**Sensory Acceptability of Dual-Fortified Milled Red and Yellow Lentil (***Lens culinaris* **Medik.) Dal in Bangladesh**

**Rajib Podder 1, \*, Mahmudul Hassan Al Imam 2,3,4 , Israt Jahan 2,3,4 , Fakir Md Yunus 5 , Mohammad Muhit 2,3 and Albert Vandenberg 1**


Received: 20 June 2020; Accepted: 22 July 2020; Published: 24 July 2020

**Abstract:** This study evaluated the sensory properties of uncooked and cooked milled lentils that were fortified with varying concentrations of Fe and Zn in the form of NaFeEDTA and ZnSO4.H2O, respectively. Our study was carried out among 196 lentil consumers residing in rural Bangladesh who experience with growing, processing, and marketing lentils. A nine-point hedonic scale was used to rate the appearance, odor, taste, texture and overall acceptability of three uncooked and two cooked lentil (dal) samples made from each of the three milled lentil product types (LPTs), red football, red split and yellow split. Preferences for sensory properties were found to be significantly different among all uncooked lentil samples, but not significantly different for cooked samples, with a few exceptions. This means that the fortification process minimally affects dual-fortified lentil sample (fortified with 16 mg of Fe and 8 mg of Zn per 100 g of lentil), which was compared to another cooked sample (unfortified control), in terms of consumers liking for all four attributes (appearance, odor, taste, and texture).

**Keywords:** dual fortification; sensory evaluation; iron and zinc deficiency; lentil

#### **1. Introduction**

Iron (Fe) and zinc (Zn) micronutrient deficiencies are two of the most prevalent nutritional threats in the world. About one third and one fifth of the human population are Fe and Zn deficient, respectively [1]. These two micronutrients share common dietary sources and are abundant in the human body [2]. Plant-based diets are becoming popular throughout the world, and legumes such as lentils, chickpeas, dry peas, beans, and fava beans are major dietary sources of protein. Among the legumes, lentils are important for human nutrition because of their relatively high amounts of protein, carbohydrates, and micronutrients compared to some of the staple cereals and root crops [3,4]. More than 50 countries in aggregate produce a global total of about 7.6 Mt of lentils, of which Canada produces about 50% (3.7 Mt) [5]. Lentils contain a substantial amount (dry weight) of protein (25.8 to 27.1%), starch (27.4 to 47.1%), dietary fiber (5.1 to 26.6%) [6–8], Fe (73 to 90 mg kg −1 ), Zn (44 to 54 mg kg −1 ), and selenium (425 to 673 µg kg −1 ) [9]. A combination of rice and lentils makes a popular

and commonly eaten dish known as "hotchpotch" in many Asian countries, for example, in Bangladesh. This dish provides all essential amino acids, carbohydrates, dietary fiber, and a number of minerals and vitamins. Although lentil has a significant amount of intrinsic Fe and Zn, some antinutritional factors, such as phytate, polyphenols, calcium, and protein can inhibit the absorption of both nutrients from food [9]. The improvement of the concentration of these micronutrients and their bioavailability using a sustainable approach is a prime area for research in order to provide an adequate amount of micronutrients and cope with micronutrient deficiency.

Several organizations are conducting research to improve the micronutrient concentration in crop or food products to cope with global micronutrient deficiency problems. Many approaches are used, including biofortification, food fortification, public health intervention, supplementation, nutrition education, dietary diversification, and food safety measures. These strategies are being employed for various staple crops or foods around the world [10]. In comparison to other approaches, food fortification is now more widely used due to its sustainability for improving the dietary quality of targeted groups or populations rapidly [10–12]. Around 84 countries have mandatory fortification programs for various food products based on their existing nutritional status [13]. Several micronutrient-fortified foods/food products are available and are mandatory in the market in different countries, for example, wheat flour in Indonesia, Philippines, Nepal, fortified rice in Papua New Guinea and Costa Rica, maize flour in the USA, soya sauce, salt and edible oil in Bangladesh, milk in Canada and China, etc. [13]. The fortification of pulse crops like lentils or chickpeas is a new research area that began in 2014 at the Crop Development Centre of the University of Saskatchewan, Canada, through the development of Fe-fortified lentils to address Fe-deficiency in humans. A laboratory-scale protocol for fortifying dehulled red lentils with the Fe fortificant NaFeEDTA (ethylenediaminetetraacetic acid iron (iii) sodium salt) was developed [14]. Fortification with 1600 ppm, NaFeEDTA provides 13–14 mg of additional Fe 100−<sup>1</sup> g in cooked dehulled lentils (dal). An in vitro bioavailability study with Fe-fortified lentils showed that dehulled lentil dal fortified with 28 mg of Fe 100−<sup>1</sup> g of lentils increased Fe bioavailability to 79% and reduced phytic acid to 25% [15]. The results from these studies led us to develop dual-fortified lentils with Fe and Zn to address Fe and Zn deficiency.

Lentil fortification with both Fe and Zn could have the potential to simultaneously reduce both Fe and Zn deficiency. In this approach, lentils are enriched with extra Fe and Zn to prevent iron deficiency in humans. In this project, research has been initiated to increase both Fe and Zn concentration and bioavailability through a fortification strategy using a modified technique of a previously developed fortification technique by Podder et al. (2017). Initially, a laboratory-based fortification protocol to develop dual-fortified lentil was established. The protocol included the selection of three lentil product types (LPTs) (dehulled red football (RF), red split (RS), and yellow split (YS)), the identification of appropriate methods of fortification, the selection of suitable dosage of added Fe and Zn, and colorimetric changes over the storage period, as well as the in vitro bioavailability of Fe from the dual-fortified lentils [16]. This report describes the results of a sensory analysis of dual-fortified lentil food products.

