all three tastes. These results suggest differences in overall cognitive expenditure between the two groups, differing based on which taste was presented.

#### *4.1. Psychophysical Data*

Caffeine consumers and caffeine non-consumers demonstrated differences in taste perception. Post-scan intensity ratings of sucrose and saccharin were significantly higher in caffeine non-consumers compared to caffeine consumers. Further, caffeine non-consumer ratings of caffeine and sucrose intensity significantly increased from before to after stimulus presentation within an fMRI scan. The latter phenomena were not present in caffeine consumers.

Psychophysical results suggest that caffeine non-consumers perceived sucrose and saccharin as being more intense than caffeine consumers after the scan. Also, caffeine and saccharin intensity ratings significantly increased after stimuli were presented during the scan for caffeine non-consumers. A plausible explanation is that the sweet taste of sucrose and saccharin may have been potentiated by caffeine [20]. The increase of perceived intensity of caffeine and saccharin after the scans in non-consumers suggests a stronger reaction to bitter taste, which is present in caffeine and in saccharin as an aftertaste. There is evidence that the perceived intensity of caffeine's bitterness may be associated with whether caffeine is regularly consumed and the expression of bitter receptors, PAV-TAS2R38 [5,36]. It is plausible that both a genetic predisposition and caffeine consumption habits contributed to caffeine non-consumers perceiving all three tastes more intensely in comparison to caffeine consumers.

#### *4.2. Reward Processing Areas*

During the hedonic evaluation of caffeine and sucrose, caffeine non-consumers demonstrated greater activation in areas associated with reward processing.

During the hedonic evaluation of sucrose, caffeine non-consumers demonstrated significantly greater activation in both hemispheres of OFC BA 10, an area associated with encoding the incentive value of a stimulus during a decision-making task [37–39]. The OFC has been activated in response to abstract internal goals, such as rewards and punishments, while other tasks are being performed [37–39]. The OFC has been reported to be responsive to the reward value of tastes, as it associates other stimuli with tastes to produce representations of expected reward value [37,40]. A reward stimulus has been found to induce increased activation in OFC BA 10 when already activated by working memory processing [41]. Further, the OFC is activated by monetary rewards and punishment, with more activation reported following a punishment outcome [38].

During the hedonic evaluation of caffeine and sucrose, activation in the anterior cingulate cortex (ACC) was significantly greater in caffeine non-consumers. During the hedonic evaluation of caffeine, only activation in the left anterior cingulate cortex was found to be significantly greater in caffeine non-consumers in comparison to caffeine consumers. Lateralization in the ACC has been found during error processing and conflict monitoring, where correct inhibitions only occurred in the right ACC [42]. Further, observational fear learning has been found to only be activated in the right, but not the left ACC [43]. The distinction that right ACC activation only occurred during the hedonic evaluation of

sucrose and not during the hedonic evaluation of caffeine suggests that sucrose may have been a more intense experience for caffeine non-consumers. Psychophysical data supports this assertion, as caffeine non-consumers provided significantly higher intensity ratings for sucrose post-scan when compared to caffeine consumers (Table 3).

Overall, the ACC has been associated with an overall neural circuit that uses past action-reward history to learn action value in order to guide voluntary choice behavior [44]. This process requires referencing a history of outcomes regarding a given choice [44]. Further, previous studies suggest that reward processing in the ACC may also guide choice behavior, as it relates actions to their consequences [45]. This suggests that ACC has an essential role in learning and using extended action-outcome histories to make voluntary choices.

It is important to emphasize that activity in the OFC is representative not merely of a reward per se, but of a detailed and information rich representation of reward [46]. Similarly, the ACC references past-action reward history and is not a direct reflection of the reward value [44,45]. Therefore, the results are not necessarily indicative of caffeine non-consumers finding tastes to be more or less rewarding than caffeine consumers. A more plausible explanation may be that greater activation in the OFC and ACC found in caffeine non-consumers suggests a greater cognitive expenditure to use past reward history and process the representation of a reward, in order to make a voluntary choice, which in this case, was the hedonic rating.

#### *4.3. Memory Processing Areas*

During the hedonic evaluation of caffeine, caffeine non-consumers demonstrated significantly greater activation in right precuneus. During the hedonic evaluation of sucrose, caffeine non-consumers demonstrated significantly greater activation in both the left and right side of the precuneus. The right precuneus has been previously linked to autobiographical memory retrieval [47]. It is of particular interest that this area was activated during the hedonic evaluation of caffeine, an experience that would not be common in caffeine non-consumers. The precuneus is an area previously associated with episodic memory retrieval, the ability to recall a previously experienced stimulus [48]. Continuous theta burst stimulation (cTBS) over the precuneus in a picture memory task was associated with a decrease in source memory errors and improvement in context retrieval, suggesting that the precuneus is integral to a memory encoding and retrieval network [48]. During a source and item-recognition memory task, the left precuneus was activated during memory retrieval [49].

