Distribution of Recognition Times to Fruity Flavor of Gummy Candies in Healthy Adults

Introduction

A psychological study revealed an important property concerning the shape of distributions of group visual reaction times: that is, exponential modified Gaussian (ex-Gaussian) distribution, which is the convolution of normal and exponential distributions (Ratcliff, 1979). The ex-Gaussian distribution allows us to consider group reaction time data by two different, normal and exponential (for outliers), distributions (Heathcote, Popiel, & Mewhort, 1991; Ratcliff, 1993). A lot of studies using the ex-Gaussian analysis has showed not only theoretical but also applicable findings (Buzy, Medoff, & Schweitzer, 2009; Feige et al., 2013; Leth-Steensen, Elbaz, & Douglas, 2000; Payne & Stine-Morrow, 2014; Roelofs, 2008; Schmiedek, Oberauer, Wilhelm, Suss, & Wittmann, 2007; White, Risko, & Besner, 2016). Some of the findings are involved in reading times (Payne & Stine-Morrow, 2014), working memory, reasoning, psychometric speed (Schmiedek et al., 2007), and applicable findings are often obtained from children with attention deficit hyperactivity disorder (Buzy et al., 2009; Feige et al., 2013; Leth-Steensen et al., 2000).

By contrast, few studies in the field of chemical senses research have analyzed the distributional properties of reaction times, particularly of reaction times to flavors. Furthermore, reaction times to gustatory stimuli have mostly been determined using simple fluids such as sucrose and sodium chloride solutions (Bonnet, Zamora, Buratti, Guirao, 1999; Bujas, Szabo, Ajdukovic, & Mayer, 1989; Reiter, Campillo Rodriguez, Sun, & Stopfer, 2015; Yamamoto & Kawamura, 1981). These studies showed very precise results, but they are far from our experiences in everyday life, because people rarely consume foods or fluids composed solely of sucrose and sodium chloride. Despite this, few previous studies have examined reaction times to real-world foods. We developed an electromyography-based measuring system for flavor reaction times (detection and recognition times) with semi-solid foods (Miyaoka & Miyaoka, 2013), whereas the term, flavor, was defined as a sensation resulting from interactions of taste, smell (aroma), and trigeminal sensations (Prescott, 1999; Small & Prescott, 2005). We observed that the overall average detection time was around 4 seconds in the used fruity gummy candies, and we analyzed gustatory factors responsible for the detection time (Miyaoka, Yamazaki, Ito, & Miyaoka, 2014; Miyaoka, Miyaoka, & Ashida, in press). However, we have not fully analyzed recognition times including their distributional property.

Thus, the major objective of this study was to examine the distributional property of recognition time of flavors. We expect that the results presented herein will have far-reaching implications for the food industry, just as with visual research.

Materials and Methods

Results

A total of 181 recognition times was recorded from the 23 participants of the male (113 records) and female (68 records) groups. Since Yate’s continuity correction test did not detect significant differences in recognition times between them, data of the two groups were combined. The combined recognition times widely distributed ranging from 2.6 seconds to 46.8 seconds (Figure 1), and goodness-of-fit test rejected the normality of the whole frequency distribution. Repeated application of the test showed that the 107 recognition times not longer than 12.32 seconds (Zone I in Figure 1) distributed normally, so the remaining 74 recognition times longer than 12.4 seconds (Zone II in Figure 1) distributed differently from Zone I. OPEN IN VIEWER

Recognition times in these two zones varied among the six gummy candies (Table 1): About 70% of the recognition times for the orange-flavored gummy candy belonged to the Zone I, while only about 35% of them for the pear-flavored. Chi-square test of the contingency table and the following multiple comparison test revealed that (a) the numbers of recognition times in the six gummy candies significantly differed between the two zones (p < .05) and (b) recognition times of the pear-flavored gummy candy were less in the Zone I (p < .05) and more in the Zone II than the other five gummy candies (p < .05; Table 1). The overall average of 107 recognition times in the Zone I for the six gummy candies was 7.5 seconds (standard deviation: 2.34 seconds), and the individual averages ranged from 6.8 seconds (for pineapple-flavored) to 8.1 seconds (for strawberry-flavored; Figure 2). One-way ANOVA detected no significant differences in the average among the six gummy candies.