Sensory analysis is a multidisciplinary science that covers a wide range of social science areas, ranging from food science to statistics to psychology [17]. By definition, "sensory analysis is the identification, scientific measurement and interpretation of the properties (attributes) of a product as they are perceived through the five senses of sight, smell, taste, touch, and hearing" [18]. It captures unbiased human response to food, which helps stakeholders to identify brand effects [19]. Taste, flavor, appearance, and texture are the major attributes of sensory evaluations of food products. The remarks from consumers provide valuable information that help in the development of recommendations for food scientists or commercial food product developers. The present study was designed to undertake an exploratory sensory evaluation to determine the acceptability of dual-fortified lentils (both uncooked and cooked) among 16 to 65-year-old consumers living in Ishurdi, a northern sub-district of Bangladesh.

Lentils are the most frequently consumed legume in Bangladesh where they are a staple food in the daily diet. Similar to other developing and some developed nations, both Fe and Zn deficiencies are common in the Bangladeshi population. Around one third (30%) of Bangladeshi adolescents are anemic, attributable mostly to Fe deficiency [20]. The 2011–12 National Micronutrient Survey of Bangladesh found that the national prevalence of Zn deficiency was approximately 45%, 52%, and 66% among preschool-age children, slum-dwelling preschool children, and non-pregnant, non-lactating women, respectively [21,22]. The expectation from the current study is that dual-fortified red and yellow cotyledon lentil dal will be equally acceptable to the lentil consumers with respect to taste, odor, appearance, texture, and overall acceptability.

The acceptability of fortified food depends on the fortificant type, dose, chemistry of the food vehicle, and interactions between different fortificants [23]. Fortification may create a metallic taste in foods, generate undesirable flavor due to fat rancidity, develop an unacceptable change in color, and degrade the quality of vitamins (e.g., vitamins A and C, which are important for absorption and utilization of Fe) [24]. The expectation of any fortification program is to contain any undesirable changes in food or food products. An earlier study of consumer-level sensory evaluations of cooked and uncooked Fe-fortified lentils (NaFeEDTA) showed that fortified lentils were well received by consumers compared to both unfortified lentils and those fortified with other Fe fortificants [25]. In this study, we hypothesized that dual fortification has a significant effect on liking for the sensory attributes of dual-fortified lentils. This hypothesis was based on the assumptions that there may be identifiable differences between dual-fortified and non-fortified lentils, and that identifying the differences in sensory properties may have major scientific implications for the food science industry.

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

#### *2.1. Study Design and Selection of Panelists*

We carried out a cross-sectional study between 1 February 2019 and 30 April 2019 at the Regional Agricultural Research Station, Ishurdi, Bangladesh. A group of 196 untrained lentil consumers, aged 16–65 years, participated in the sensory evaluation. A total of 50–100 responses are desirable for sensory evaluation according to the sensory evaluation guidelines of the Institute of Food Technologists' Sensory Evaluation Data [26].

Panelists were included on the basis of their willingness to participate in the study and their general health. The exclusion criteria were (i) having a fever, cold or, gum inflammation; (ii) taking medicines for cancer, thyroid, neurologic, or psychotropic treatment; (iii) being susceptible to an allergic reaction to lentils, iron or zinc; (iv) pregnancy, (v) having chewed betel leaf with betel nut and tobacco (locally known as paan/jarda) less than an hour before the sensory evaluation. A face-to-face interview technique was adopted since it was the appropriate method for filling in the sensory evaluation data by trained research assistants. With the proposed sensory trials, a preliminary assessment of consumer acceptability was conducted prior to carrying out a large-scale study with consumers.

#### *2.2. Preparation of Cooked and Uncooked Lentil Samples for Evaluation*

The most suitable Zn and Fe fortificants were selected after a series of experiments at the University of Saskatchewan Lab [16]. Based on those results, this sensory acceptability study for dual-fortified lentils was conducted. Two dual-fortified uncooked and one unfortified control sample from each of the three milled lentil product types (LPTs) (red football (RF), red split (RS), and yellow split (YS)) were evaluated by the consumers (Figure 1). One randomly selected dual-fortified sample (fortified with 16 mg Fe and 8 mg Zn per 100 g of lentil) and one unfortified control sample from each of the three LPTs were used to prepare a popular traditional recipe [25,27] commonly consumed in Bangladesh (Figure 2).


− − **Figure 1.** Images of uncooked lentil samples of three lentil product types (red split, left column; red football, middle column; and yellow split, right column), including the unfortified control (**upper row**) and two dual-fortified samples (i) fortified with 16 mg Fe from NaFeEDTA and 8 mg Zn from ZnSO<sup>4</sup> .H2O 100−<sup>1</sup> g of lentil (**middle row**), and (ii) fortified with 24 mg Fe from NaFeEDTA and 12 mg Zn from ZnSO<sup>4</sup> .H2O 100−<sup>1</sup> g of lentil (**lower row**).

− **Figure 2.** Samples of cooked dal prepared from each of the three product types (red split, left column; red football, middle column; and yellow split, right column) of lentil, including the unfortified controls (**upper three**) and dual-fortified samples (**lower three**) fortified with 16 mg of Fe and 8 mg of Zn 100−<sup>1</sup> g of lentil.