During the evaluation of sucrose, caffeine non-consumers also demonstrated greater activation in both the left and right of the posterior cingulate and cingulate gyrus. The posterior cingulate cortex has been associated with memory retrieval, namely autobiographical memory retrieval [50]. The posterior cingulate cortex also subserves evaluative functions such as monitoring sensory events and behavioral actions in the service of spatial orientation and memory [51].

These results support the hypothesis that caffeine non-consumers demonstrate greater cognitive expenditure in memory processing areas. We speculate that greater activations in the caffeine non-consumers while evaluating caffeine and sucrose could indicate a greater source memory retrieval expenditure. It is possible that caffeine non-consumers may have had less exposure to these tastes due to their dietary choices, and therefore, require greater cognitive effort to process them. Further, while sucrose is ubiquitous in all types of food, it is possible that experiencing caffeine's bitter taste is a new experience for caffeine non-consumers, not only in experiencing caffeine's flavor profile, but also its subsequent impact on other tastes.

#### *4.4. Information Processing*

Activation in information processing pathways was observed during hedonic evaluation of all three tastants. Activation in the right superior frontal gyrus (SFG) was significantly higher during the hedonic evaluation of sucrose. The right SFG has been linked to functioning in cognitive control, such that greater activation was linked to more efficient response inhibition, less motor urgency, as well

as greater self-regulation [52,53]. The left SFG was significantly higher during the hedonic evaluation of caffeine in caffeine non-consumers. The superior frontal gyrus, particularly the left SFG, has been associated with performing higher cognitive functions associated with working memory retrieval, especially in relation to task-related behavioral goals [54].

Both sides of the medial frontal gyrus were significantly activated in caffeine non-consumers during the hedonic evaluation of caffeine and sucrose, but not in the saccharin condition. Previous studies have linked activation in the left dorsolateral prefrontal cortex to processing and rating multimodal flavor stimuli [55]. Further, this is an area where the consequences of actions directly affect cognition in the preparation for and selection of response [55]. Results suggest a greater cognitive effort during the hedonic evaluation of caffeine and sucrose for caffeine non-consumers and during saccharin for caffeine consumers in information processing pathways coinciding with results previously stated. Due to the variability in between group activation within information processing areas, it is difficult to make a conclusive decision whether or not caffeine non-consumers activate a larger network than consumers in order to perform the hedonic evaluation task. While there was primarily more activation within the overall study in caffeine non-consumers, caffeine consumers demonstrated greater activation during the saccharin condition. We speculate that greater activation for caffeine consumers during saccharin evaluation may have occurred because saccharin evokes both sweet and bitter taste [56]. In addition to the stimulation of both sweet and bitter receptors, additional expenditure of cognitive effort may be required to hedonically evaluate this taste experience.

#### *4.5. Further Considerations*

There are limitations to this study. We did not investigate the potential differences in response between caffeine consumers who regularly consume caffeinated beverages with a higher sugar content and caffeine consumers who more regularly consume more bitter tasting beverages. Future studies may differentiate between the impact of taste processing for habitual consumers that drink primarily bitter tasting beverages (i.e., black coffee and tea) or items greater in sugar content (i.e., energy drinks). Further, in the caffeine consumers group, there were varying levels of caffeine consumption. Future studies may choose to expand on this paradigm, considering the effect of varying types and levels of caffeine consumption on taste perception.

We did not specify whether the taste stimuli were administered to the left or right side of the tongue. While we could not locate literature detailing a lateralization in processing of sweet and bitter taste alone, previous studies have reported laterization when discriminating tastes and rating taste quality [57,58]. Stevenson, Miller, and McGrillen [58] reported that when administering sour, sweet, salty, bitter, and umami solutions, discrimination among tastes was better when stimuli were applied to the right tongue tip and participants were better at taste quality judgements when tastants were applied to the left tongue tip [58]. All stimuli in the present study were administered to the tip of the tongue and whether the stimuli were more exposed to the left or the right side on trials was not specified. However, future studies could elaborate on this paradigm by taking this lateralization of gustatory processing into account.

The effects of caffeine consumption on taste perception are of considerable interest. Following a report that adenosine can enhance sweet taste in mice through its actions on A2B receptors in the taste bud, a recent report of a human psychophysical study suggested that caffeine, which is an adenosine-receptor antagonist, may decrease the perceived intensity of sweet taste through its action on adenosine receptors in sweet-sensitive taste cells [19,59]. Early studies of the effects of caffeine on taste had reported that in aqueous solutions of two component mixtures, caffeine decreased the sweetness of sucrose; and that when applied directly to the tongue with filter paper, caffeine enhanced the intensity of quinine HCl, NaCl, and a number of nonnutritive sweeteners, particularly those with bitter components (e.g., saccharin), but not the nutritive sweeteners sucrose and fructose [20,60]. The acute ingestion of caffeine has been reported to reduce the intensity of saccharin but not other taste stimuli, and that raising caffeine levels in the saliva for a period of three weeks had no measurable

effects on reported intensity of caffeine, denatonium benzoate or NaCl [61,62]. Differences in the effects of caffeine on sweetness intensity may be related to the stimuli, their concentrations, the route of administration or other methodological differences in these studies [19,20,60–62]. The current study focused on the effects of habitual caffeine consumption on fMRI of central brain response and found differential activation between caffeine consumers and non-consumers during hedonic evaluation of sucrose, caffeine, and saccharin stimuli. Further research on both intensity and hedonics of bitter and sweet stimuli, including natural as well as artificial sweeteners, in caffeine consumers and non-consumers will be of great interest to better understand the nature of caffeine's influence on taste perception.