OPEN IN VIEWER

Table 1.

Percentages of Recognition Times in Zones I and II.

Flavors of gummy candies	Apple	Grape	Orange	Pear	Pineapple	Strawberry	Total
Zones I and IIa	31	43	33	26 c	17	31	181
Zone Ib (<12.34 seconds)	15	30	24	9 d	11	18	107
b/a × 100 (%)	48.4	69.8	72.7	34.6	64.7	58.1	59.1

OPEN IN VIEWER

Figure 2.

The average recognition times for six fruity flavored gummy candies. The values shown were collected only from the Zone I (see Figure 1). The values present the means ± standard errors of means. There are no significant differences in the average among the six fruity flavored gummy candies.

Eighteen of the 23 participants (78.3%) answered 41 (22.7%) inconsistent reports, which did not coincide with the flavors that were provided by the manufacturer. The most frequently observed report was an apple-flavored gummy candy (14 of the 41 reports, 34.1%) mainly for a pear-flavored (8 of the 13 reports, 61.5%). The report was followed by peach-flavored (6 reports, 14.6%) mainly for grape-flavored (3 reports, 50.0%) and lemon-flavored (5 reports, 9.8%) mainly for orange-flavored (4 reports, 80.0%) gummy candies, respectively. However, chi-square test of the contingency table showed that there was no significant correlation between the numbers of recognition times and those of inconsistent reports among the used six flavored gummy candies.

Discussion

There were two, I and II, zones for the 181 flavor recognition times measured in this study (Figure 1, Table 1) and the two zones suggested an ex-Gaussian distribution, which consisted of normal and exponential distributions. Actually, the statistical analysis indicated the normality of the recognition times in the Zone I. It is difficult for us to describe mathematically the precise distributional property of the recognition times in the Zone II, which we can state just distributed non-normally due to the limits of our statistical abilities. Although the distributional property shown in this study is needed to be confirmed precisely with some suitable statistical methods, it must be the first report in research field of chemical senses. Distributional analysis may be useful not only in visual and cognitive research fields (Buzy et al., 2009; Leth-Steensen et al., 2000; Payne & Stine-Morrow, 2014; Roelofs, 2008; Schmiedek et al., 2007; White et al., 2016) but also in chemical senses and food science research fields. In visual and cognitive research fields, the congruence of stimuli, for instance, was evaluated by using distributional analysis of Stroop task data (Heathcote et al., 1991; Mewhort, Braun, & Heathcote, 1992). Recent studies in chemical senses applied a model of probability summation to simple reaction time data (Shepard, Veldhuizen, & Marks, 2015; Veldhuizen, Shepard, Wang, & Marks, 2010). In one of the studies (Shepard et al., 2015), the authors used citral, monosodium glutamate, and their mixtures as test stimuli and concluded that ‘gustatory and olfactory flavorants integrated their effects in producing rapid responses, but only if, or especially when, the flavorants are congruent.’ Distributional analysis may be a novel approach for evaluation of the congruence of stimuli in chemical senses.

We reported detection times for fruity flavors which were emitted by chewing of test foods (gummy candies) in the mouth, and the average detection times ranged around 4 seconds (Miyaoka et al., 2014), which were much longer than those measured in fluids as test stimuli (e.g., Veldhuizen et al., 2010). Differences in attributes of test stimuli (fluids and solid foods) between these studies are likely to be responsible for the large differences in measured detection times. Recently, we showed that the hardness of semi-solid foods strongly affected detection times: that is, the overall average detection and recognition times across the used flavors were 2.1 seconds and 4.6 seconds for the fruity jellies (average hardness was about 170 kPa/m²), while those were 4.0 seconds and 7.5 seconds (Figure 2) for the gummy candies (average hardness was about 800 kPa/m² (Miyaoka et al., 2014; Miyaoka et al., 2018). The present result clearly suggested that the hardness of test foods affects not only detection but also recognition time.