9-

Food samples were cooked in the food processing laboratory of the Bangladesh Agricultural Research Institute (BARI), Ishurdi, Pabna, Bangladesh. Hygiene and quality were maintained by using stainless steel cookware to prepare all cooked samples. We prepared a semi-thick lentil soup with each of the 6 different lentil samples. A portion (500 g) of each lentil sample was cooked for about 25 min using a local recipe, i.e., de-ionized water (2.5 L), turmeric (10 g), table salt (20 g), canola oil (30 mL) and chopped onion (100 g). All of the nine uncooked samples were separated into 4-oz white-colored foam cups, labelled with 3-digit codes, for evaluation by individual participants. After completing the uncooked sample evaluation, each participant was given one tablespoon of a cooked lentil dish or lentil soup from each of the six samples separately in 3oz plastic cups labelled with 3-digit codes. Water for rinsing the mouth between tastings was provided to the participants before and after testing each of the dishes.

#### *2.3. Data Collection Tools and Techniques*

Data were collected at two stages. At the first (screening) stage, a sampling frame was created among the interested participants and we used a simple random sampling technique to finalize the participants. At first, a screening questionnaire was used to collect the information from 200 lentil consumers (aged 16–65 years, who expressed interest in participation) with selected sociodemographic variables. A total of 196 study participants were selected for the final sensory evaluation study. In the second stage, a separate structured questionnaire was used for sensory evaluation. Both questionnaires followed forward–backward translation (English and Bengali). The sensory evaluation form had three parts. Part I covered demographic information, Part II included an evaluation of liking for the appearance, odor, and overall acceptability of the uncooked lentil samples. Similarly, an evaluation of liking for four sensory attributes (appearance, odor, taste, and texture), and the overall acceptability of the cooked dual-fortified lentil samples was also included. Participants' responses were captured using a 9-point hedonic scale (1 = dislike extremely and 9 = like extremely). In Part III, any opinions/comments from the participants regarding the tasted sample were documented (verbatim), whether they were positive or negative.

Sensory evaluation was carried out in a single day from mid-morning to mid-afternoon. A total of 20 research assistants (RA) were recruited a day before the interviews and were trained by a senior research investigator on the day of evaluation. The training mainly emphasized interview techniques and understanding the sensory evaluation form. After the training session, the data collection team practiced the administration of sensory evaluation forms to ensure the complete understanding and uniformity of the whole data collection process. We organized a total of 20 dual-fortified lentil booths that had uniform white light conditions and furniture for the testing of sensory attributes by the participants. Each participant scored the samples while seated face to face with the research assistant. Twenty participants took a test at one time and the sensory evaluation was conducted in single sessions to avoid reporting bias. Initially, uncooked samples were presented in a white tray for scoring. Then each participant was given one tablespoon of the cooked dishes or lentil soup from each of the samples separately. Deionized water was provided to the participants for oral rinsing before testing the first dish and after testing each of the dishes to cleanse the palate [25,28].

#### *2.4. Ethical Considerations*

The study was approved by the Research Ethics Office, the University of Saskatchewan, Canada (BH 14–729), the Bangladesh Medical Research Council, Bangladesh (BMRC/NREC/2016-2019/14) and the Asian Institute of Disability and Development (AIDD) Human Research Ethics Committee (HREC) (southasia-hrec-2019-3-01).

The anonymity and confidentiality of the study participants were strictly maintained. Written informed consent was received from each respondent. Unique identification numbers (UID) were assigned to each participant to maintain anonymity and confidentiality. Study participants had the right to withdraw from the study at any time during the interview or sensory evaluation

process. No side effects were expected in consideration of the amounts of Fe and Zn fortificants that respondents would consume during the evaluation study. All fortificants were food-grade quality. The toxicity level for Fe in the human body compared to the dose provided was negligible. However, monitoring was undertaken, and an adequate supply of water and necessary precautions were taken before initiating the sensory evaluation. Consent forms were stored separately from the collected data, which was stored on a password-protected computer and all associated computers were also password protected. Hard copies were stored in a locked cabinet. Data will be stored for 5 years after submission of the final report, at which point the soft copies will be deleted from computers and hard copies will be shredded.

#### *2.5. HunterLab Colorimetric Measurements of Unfortified and Dual-Fortified Uncooked Lentil Samples and Correlation with Sensory Attributes*

The lightness (L\*), redness (a\*), and yellowness (b\*) score of uncooked dual-fortified lentil samples from three LPTs were measured using a HunterLab instrument (Hunter Associates Laboratory Inc., Reston, VA, USA). L\* indicates the darkness to lightness, ranging from 0 to 100; a\* indicates greenness to redness, ranging from −80 to +80 and b\* indicates blueness to yellowness, ranging from −80 to +80 (Wrolstad and Smith, 2010). The HunterLab L\*, a\* and b\* scales were used for measurements three times per sample and the scores were analyzed using ANOVA in SAS 9.4 (SAS Inc. Cary, NC, USA). The sensory data of three attributes (appearance, and overall acceptability) of three LPTs of uncooked lentil samples were correlated with the L\*, a\*, and b\* scores using Pearson's correlation test.

#### *2.6. Statistical Analysis*

After data collection was completed, a dataset was prepared in SAS (Statistical Analysis Software, SAS Institute Inc., Cary, NC, USA) version 9.4. Datasets were reviewed by first entering the pretesting questionnaire data as a means of testing the practicability, and to check whether it covered every variable mentioned in the questionnaire. Scores for appearance, odor, taste, texture, and the overall acceptability of the fortified lentils were presented as means with standard deviations (SD). A One-Way Analysis of Variance (ANOVA) was performed to determine mean score differences among food samples, including the control. Statistical significance was set at *p* < 0.05. We tabulated the frequency and percentage as appropriate and used box plots to present sensory data using a 1–9 scale.