#### **5. Conclusions**

In summary, we administered three tastants, caffeine, sucrose, and saccharin, to investigate differences in neuronal activation between those who were self-reported caffeine consumers and caffeine non-consumers. We found differences in intensity ratings between groups. We also found differences in activation patterns during a hedonic evaluation task. Our results suggest that there is greater activation for caffeine non-consumers while processing caffeine and sucrose and greater activation for caffeine consumers while rating saccharin. The results support differential memory, reward, and information processing of taste between those who habitually consume caffeine and those who do not. These results suggest that further research into the link between caffeine consumption and taste perception is warranted.

**Author Contributions:** Conceptualization, L.G., E.K., and C.M.; Formal analysis, E.K., Funding acquisition, C.M.; Investigation, L.G. and C.M.; Methodology, L.G. and C.M.; Project administration; C.M., Resources, C.M.; Supervision, C.M.; Visualization, E.K. and C.M.; Writing—original draft, L.G., E.K., and C.M.; Writing—review & editing, E.K. and C.M.

**Funding:** This research was supported by NIH Grant No AG004085-26, from the National Institute on Aging (Claire Murphy).

**Acknowledgments:** We would like to thank Aaron Jacobson, and Erin Green and Lori Haase for fMRI acquisition and expertise, as well as the Lifespan Human Senses Laboratory assistants for research assistance. We also thank Thomas Liu and the UCSD Center for fMRI.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Sweet and Umami Taste Perception Differs with Habitual Exercise in Males**

#### **Emma L. Feeney 1, Laura Leacy 2, Mark O'Kelly 2, Niamh Leacy 2, Abbie Phelan 2, Leah Crowley 2, Emily Stynes 1, Aude de Casanove <sup>1</sup> and Katy Horner 2,\***


Received: 29 November 2018; Accepted: 29 December 2018; Published: 12 January 2019

**Abstract:** Taste is influenced by several factors. However, whether habitual exercise level is associated with differences in taste perception has received little investigation. The aim of this study was to determine if habitual exercise is associated with differences in taste perception in men. Active (*n* = 16) and inactive (*n* = 14) males, between ages 18–55, underwent two days of sensory testing, using prototypical taste stimuli of high and low concentrations for sweet, salt, bitter, sour, umami, and carbohydrate (maltodextrin). Mean perceived intensity and hedonic ratings were recorded. Eating behaviour was assessed by the three factor eating questionnaire and food intake by EPIC food frequency questionnaire (FFQ). There were moderate to large differences between the two groups in perceived intensity for sweet taste at the high concentration and umami taste at both high and low concentrations, with active males recording a higher perceived intensity (*p* < 0.05 for all). The active group also recorded a greater dislike for umami low and carbohydrate low concentration (*p* < 0.01). Salt, bitter and sour perception did not significantly differ between the two groups. FFQ analysis showed no difference in % energy from macronutrients between the groups. Eating behaviour traits correlated with sweet taste intensity and umami taste liking, independent of activity status. Results indicated that sweet and umami taste perception differ in active compared to inactive males. Habitual exercise level should be considered in taste perception research and in product development. Whether differences in taste perception could be one factor influencing food intake and thus energy balance with habitual exercise warrants further investigation.

**Keywords:** taste perception; umami; carbohydrate; sweet; salt; bitter; physical activity; intensity; liking

#### **1. Introduction**

The sense of taste allows us to identify and distinguish between sweet, sour, salty, bitter, and umami qualities [1], perceived on the tongue in the absence of odour. In addition, carbohydrate has recently been described as a taste [2] exemplified by maltodextrin. Taste sensitivity differs between individuals for different taste qualities [3,4]. There is considerable variation in the degree of taste perception, and a wide range of factors, including genetics [3,5], age [6], sleep [7] body mass index [8], anxiety level and neurotransmitters [9], hormonal factors [10], and habitual diet [11], among others, have been associated with differences in taste perception between individuals. Physical activity could potentially influence several of the modifiable factors associated with differences in taste perception. Although the outcomes are variable, some studies have reported alterations in taste perception during

and after a single bout of exercise (see References [12,13] for reviews). There is also some limited evidence that habitual exercise may be associated with differences in taste perception [14]. In a study of female swimmers and inactive females, swimmers were found to perceive high-sucrose stimuli as sweeter [14]. However, little other research to date has investigated taste perception and habitual exercise.