Our previous study documented the overall odor intensity of the test foods and detected differences in average intensity (Miyaoka et al., 2014). However, supplementary statistical analyses revealed that odor intensity did not significantly correlate with the following three parameters: (a) average recognition time (Figure 2), (b) percentages of recognition times in either of the two zones (Table 1), or (c) the rate of inconsistent reports (Table 2). In addition, chemical (gustatory) components differed significantly among the gummy candies (see ‘Test Foods’ section), but supplementary analyses showed that these differences also did not correlate with the three aforementioned parameters. In summary, these additional analyses indicated that odor intensity and the chemical (gustatory) components of the test foods did not significantly affect the results of the present study.

OPEN IN VIEWER

Table 2.

Inconsistency of Flavors Between the Participants’ Reports and the Manufacturer’s Description of the Products.

Flavors of gummy candies	Apple	Grape	Orange	Pear	Pineapple	Strawberry	Total
Recognition timesa	31	43	33	26	17	31	181
Inconsistent reportsb	4 (0)	8 (1)	11 (0)	10 (2)	2 (0)	6 (0)	41 (3)
b/a × 100 (%)	12.9	18.6	33.3	38.5	17.6	19.4	22.7

There was considerable interindividual variability among the recognition times (Figure 2). A previous study investigated interindividual variability in aroma release from milk gels of different hardness using hierarchical cluster analysis (Gierczynski, Laboure, & Guichard, 2008). The authors found that the aroma of firmer gels was perceived as more intense and also that their 14 panelists could be separated into two groups. Aroma-release profiles differed greatly between the two groups of panelists, and the authors suggested that the opening of the velum–tongue barrier during chewing causes this difference. However, it is difficult for us to judge whether or not physiological factors are also responsible for the interindividual variability observed in recognition times in the present study, because the differences in hardness of the gummy candies used here are likely much less pronounced than the differences in the foods used in the previous study (Gierczynski et al., 2008).

There were two unexpected results for us in this study: one was extremely long recognition times in the Zone II (Figure 1) and the other was the fact that 22.7% of the reports by more than the 23 participants were not consistent with the flavors that were provided with the manufacturer. The present participants were not informed the flavors of gummy candies before each trial, and this situation was likely to be related the unexpected result. A previous study indicated that the average scores of correct responses for six fruity odors in a cued identification task (participants were informed names of the odors) were much higher than those in a free identification (de Wijk & Cain, 1994). Just like this study, we did not use the real-world substances as test foods, but we used commercially available gummy candies. Thus, the recognition time task in the present study had no correct responses, and we could merely state that individual recognition reports were or were not consistent with the flavors provided with the manufacturer. A study in food science examined 18 apple cultivars by using sensory (quantitative descriptive analysis) and instrumental (gas chromatography/mass spectrometry) methods to characterize perceived quality and its relation to odor and flavor (Aprea et al., 2012). This study identified 72 volatile compounds and predicted 9 odors including apple and pear. The authors suggest a possibility that an interaction of the same volatile compounds at different concentration can produce an alternative perceived odor/flavor: for example, mixing a different ratio of acetate esters contributes to the development of apple and pear odors. The possibility may relate to the present result that there were a lot of inconsistent responses (n = 13) to apple (by the manufacturer) and pear (by the participants) flavors.

In conclusion, there were two groups of the 181 recognition times measured by an electromyography-based system, I (distributed normally) and II (distributed non-normally), for six fruit flavored gummy candies in 23 young healthy participants. The overall average recognition time in Zone I was 7.5 seconds (n = 107 responses), while the ratios of the six averages did not differ among the six flavored gummy candies. Forty-one (22.7%) of the total 181 recognition reports by 18 of the 23 participants were not consistent with the flavors that were provided with the manufacturer.