#### *2.7. Consistency Assessment for Sensory Data Based on Cronbach's Alpha*

Cronbach's alpha (CA) coefficient was used to measure the consistency of the panelists' responses since it measures the internal consistency reliability (ICR) of a sensory panel in multi-item evaluation scores [29]. It assessed the measurement error (between zero and one) by squaring correlation (α values) and by subtracting the end results from one, which provides the variation in the error that occurred in the measurement [30–32]. The value after subtraction represents the error variance in the score. We assessed the ICR of the liking scores for sensory attributes of 196 panelists in Bangladesh, for the nine uncooked and six cooked samples. Although there is no strict cut-off for CA, several studies report acceptable ICR ranges from 0.70 to 0.95 [33,34].

#### **3. Results**

#### *3.1. Demographic of the Study Participants*

The sociodemographic profile of the consumers is presented in Table 1. Among the participants, 59.2% and 40.8% were male and female, respectively, with an age range from 16–65 years, with a major portion (40%) in the 26–35 age range group. In total, 77.7% of the participants were from households where between one and five people were employed. Almost half (48.0%) of the participants had a monthly income ranging between BDT 10,000 and 19,000 (USD ~121–240).


**Table 1.** Socio-demographic profile of consumers who participated in the dual-fortified lentil sensory evaluation study in Bangladesh.

#### *3.2. Consumer Attitudes toward Lentil Consumption*

Among the participants, 52.0% and 13.8% of the respondents purchased 251–500 g and 751–1000 g of lentils per week, respectively (Table 2). Participants also bought other pulses at lower quantities compared to lentils—46.2% purchased 100–250 g of other pulses (chickpeas, mung beans, black gram, field peas, etc.) weekly, and 38.8% of the participants bought 251–500 g per week. Local markets were the primary source of purchased lentils (89.8%) followed by 8.1% from neighborhood grocery stores. The majority (76.5%) of panelists purchased lentils on a monthly basis and 89.9% preferred to buy red football LPT, followed by 9.7% who preferred red split LPT.


**Table 2.** Consumer habits and patterns of lentil consumption.

### *3.3. Liking for the Uncooked Fortified Lentil Dal*

Figure 3 show the mean, range, dispersion and outliers of the sensory attributes for the nine uncooked samples. For all three LPTs (RF, RS, and YS), consumer responses varied significantly for appearance, odor, and overall acceptability. The liking scores for sensory attributes, and for overall acceptability were significantly different between control and fortified LPTs for the three samples of both RF and RS lentils; however, insignificant differences were observed within the fortified samples. In YS lentils, the odor and overall acceptability scores significantly varied between fortified and unfortified lentil samples as well as within fortified YS lentil samples. For all attributes and product types, the highest preference score was observed for unfortified control lentil samples, followed by samples fortified with 8 mg Zn from ZnSO4H2O and 16 mg Fe from NaFeEDTA. The lowest score was recorded for the sample fortified with 12 mg Zn from ZnSO4H2O and 24 mg Fe from NaFeEDTA.

**Figure 3.** Box plot analysis of hedonic scores (1 = dislike extremely, 9 = like extremely) obtained for

−

−

three uncooked lentil dal samples (unfortified control lentil polished with 0.5% canola oil; dual-fortified with 8 mg Zn from ZnSO4H2O +16 mg Fe from NaFeEDTA (100−<sup>1</sup> g of lentils)); dual fortified with 12 mg Zn from ZnSO4H2O + 24 mg Fe from NaFeEDTA (100−<sup>1</sup> g of lentils) from each of the three product types, red football (**A**), red split (**B**) and yellow split (**C**), evaluated for appearance (**A**–**C**(**a**)), odor (**A**–**C**(**b**)), and overall acceptability (**A**–**C**(**c**)), by 196 panelists in Bangladesh. Different letters after mean values in the right column indicated significant differences between three samples within each attribute. Each box plot displays the distribution of data for each sample type separately based on a five-number summary, "minimum", first quartile (Q1), median, third quartile (Q3), and "maximum".

In general, the box plots for the control samples had a smaller range and less dispersion than those of the two fortified samples for all three LPTs. The box plot skewed either to the right (positive skew) or was neutral for unfortified control, with the average score significantly (*p* < 0.05) higher than that of fortified samples for each of the product types and attributes. In all three LTPs, the mean liking scores for the dual-fortified sample, fortified with 8 mg Zn from ZnSO4H2O and 16 mg Fe from NaFeEDTA, were significantly (*p* < 0.05) different but closer to the unfortified control compared to the other dual-fortified sample fortified with 12 mg Zn from ZnSO4H2O and 24 mg Fe from NaFeEDTA.

#### *3.4. Liking for the Cooked, Fortified Lentil Dal*

For all three LPTs, unfortified cooked control samples received the highest mean score for all five attributes (appearance, odor, taste, texture, and overall acceptability) compared to the fortified samples (fortified with 16 mg Fe from NaFeEDTA and 8 mg Zn from ZnSO4H2O) (Figure 4). An insignificant variation was observed for the two cooked lentil dal samples from all three LPTs evaluated by panelists, except for texture and overall acceptability of RF, and for appearance and overall acceptability of YS lentils. The numerical differences between scores across all samples of each of the three LPTs were very low for all five attributes. Specifically, the box plots for cooked samples showed less dispersion and a narrower range of liking scores for all attributes compared to those for the uncooked samples. All samples scored well (~7.0 = like moderately) for all five attributes.