Characterising factors influencing food intake in active and inactive individuals is important to gain a greater understanding of the role of physical activity in energy balance [13,15,16]. Sedentary individuals have been proposed to be at a greater risk of overeating due to a lack of physiological regulation of appetite [17] and several aspects of appetite and food intake regulation have been shown to vary depending on habitual physical activity level [18–20]. Evidence from both cross-sectional and longitudinal studies suggests physical activity is associated with improved short-term appetite control [19]. Moreover, hedonic responses for high- or low-fat and sweet or savoury foods have been shown to differ between habitual exercisers and inactive individuals [20]. The underlying factors and mechanisms associated with differences in appetite control and food intake with physical activity, however, remain to be fully elucidated.

Understanding whether taste perception differs depending on physical activity level is important, as differences in taste perception could influence food choice or eating behaviour [3,4,21] and may be related to weight status [22,23]. Alterations in taste perception have been linked to weight gain, with a recent longitudinal study demonstrating attenuated sweet and salty taste perception was associated with weight gain in college-aged males [24]. Moreover, taste perception has been proposed as a factor that may influence athletes' food choices [25]. Determining whether taste perception differs in active and inactive individuals could, therefore, provide greater insight into factors influencing food choice and energy balance. For example, a reduced sensation of sweet or salty taste could potentially render inactive individuals more susceptible to weight gain.

The present study aimed to compare taste perception (taste intensity and liking) between active and inactive males for the five 'basic' taste qualities of sweet, sour, salty, bitter, and umami tastes, as well as the more recently proposed carbohydrate taste.

#### **2. Materials and Methods**

#### *2.1. Study Design*

Participants in this between-groups cross-sectional design study undertook two separate test mornings one week apart. Ethical approval for the study was provided by the University College Dublin School of Public Health, Physiotherapy and Sports Science undergraduate research ethics committee. The primary outcome measure was taste perception (intensity) and secondary outcomes were liking and identification of taste, anthropometry and body composition, eating behaviour, and habitual dietary intake.

#### *2.2. Participants*

Thirty men were studied (*n* = 14 inactive and *n* = 16 active) between the ages of 19 and 51 years. Inclusion criteria were: Male, aged 18–55 years, nondiabetic, no medical conditions and not taking medication known to influence taste perception, willing to consume study taste solutions, and nonsmokers. Participants were classified through a screening questionnaire based on their self-reported physical activity patterns over the last 6 months as either inactive (undertaking ≤1 structured exercise session per week and not engaged in strenuous work) or active (undertaking ≥4 structured exercise sessions per week). Individuals who did not fit either category were excluded. One exercise session was defined as at least 40 min of moderate to high intensity activity. Participants were also asked to record the typical intensity, frequency, and duration of each activity per week. These criteria were used as identical to previous studies showing differences in appetite control in active versus inactive individuals [18,26]. We have previously shown these categories to differ in

objectively measured physical activity [27]. Sample size calculations were conducted in G\*Power [28] using data from a recent study assessing sweet intensity in adults [29], as taste intensity was the primary outcome measure, differences with physical activity in females were previously shown in relation to sweet taste [14], and the most similar literature that provided quantitative data to allow sample size calculation related to sweet taste. To detect a 10-point difference in sweet taste intensity ratings between the two groups in the present study with a power of 80% and a significance level of 5%, 14 individuals were required per group.

#### *2.3. Recruitment and Setting*

Recruitment was conducted through the distribution of recruitment flyers and emails throughout the university campus. Participants who were eligible to participate based on information provided in the screening questionnaire were invited to participate in the testing sessions. The testing sessions took place in the sensory evaluation suite at UCD's Institute of Food and Health. Participants were recruited from the 1 January 2018 until the 15 March 2018.

#### *2.4. Body Composition Measurement and Taste Perception Assessment Day Protocol*

On arrival, all participants provided written informed consent. On both test days, participants attended the laboratory in the morning, having avoided strong-flavoured foods/drinks, such as spicy foods and coffee, for 12 h and strenuous exercise for 24 h, and were instructed to wear light clothing for body composition measurements. Participants' height measurements were first taken using a stadiometer, followed by weight and body composition using a Tanita body composition analyser (BC-420MA, Tanita Ltd., Yiewsley, UK), which uses bioelectrical impedance (BIA) to assess body composition. Participants were then familiarised with the generalised labelled magnitude scale to assess perceived intensity (gLMS) [30]. The gLMS is a validated scale for assessing taste intensity, according to the standard protocol outlined previously by Green, Schaffer, and Gilmore [31] and Green et al. [32]. A generalised degree of liking scale (gDOL) with labels of 'neutral' and 'strongest liking/disliking of any kind' was used to assess liking of the stimuli.

#### 2.4.1. Taste Stimuli

Food-grade, prototypical taste stimuli in water were prepared as follows: Sucrose (sweet) (27 mmol/L and 243 mmol/L), citric acid (sour) (1 mmol/L and 9 mmol/L), sodium chloride (salt) (33 mmol/L and 300 mmol/L), quinine (bitter) (0.056 mmol/L and 0.498 mmol/L), monosodium glutamate; MSG (umami) (0.51 g/L and 4.566 g/L) [24], and maltodextrin (carbohydrate) (dextrose equivalent 4.0–7.0, Sigma Aldrich, Arklow, Ireland) (35.5 g/L and 112.4 g/L) [2]. These concentrations were selected to provide 'low' and 'high' concentrations to allow comparison to recent research [2,24].

#### 2.4.2. Taste Perception Rating

Participants undertook two identical taste sessions spaced one week apart. Previous work has indicated that at least two taste intensity ratings are necessary to achieve reliable estimates of individual taste responsiveness when using the gLMS [33]. Both sessions took place in the mornings, and participants followed identical instructions prior to each visit. The taste stimuli were the same in each session and mean results from the two ratings for each concentration of each taste were the primary outcome used in analyses. Results from the individual test days were also explored to examine the reliability of results at the individual test session. At each session, participants tasted 12 samples (high and low concentrations of the six tastes), served at room temperature in 20 mL medicine cups in a sip-and-spit manner in a randomised block design, presented blinded, using 3-digit randomised codes. The tests were administered on computers located in each sensory booth, using RedJade software (RedJade Software Solutions, LLC, Boulder, CO, USA). Participants were requested to identify the taste from a specified list ('sweet', 'salt', 'sour', 'bitter', 'umami', 'carbohydrate' or 'unsure') and then to rate the perceived intensity and then liking of the stimuli on a gLMS and gDOL respectively presented on

screen. A 30 s break was enforced with the software between each solution, with a 2-min break after every 4 samples, with water for rinsing provided between each sample.

#### 2.4.3. Eating Behaviour and Habitual Diet

At the end of the taste protocol on the second test day, participants completed a paper-based version of the three factor eating questionnaire (TFEQ) [34] and European Prospective Investigation into Cancer and Nutrition (EPIC) food frequency questionnaire (FFQ) [35].

#### *2.5. Statistical Analysis*

Statistical analysis was performed using PASW Statistics 24.0 (SPSS Inc., Chicago, IL, USA) and GraphPad Prism version 7.0 for Mac (GraphPad Software, San Diego, CA, USA). To determine if the data of the two groups was normally distributed, the Shapiro–Wilk test was used. For normally distributed data, a parametric independent *t*-test was used; otherwise, the Mann–Whitney U test was used. Pearson's correlation or Spearman's rank correlation coefficient were used to determine relationships between variables where appropriate. Effect size (ES: Cohen's d or r where appropriate) was also assessed. Multiple regression analysis was undertaken to identify the effects of confounding variables such as age and body composition on taste perception. There were no missing data for any outcomes, except for the TFEQ and FFQ for two participants in the active group. Only complete data for the TFEQ was used in analyses (*n* = 14 in both groups). Statistical significance was considered at *p* < 0.05 and data are reported as mean (SD) unless otherwise stated.

#### **3. Results**

#### *3.1. Subject Characteristics*

Descriptive data for the active compared to inactive groups are shown in Table 1. Age, height, weight and body mass index (BMI) did not significantly differ between the two groups; however, the active group had a lower body fat percentage.


**Table 1.** Subject characteristics of active (*n* = 16) and inactive (*n* = 14) groups.

<sup>1</sup> As data were not normally distributed, *r* is used for effect size.

#### *3.2. Taste Identification*

Although the study was not designed to assess identification of tastes, we explored whether it differed between the two visits and two groups, as it could potentially influence the results. Overall, taste identification did not differ significantly for any taste and concentration between the first and second session, suggesting no learning effect for taste identification occurred. Mean percentage of tastes correctly identified was greater at the high concentrations than at the lower concentrations for both groups (*p* < 0.05) but did not differ between the two groups (*p* > 0.05). When the individual tastes were compared at the two visits, a greater percentage of the active group correctly identified the umami taste compared to the inactive group (*p* = 0.03). However, there were no significant differences in the identification of all tastes between the two groups for all other tastes and concentrations at the two visits.

#### *3.3. Taste Intensity*

Perceived intensity in the active compared to inactive groups for the six tastes studied are shown in Figure 1. There was a large difference (ES: *d* = 1.63, *p* < 0.05) in perceived intensity for the high-concentration sweet (sucrose) taste between the two groups, with the active group recording a significantly higher intensity rating compared to the inactive group (Figure 1A). Significantly higher intensity ratings were also observed in the active group for the umami (MSG) taste (Figure 1C), with a large difference between the two groups for the low concentration (ES: *d* = 1.18, *p* < 0.01) and moderate difference at the high concentration (ES: *r* = 0.38, *p* < 0.05). Perceived intensity did not significantly differ between the two groups for the low concentration of sucrose, nor for either high or low concentrations of citric acid, quinine, sodium chloride, and maltodextrin (*p* > 0.05 for all; Figure 1). For the latter comparisons, effect sizes were small for all (*d* < 0.30), except for the low concentration of quinine (*d* = 0.60) and low concentration of maltodextrin (*d* = 0.66).