#### *3.5. HunterLab Colorimetric Measurements of Unfortified and Dual-Fortified Uncooked Lentil Samples and Correlation with Sensory Attributes*

The results of the lightness (L\*), redness (a\*) and yellowness (b\*) scores of unfortified and dual-fortified lentil samples from three LPTs are shown in Table 3. For all three LPTs, a significant variation was observed between control and dual-fortified lentil samples for all L, a\* and b\* scores. Again, in all three LPTs, the highest and lowest L, a\* and b\* values were observed in unfortified-control and dual-fortified samples fortified with 24 mg Fe and 12 mg of Zn 100−<sup>1</sup> g of lentils. Among the two red football and red split dual-fortified samples, insignificant differences were observed for the L value, but for a\* and b\* values there were significant differences. Non-significant differences were observed between two dual-fortified samples for all thee scales.

The correlation coefficients between L, a\*, and b\* scores obtained from HunterLab and sensory acceptability scores were significant at *p* < 0.0001 with a range from 0.92 to 0.99 (Table 4). In the previous study, when we added different doses of Fe solution, the colorimetric test showed that with the increase in Fe dose, the red color of the lentil also became darker [14]. This result also showed a significant correlation with the sensory evaluations of uncooked samples by panelists. The appearance, odor, and overall acceptability were influenced by the increase or decrease in the Fe doses.

− **Figure 4.** Box plot analysis of hedonic scores (1 = dislike extremely to 9 = like extremely) for two cooked lentil dal samples [unfortified control lentil polished with 0.5% canola oil; dual-fortified with 8 mg Zn from ZnSO4H2O +16 mg Fe from NaFeEDTA (100−<sup>1</sup> g of lentils) for each of the three lentil product types—red football (**A**) red split (**B**) and yellow split (**C**). Samples were evaluated for appearance (**A**–**C**(**a**)), texture (**A**–**C**(**b**)), odor (**A**–**C**(**c**)), taste ((**A**–**C**)(**d**)), and overall acceptability (**A**–**C**(**e**)), by 196 panelists in Bangladesh. Different letters after mean values in the right column indicate significant differences between two samples within each attribute. Each box plot displayed the distribution of data for each sample type separately based on a five-number summary, "minimum", first quartile (Q1), median, third quartile (Q3), and "maximum".


**Table 3.** Lightness (L\*), redness (a\*) and yellowness (b\*) scores of one unfortified and two dual-fortified dehulled red football, red split and yellow split lentil samples prepared using Fe and Zn from NaFeEDTA and ZnSO4H2O, respectively.

<sup>a</sup> Mean <sup>±</sup> SD. Mean scores for lightness (L\*), redness (a\*) and yellowness (b\*) score followed by different Roman letters within columns are significantly different (*p* < 0.0001). <sup>b</sup> Unfortified control lentil; <sup>c</sup> Dual-fortified lentil with NaFeEDTA and ZnSO4H2O; <sup>d</sup> polished with 0.5% canola oil.

**Table 4.** Correlation coefficients between colorimetric data lightness (L\*), redness (a\*) and yellowness (b\*) score obtained from HunterLab and sensory acceptability scores from Bangladeshi consumers for three attributes (appearance, and overall acceptability) of each of three uncooked product types (red football, red split and yellow split) of lentil samples. all the correlation coefficients were found significant at *p* < 0.0001.


L\*, Lightness; a\*, redness; b\*, yellowness.

#### *3.6. Consistency Assessment for Sensory Data Based on Cronbach's Alpha*

Cronbach's alpha (CA) was used to evaluate the reliability of the sensory data. It creates a "proximity measure between evaluation profiles" by considering both variance and covariance relationships [29]. Table 5 presents the CA scores of both fortified and unfortified cooked and uncooked samples. The CA was ≥ 0.75 for uncooked samples. All the CA scores for cooked samples, except for unfortified YS control lentils polished with 0.5% canola oil, were greater than or equal to 0.80. Mean CA scores for uncooked and cooked samples were 0.84 and 0.81, respectively, which represents a high consistency in the evaluations of all samples using the hedonic scales.



*Foods* **2020**, *9*, 992

#### **4. Discussion**

Sensory evaluation encompasses effective measurements from consumers in terms of their liking, preference, and acceptability of food or food products [35]. The current study was undertaken to understand and evaluate the sensory attributes of dual-fortified lentils among lentil consumers in Bangladesh. The choice of Bangladesh as a study site was made for specific reasons. Lentils are considered a staple or partially staple food in many countries. About 56% of the lentils produced in the world are consumed in Asia [19], with a very high consumption in Bangladesh. Lentils are consumed frequently in daily meals due to their fast cooking properties, and they are also an inexpensive source of protein, carbohydrates, and micronutrients compared to animal sources. This study was conducted in one of the most important lentil-growing regions of Bangladesh. Most farmers of this region have experience with growing, processing, and marketing lentils. Moreover, the national Pulses Research Centre (PRC) of the Bangladesh Agricultural Research Institute (BARI) is located in this region. Several national and international organizations are actively involved with the Bangladesh national health sector in conducting research studies, sensory evaluations, and field trials with fortified foods, e.g., fortified rice. "Daal (pulses), vhat (rice)" or "hotchpotch", made with pulses (mostly lentils) and rice are common and popular dishes in most South Asian countries, including Bangladesh. Around 60% and 12% of Bangladeshi women consume lentils at a frequency of 3 and 4 days per week, respectively [36]. A similar study reported that 92% of the 384 respondents consumed hotchpotch at least once per week [36]. More than 80% of the lentil dal in the Bangladeshi market is imported from other lentil-growing countries, mostly from Australia and Canada. This provide an enormous opportunity to export dual-fortified lentil products to cope with both Fe and Zn deficiency problems in Bangladesh.