**Figure 1.** Differences (mean (SE)) in perceived intensity responses on a generalised labelled magnitude scale (gLMS) for high and low concentrations of: (**A**) Sweet, (**B**) sour, (**C**) umami, (**D**) bitter, (**E**) salt, and (**F**) carbohydrate (maltodextrin) taste intensity ratings with physical activity. Solid dark line indicates active group, dashed line indicates inactive group. \* *p* < 0.05, \*\* *p* < 0.01. MSG, monosodium glutamate.

#### *3.4. Hedonic Response*

Hedonic ratings in the active compared to inactive group for the six tastes studied are shown in Table 2. Apart from the sweet taste, the majority of tastes had negative ratings, indicating varying levels of dislike on the gDOL. There was a large difference between the two groups in liking of the umami low concentration with the active group recording a dislike of the taste, compared to a mean response of a weak liking in the inactive group. Similarly, the active group recorded a dislike for the low-concentration carbohydrate taste, compared to a mean response of a weak liking in the inactive group (Table 2). Although there were no statistically significant differences in hedonic ratings between the two groups for the other solutions, moderate effect sizes were observed for the bitter taste, with a trend towards a greater dislike of the bitter taste in active individuals.



<sup>1</sup> The labels of the scale were 'neutral' and 'strongest liking/disliking of any kind'. <sup>2</sup> MSG, monosodium glutamate.

#### *3.5. Reproducibility of Taste Intensity Comparisons between Groups at Individual Test Days*

Interestingly, carbohydrate perceived intensity was higher at the first visit in the active group (*p* < 0.05) but did not differ at the second visit and therefore was not significantly different when mean ratings were compared. By contrast, umami high concentration intensity ratings did not significantly differ statistically between groups at the individual test days but differed when mean ratings were compared. However, most differences that were observed in mean ratings were also evident at both individual test days, with significant differences or similar trends observed (*p* < 0.1) for high-concentration sweet and low-concentration umami. Moreover, similar to mean ratings, perceived intensity for low-concentration sweet, and both concentrations of sour, bitter, and salty were not different between the two groups at the separate test visits.

#### *3.6. Habitual Dietary Intake*

There was no difference observed in the percentage of energy from macronutrients, or carbohydrate between the active and inactive group (Table 3), except for fructose, which was higher in the inactive group (*p* = 0.04), and fibre, which trended towards being higher in the inactive group (*p* = 0.05).


**Table 3.** Mean energy intake and percentage of energy from macronutrients for active and inactive men (FFQ data).

<sup>1</sup> FFQ (food frequency questionnaire) data were available for *n* = 26 individuals (*n* = 13 per group). Data were missing for two participants and data for two individuals were removed due to energy misreporting (energy intake >2 SD above or below the mean energy intake were removed as per Low et al. [2].

#### *3.7. Regression Analysis Including Age, BMI, and Body Composition*

#### 3.7.1. Taste Intensity

For the sweet taste, physical activity status was the only variable associated with perceived intensity at the high concentration (model adjusted *<sup>R</sup>*2: 0.13; *<sup>ß</sup>* <sup>=</sup> −0.39, *<sup>p</sup>* = 0.03). Age, BMI or percentage of body fat were not independently associated with sweet taste intensity or when included in models for either concentration (*p* > 0.1 for all).

For the umami taste at both concentrations, physical activity status showed the strongest association with perceived intensity. BMI, body fat or age were not associated with differences in umami perceived intensity (*p* > 0.1).

For the sour taste, at the low concentration, both age and body composition were significantly associated with perceived intensity (*p* < 0.01) but not independently in the same model. Moreover, there were no associations between sour taste at the high concentration and age, BMI, percentage of body fat or physical activity status. Bitter and salty tastes also showed no significant associations with these variables at either the high or low concentrations (*p* > 0.1 for all). For the carbohydrate taste, BMI and activity status together in the same model significantly predicted perceived intensity at the low concentration (model adjusted *<sup>R</sup>*2: 0.14; *<sup>p</sup>* < 0.05; activity status *<sup>ß</sup>* <sup>=</sup> −0.30, *<sup>p</sup>* = 0.099; BMI *<sup>ß</sup>* <sup>=</sup> −0.31, *p* = 0.08).

#### 3.7.2. Liking

Sweet and sour taste liking were not associated with physical activity status, age or body composition (*p* > 0.1). However, activity status was associated with liking of the low-concentration umami (*ß* = 0.44, *p* = 0.02), high-concentration bitter (*ß* = 0.37, *p* < 0.05), and low-concentration carbohydrate (*ß* = 0.43, *p* = 0.02) tastes. Liking of the low-concentration salty taste (*ß* = 0.45, *p* = 0.01) and of the high-concentration umami taste (*ß* = 0.45, *p* = 0.01) were both associated with age. Liking of the low-concentration carbohydrate solution was the only variable associated with percentage of body fat (*ß* = 0.45, *p* = 0.01) and, together with activity status in the same model, was associated with 20% of the variance in liking for carbohydrate at the low concentration (model adjusted *R*2: 0.20; *p* < 0.05; activity status *ß* = 0.27, *p* = 0.18; body fat *ß* = 0.31, *p* = 0.13).

#### *3.8. Regression Analysis with Eating Behaviour*

#### 3.8.1. Hunger

Sweet taste perceived intensity for the high concentration was positively associated with the trait hunger (*ß* = 0.39, *p* = 0.04), (i.e., a higher hunger score was associated with greater perceived intensity). This remained significant when included in the same model as activity status. Together, activity status and hunger were associated with 27% of the variance in perceived intensity for sweet taste at the high concentration (model adjusted *R*2: 0.27; *p* < 0.01). In addition, hunger was associated with perceived intensity for the high-concentration bitter taste (*ß* = 0.44, *p* = 0.02) and liking for the high-concentration salt taste (*ß* = 0.43, *p* = 0.02).

#### 3.8.2. Disinhibition

Disinhibition was also associated with liking of the high-concentration salt taste (*ß* = 0.49, *p* < 0.01), but not with any other variables.

#### 3.8.3. Restraint

Perceived intensity for the high-concentration bitter taste (*ß* = −0.41, *p* = 0.03), and liking of the low- (*ß* = −0.46, *p* = 0.02) and high- (*ß* = −0.43, *p* = 0.02) concentration umami taste were associated with dietary restraint. As activity status (active or inactive) was also significantly associated with liking for umami low concentration (model adjusted *R*<sup>2</sup> = 0.19, *ß* = 0.439, *p* < 0.015), they were included in the same model. Together, activity status and restraint accounted for 38% of the variance in umami low concentration liking (model adjusted *<sup>R</sup>*<sup>2</sup> <sup>0</sup>·38, *<sup>p</sup>* < 0·001; activity: *<sup>ß</sup>* = 0·46, *<sup>p</sup>* < 0·01; restraint: *<sup>ß</sup>* <sup>=</sup> −0.47, *p* < 0·001). Dietary restraint was also inversely associated with liking of the high-concentration carbohydrate taste (*ß* = −0.41, *p* = 0.03).

#### **4. Discussion**

The present findings demonstrate that taste perception intensity differed between active and inactive males. In this cohort, active males reported a greater perceived intensity for both sweet and umami tastes. Given previous evidence of associations between taste perception and food choice [3,4,21], and weight gain [24], these findings may have implications for understanding factors influencing the control of food intake and energy balance with habitual exercise.

Although limited research has investigated associations of habitual exercise with taste perception, our finding of a greater perceived intensity of sweet taste in active males is comparable to a previous study in females. Crystal, Frye, and Kanarek [14] found female swimmers perceived a high-concentration sucrose solution as sweeter compared to inactive females' using visual analogue scales (VAS). In response to acute exercise, increases, no change, and decreases in acuity of taste and rated preference for tastes have been previously reported, with results appearing to depend on differences in length and intensity of the exercise session and the taste [13]. Regarding sweet taste specifically, Westerterp-Plantenga et al. [36] observed an increase in perceived intensity of taste using VAS for a low-concentration sucrose solution (but not high-concentration) following 2 h of moderate intensity cycling. By contrast, others have reported no change in sweet taste intensity for a sucrose solution, but an increase in intensity of sour taste following 10 minutes of cycling to generate a 'light sweat' [37]. However, assessing differences with longer-term interventions and with habitual exercise is also essential, as the repeated effects of regular exercise on physiological and psychological processes of appetite control do not always mimic the acute effects of exercise. Participants in the present study were instructed to avoid strenuous exercise for 24 h before the test sessions to avoid influence of acute exercise on results.

Perceived intensity of umami taste also differed between the active and inactive groups. Perceived intensity of both low and high concentrations of MSG was rated as significantly higher in active males. In a previous study, along with the other 'basic' tastes, Horio and Kawamura [38] assessed umami

threshold and liking using six different concentration solutions of MSG after moderate-intensity cycling and found no difference compared to pre-exercise in healthy university students. Generally, the effects of both acute and chronic exercise on umami taste perception, however, have not been extensively studied previously.

Several factors could contribute to the differences in sweet and umami taste perception we observed with habitual exercise. Some previous studies have shown habitual diet to be associated with taste perception. For example, sweet taste intensity has been shown to negatively correlate with total energy and carbohydrate and sweet food intake [11], although this was not demonstrated elsewhere [39]. By contrast, higher carbohydrate taste intensity has been positively associated with greater energy and starch intakes, assessed by either FFQ or food diary [2]. In the present study, however, we did not observe differences in the percentage of energy consumed from starch, sugar or other carbohydrate forms, apart from fructose, which was higher in the inactive group, although the limitations of FFQ are recognised.

Eating behaviour traits could also contribute to differences in taste perception between individuals. Dietary restraint and disinhibition have been identified as factors that may influence relationships between adiposity and taste sensitivities to 6-n-propylthiouracil [40]. Therefore, we investigated whether eating behaviours could influence associations between habitual exercise and taste perception. The trait hunger and physical activity were both independently associated with sweet taste intensity perception at the high concentration in the same model. Hunger and restraint were both also associated with perceived intensity of the high-concentration bitter taste (quinine); however, eating behaviours were not related to other perceived intensity ratings. These findings suggest eating behaviour traits may be linked to taste perception of some tastes; however, the traits hunger, restraint or disinhibition do not explain the differences in taste perception observed with habitual exercise.