The concept of fortification is emerging in Bangladesh, although few fortified foods are available in the market, and some are under consideration. Two mandatory fortified foods, vegetable oil and salt with vitamin A and iodine, respectively, are now available in Bangladesh [22]. Research studies and evaluations of other fortified food products including rice, lentils, wheat flour, and sugar are underway in Bangladesh. A feasibility study of the field implementation of Fe-fortified lentil with adolescent girls in Bangladesh showed that respondents willingly consumed Fe-fortified lentil meals [22]. A large-scale double-blind community-based randomized controlled trial using Fe-fortified lentils with ~1200 adolescent girls in Bangladesh was recently completed, and results showed a significant effect of Fe-fortified lentils in improving the Fe-status of adolescent girls [37].

In any sensory evaluation study, consumers play a significant role in the preference assessments of product differences and characteristics [38]. The selection of the number of respondents in any consumer test depends on food/food products that need to be evaluated, the purpose of the test, the time frame, and the cost [39]. The recommended sample size for consumer acceptability tests suggest that 50–300 respondents are required for an acceptability test [40]. Suresh and Chandrashekara (2012) described a formula to calculate the sample size and showed that ~96 participants are acceptable to conduct research at the consumer level [41]. In this study, data from 196 participants were used to describe the objectives with statistical significance.

In a sensory analysis at the consumer level, sociodemographic data can be very useful to provide an insight as to whether or not the participants are representative of the total population when a specific food product is evaluated. An earlier study reported that socio-cultural diversity, socio-demographic factors and economic status affect consumer choice regarding functional foods [42]. In the current study, data recorded on participant diversity in terms of age, gender, monthly income, employment status, education, and lentil consumption attitudes confirmed the representativeness of the general consumers (Table 1).

Consumer attitudes toward lentil consumption showed that Bangladeshi consumers preferred red lentil dal compared to other pulses. Among the two product types of red lentil dal, the football type was more preferred (89.8%) than the split type (Table 2). Unlike red lentil dal, dehulled yellow cotyledon lentil dal is usually produced from lentils with green coats, and is not yet well known in

the Bangladeshi market. Whole (not decorticated) green lentils have been using in the snack industry for several years, but not for soup preparation at the household level. As lentil demand is increasing around the world [43] and market research for green lentil products has been initiated in different South Asian countries, our goal was to introduce the dehulled yellow lentil dal to the lentil consumers and evaluate consumers' attitudes to this type of lentil along with the red type. Most of the consumers (around 90%) in Ishurdi bought lentils from the local market, where lentils are sold by scooping from open sacs or in 1–2 kg plastic bags. The previous study [43] in an urban market showed that 37% of consumers bought lentils from local markets or retail shops. This difference could be due to the sociodemographic differences between urban and suburban areas. Fortified lentil is considered a value-added food product that requires packaging in sealed bags to ensure quality and to reduce the risk of adulteration.

Sensory responses to uncooked lentil dal samples revealed significant differences between unfortified and dual-fortified samples for all three LPTs (Figure 3A-C). Although the differences were numerically very low, liking scores from all three attributes (appearance, odor, and overall acceptability) decreased significantly with the increase in Fe and Zn concentration. In all three LPTs, the unfortified controls received higher scores than the fortified samples for all three attributes. Among the three control samples from three LPTs, the RF control got the highest score compared to the other two control samples of RS and YS, indicating the preference for RF lentils compared to RS and YS lentils. Overall acceptability scores for RF, RS, and YS lentils ranged from 7.0 to 8.0, 6.8 to 7.5, and 7.3 to 8.0, respectively. For all the three LPTs, insignificant differences were observed between two dual-fortified lentil samples for all three attributes, except for the order and overall acceptability of YS lentils. A previous study [43] showed that with the increase in Fe fortificants, liking scores decreased in Fe-fortified lentils. The results from this study indicate that Zn fortificants might help to protect the lentil from darkening, even with higher doses of Fe (24 mg of Fe 100−<sup>1</sup> g of lentil). The results also show that the dual fortification of YS lentils is more susceptible to the development of an off-color appearance than RF and RS lentils with higher doses of Fe and Zn fortificants. In three LPTs, three attributed mean scores of uncooked samples ranged from "like moderately, a score of 7" to "like very much, a score of 8". Moreover, from all three LPTs and three attributes, several participants scored "9, like extremely" for control samples and a dual-fortified sample fortified with 16 mg of Fe and 8 mg of Zn per 100 g. Overall, the results indicated that dual-fortification with Fe and Zn did not have a large adverse effect on the sensory characteristics of any LPTs.

Non-significant differences were observed between cooked control and dual-fortified lentil samples for all five attributes (appearance, odor, taste, texture and overall acceptability) of the three LPTs, except for texture and the overall acceptability of RF lentils and the appearance and overall acceptability of YS lentils. For all three LPTs, the control lentils received a numerically higher score for all five attributes compared to dual-fortified products. Overall, liking scores for all three LPTs indicated that both cooked samples from each of the three LPTs were accepted equally by the participants. Boxplot comparisons of both uncooked and cooked samples showed that some outlier scores might have greatly influenced the average score of the lentil samples. Some consumers scored the uncooked dual-fortified sample (fortified with 24 and 12 mg of Fe and Zn, respectively) with the two lowermost hedonic scores (dislike extremely, a score of one; dislike very much, a score of two). Some consumers also noted the floating of a black-colored substance, and black spots in the cooked and uncooked samples, respectively. The black spot is the micropylar region of dehulled lentils that is insoluble in water and, after cooking, this region detaches from the cotyledon and floats in the soup. During fortification, this whitish embryonic tissue absorbs fortificant from the solution, resulting in a slight discoloration caused by oxidation [25]. This dark micropylar region could, however, be used as an indicator to help consumers distinguish fortified lentils from unfortified lentils.