Although we did not assess hormonal status, hormonal differences could be another mechanism contributing to the differences in taste perception we observed in active compared to inactive males. It is interesting that perception of sweet and umami were the two tastes that differed between the active and inactive groups, indicating the differences in perception were specific mainly to these tastes and not an overall effect on taste function. Both sweet and umami share the same class of taste receptor in the mouth, which initiate a G protein-coupled signalling cascade [10,41,42]. There is now strong evidence that in addition to intestinal signalling, glucagon-like peptide-1 (GLP-1) signalling also occurs within the taste bud [43], and evidence from animal studies indicates GLP-1 signalling has an important role in the modulation of both sweet and umami taste [42]. In the present study, a sip-and-spit technique was used for the tasting of solutions, suggesting any differences in the taste responses observed are influenced by orososensory mechanisms and not a post-oral response to nutrients. Other hormones modified by regular exercise or body composition could also potentially contribute to the differences we observed (see Reference [10] for a review). For example, leptin can inhibit the response to sweet taste [10,44]. An increased circulating leptin (due to a greater body fat percentage) in less active individuals could be one potential mechanism contributing to a reduced intensity of sweet taste. However, BMI or percentage body fat did not moderate the relationship between activity status and sweet taste perception in the present study. Endocannabinoids [45] and glucose levels [46] also influence sweet taste responses and are altered with physical activity [47]. Characterising multiple hormonal factors is warranted in future studies, as it may provide mechanistic insight into differences in taste perception with habitual exercise.

Regarding liking, sweet was the only taste to be positively rated by both groups, adding support that sweet tastes are liked by most individuals [48]. We also observed that in addition to a greater perceived intensity in the active group, active males also had a lower (negative) hedonic rating for the low-concentration umami taste, suggesting the greater perceived intensity could have contributed to a dislike for the taste. In contrast, liking of the sweet and umami high concentration tastes did not differ between the active and inactive groups, despite differences in perceived intensity. Others have also observed no difference in hedonic rating, despite changes in taste perception of

simple solutions, including sucrose and quinine sulphate, with acute exercise [36]. One explanation could be the form in which the taste stimuli are provided. For standardisation, all samples here were provided to participants in solutions made up with water. However, when compared to more complex food and beverage matrices, possible differences are likely [49]. The implications of differences in perceived taste intensity with habitual exercise for liking of different foods and food choice warrant further investigation.

Various methodological aspects of the present study should be considered. Participants included males only to eliminate the possible interference of the menstrual cycle on taste perception, and to add to previous research in females [14]. Further research in females is warranted. As body composition has a role in some hedonic aspects of appetite control [15,20], body composition (fat and fat free mass) should also be further considered. Links between taste and body weight or BMI have been previously studied [8,22,23]; however, fewer studies [50] have examined fat or fat free mass. In the present study, BMI did not differ between active and inactive groups, while percent body fat differed significantly. These and other potential modifying variables were included in regression models but did not modify the relationship between physical activity status and taste perception. For logistical reasons, body composition was assessed using BIA. Relationships between taste and body composition have previously been reported using BIA [51]. However, studies have shown variable findings regarding the accuracy of BIA [52,53]. Future studies should explore relationships between taste perception and body composition (fat and free mass) on an individual level using more accurate measures. Measurement of waist circumference would also be relevant. In addition, objective measurement of physical activity should be considered in future studies, to further elucidate relationships with aspects such as energy expenditure and sedentary behaviour. Finally, a significant strength of the study is the use of six tastes, low and high concentrations and two taste sessions, allowing a comprehensive assessment of potential differences in taste perception with habitual exercise. Therefore, the findings can be considered to provide reliable estimates of individual taste responsiveness in active versus inactive males.

In conclusion, these data show sweet and umami taste perception differ in habitual exercisers compared to inactive individuals. There is evidence elsewhere that habitually active individuals have improved energy compensation for energy density of foods [19]. Alterations in taste perception could be one potential mechanism contributing to the regulation of energy balance with exercise. While causal inferences cannot be drawn due to the cross-sectional nature of this study, the findings have implications for researchers and for product development—indicating habitual exercise level should be considered in studies examining taste perception and for consumer selection for product development. Further studies are needed to examine longitudinal responses to exercise intervention and to further explore the underlying mechanisms and implications for food intake.

**Author Contributions:** Conceptualisation, K.H., E.L.F., L.L., M.O., N.L., A.P., L.C.; formal analysis, K.H., E.L.F.; investigation, L.L., M.O., N.L., A.P., L.C., A.d.C., E.S.; writing—original draft preparation, L.L., M.O., N.L., A.P., L.C., K.H., E.L.F.; writing—review and editing, K.H., E.L.F., supervision, K.H., E.L.F., A.d.C., project administration: L.L., M.O., N.L., A.P., L.C., E.S., A.d.C.

**Funding:** This research received no external funding.

**Acknowledgments:** We are grateful to all of the participants in this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