In this study, two samples from each of the three LPTs were selected for evaluation by consumers, including one control and one dual-fortified sample with 16 mg and 8 mg of Fe and Zn, respectively. A previous study [16] showed that dual-fortified RF, RS, and YS lentil products fortified with 16 mg Fe

and 8 mg Zn per 100−<sup>1</sup> g of lentils, can provide Fe and Zn at 27.1 to 13.9 mg, 28.0 to 13.4 mg, and 29.9 and 12.1 mg per 100−<sup>1</sup> g of lentil, respectively. The control samples from each of the three LPTs contain Fe and Zn at 7.5 to 4.3, 7.1 to 4.4 and 5.9 to 3.9 mg per 100−<sup>1</sup> g of lentil. Each of the 196 participants consumed 15 mg of lentils from each of the cooked samples. Each of the participants consumed a total of 90 g of lentils from both fortified and control lentil samples. From 90 g of lentils, each participant consumed 15.3 mg (4.06 + 4.20 + 4.5 + 1.1 + 1.1 + 0.89) of Fe and 7.76 mg (2.08 + 2.01 + 1.82 + 0.64 + 0.63 + 0.58) mg of Zn. The tolerable upper intake level of iron and zinc per day for males and females (19+ years) is 45 mg/day and 40 mg/day, respectively.

Liking scores for all sensory attributes and for overall acceptability from both uncooked and cooked samples showed that consumers scored differentially for similar samples when cooked lentils were compared to uncooked lentils. The wider range of scores observed for uncooked samples was narrowed down after cooking. The reduced score range could be due to cooking the lentils following a traditional lentil soup preparation recipe [27]. Dry turmeric (*Curcuma longa* L.) powder and onion (*Allium cepa* L.) are the two common ingredients used to cook lentils. The yellow color of turmeric would change the soup's appearance and suppress the darkness of fortified lentils. The pungent smell of onion also has a significant effect on changing odor and taste profiles and can suppress the metallic taste (if any is detectable) of fortified lentils after cooking [25]. Insignificant differences in sensory attributes were also reported for cooked conventional and fortified rice [44]. Iron and Zn from the fortificants may affect the taste. Since we did not measure biological assessments that affect taste, and as this study was conceptualized to capture a real working scenario in the study population, our study cannot address this issue. In addition, the fortified lentils used in this study were produced in the Saskatchewan Food Industry Development Centre, Canada. In Canada, canola oil is commonly used to polish the lentils after dehulling and cleaning to give them a shiny look that increases consumer attraction. In this study, we did not use palm oil or soybean oil to avoid any interaction between the two different oils, which may have altered the taste and odor. Moreover, participants had the recipe explained to them before the sensory evaluation started.

Sensory analysis helps to evaluate products in a relatively short time and at a low cost with representative consumers who consume the identified product and have sensory skills [45]. The effects of dual fortification on the sensory properties of food are highly variable and depend on the Fe and Zn fortificants and food items [23]. In this study, although consumers could easily distinguish the fortified samples from the control, the overall acceptability was more similar when the samples were cooked. The recommended intake of pulses is 50 g/day/person [46] and the estimated average requirement (EARs) of the Fe and Zn is 29.4 mg and 4.9 mg for males and 18.8 mg and 7.0 mg for females, respectively [23]. Consumption of dual-fortified lentils instead of unfortified lentils could be a prime option to provide a sufficient amount of Fe and Zn in a rapid manner in comparison to other micronutrient intervention approaches mentioned by Northop-Clewes (2013) [23].

The consistency of sensory data was assessed by calculating the Cronbach's alpha (CA) value, which showed that panelists were consistent in scoring both uncooked and cooked samples and that the CA values were within the acceptable range (0.75 to 0.95) [33,34]. Only one LPT sample (YS control, 0.64) was below the suggested range. This could be due to the inconsistent scoring of consumers for this sample. Although YS lentils were introduced to participants before scoring, some participants did not score the YS sample. One study reported that missing values have an effect on the psychometric properties of any test [47]. However, generalizability cannot be explained through this study since data were cross-sectional in nature. We therefore advise caution when interpreting these specific results.

#### **5. Conclusions**

Overall, dual fortification decreased consumers' liking for uncooked lentils, but not cooked ones. We also found high acceptability of the dual-fortified red lentils and no major issues related to acceptability were observed for sensory attributes. We estimated that the dual-fortified samples used in a cooked dal preparation for the three lentil product types can provide approximately 14 mg of Fe and 6.5 mg of Zn from 50 g of lentils. This represents a major part of the estimated average requirement (EARs) of Fe and Zn currently recommended by the World Health Organization (WHO).

**Author Contributions:** R.P., M.H.A.I., I.J. and F.M.Y. designed the study. R.P., M.H.A.I., I.J. conducted the study and analyzed the data. The manuscript was drafted by R.P. R.P., M.H.A.I., I.J. and F.M.Y., M.M. and A.V. helped to interpret the data, edited the draft, reviewed all documents critically, and approved the final manuscript for submission to the journal. All authors have read and agreed to the published version of the manuscript.

**Funding:** Saskatchewan Ministry of Agriculture (Agriculture Development Fund), Fund number: 420088.

**Acknowledgments:** The authors would like to acknowledge financial assistance received from the Saskatchewan Ministry of Agriculture (Agriculture Development Fund) and Grand Challenges Canada. The authors are grateful for the technical assistance provided by the research scientist and scientific assistants of the Pulses Research Centre, Bangladesh Agricultural Research Institute, Ishurdi, Bangladesh and for their invaluable support in conducting the sensory trial in Bangladesh. We also acknowledge all of the participants who participated in this study.

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

#### **References**


© 2020 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* **Legume Flour or Bran: Sustainable, Fiber-Rich Ingredients for Extruded Snacks?**

#### **Cristina Proserpio \* , Andrea Bresciani , Alessandra Marti and Ella Pagliarini**

Department of Food, Environmental and Nutritional Sciences (DeFENS), Università degli Studi di Milano, 20133 Milan, Italy; andrea.bresciani@unimi.it (A.B.); alessandra.marti@unimi.it (A.M.); ella.pagliarini@unimi.it (E.P.)

**\*** Correspondence: Cristina.proserpio@unimi.it

Received: 16 October 2020; Accepted: 14 November 2020; Published: 17 November 2020

**Abstract:** The impact of using legume flour and bran on both sensory and texture properties in extruded, sustainable snack formulations was investigated. Sensory attributes determining consumer preference or rejection of legume-based snacks, as well as food neophobia and food technology neophobia were also explored. Seven samples of extruded snacks (R = 100% rice flour; C = 100% chickpea flour; P = 100% green pea flour; C30 = 30% chickpea bran and 70% rice flour; C15 = 15% chickpea bran and 85% rice flour; P30 = 30% green pea bran and 70% rice flour; P15 = 15% green pea bran and 85% rice flour) were subjected to the three-point bend method using a TA.XT plus texture analyzer. Seventy-two subjects (42 women; aged = 29.6 ± 9.3 years) evaluated the samples for liking and sensory properties by means of the check-all-that-apply (CATA) method. The sample made with 100% rice flour obtained the lowest liking scores, and it was not considered acceptable by the consumers. Samples P, C, C15, and P15 were the preferred ones. Crumbliness and mild flavor attributes positively influenced hedonic scores, whereas stickiness, dryness, hardness, and to a lesser extent, visual aspect affected them negatively. Neophilic and neutral subjects preferred the snacks compared with the neophobic ones, while no differences in liking scores were found regarding food technology neophobia. Extruded snacks with legume flour and bran were moderately accepted by consumers involved in the present study, albeit to a lesser extent for neophobic subjects, and could represent an interesting sustainable source of fiber and high-value proteins, as well as a valuable alternative to gluten-free foods present on the market.

**Keywords:** acceptance; sensory descriptive analysis; CATA; texture analyzer; pulses; green peas; chickpea; rice

#### **1. Introduction**

One of priorities of the food industry is to reduce the environmental impact of its production. This objective can be achieved using several strategies, including the improvement of food chains that have less of an impact than others and focusing on a "circular economy" to reintroduce bioactive components from waste or by-products into new food formulations.

Legume production can satisfy both the abovementioned strategies. Firstly, a plant-based food system requires less resources in terms of water, land, and energy compared with a meat-based food system [1]. Legumes supply nitrogen for fertilization, since they can fix atmospheric nitrogen, thus reducing the amount of fertilizer used on the crops and increase proteins in animal feeding [2,3]. Secondly, milling by-products could be recovered to obtain bioactive components to be used as value-added ingredients in innovative food products. Among these components, legume bran has a high amount of dietary fiber, ranging from about 75% to 90% for chickpea and pea, respectively. Specifically, legume hull fiber is mostly insoluble fiber, whose purity is above the 80% [4].

It is well known that potential health benefits have been associated with the consumption of an appropriate amount of fiber [5]. Specifically, several epidemiological studies have highlighted that dietary fiber decreases the incidence of various diseases such as some types of cancers (e.g., colon and ovarian) [6,7], cardiovascular disease [8], and, in general, decreases the risk of mortality [9]. Moreover, it has been reported that the significant reduction in fiber consumption observed in industrialized countries is related to a worrying increase in cases of overweight people. Fiber intake in place of other macronutrients would lead to a decrease in calorie intake that could be extremely important for the overall health of the Western world [10]. However, the total dietary fiber consumed by the average individual is rather low, about <50%, of the recommended daily amount [11]. Even if proper nutrition requires legumes as a staple food in consumer diets, this recommendation is not often followed because cooking legumes can require a rather long preparation and because legumes are often not generally appreciated from the sensory point of view [12]. In recent years, a great deal of interest has been placed on partially replacing cereal-based products, such as pasta and bread, with legume flours to increase their nutritional profile [13].

Even if fiber-enrichment adds value in the eyes of the consumer [14], the addition of fiber in a food matrix causes changes in the production process, sensory properties, as well as texture and rheological parameters [15,16]. Therefore, the impact of adding fiber to a specific product needs to be studied. From a technological standpoint, fiber breaks down the starch–protein matrix, leading to important structural changes. Adding a small concentration of fiber could improve the structure of some products thanks to its ability to bind with water, but a high amount of fiber is almost always associated with a worsening of structural characteristics [17]. The types of fibers, as well as the food matrices in which these components are added, influence hardness, adhesiveness, and sensory attributes that cannot be generalized [18]. Previous findings have highlighted that pulse flours can be added up to of 40% (based on flour) in baked products without reducing their sensory quality [19], while other results have revealed that acceptability starts to decrease when more than 20% of wheat flour is replaced with that of legumes [20]. In addition to baked goods (e.g., bread and crackers), legumes might partially or totally replace cereals in the production of extruded snacks, products for which consumer demand is increasing since they can satisfy the demand for healthy, minimally processed, ready-to-eat foods.

Therefore, the aim of the present study was to investigate the impact of using legumes and legume bran on the sensory and texture properties of extruded snack formulations. The sensory attributes determining consumer preferences with regards to the experimental samples were studied. Food neophobia, which is the fear to try new and unfamiliar foods [21], and food technology neophobia, which refers to new food technologies [22], were also explored as behavioral attitudes playing a key role in defining consumer behavior.

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

#### *2.1. Materials*

Flours from milled rice (82.00% carbohydrates; 9.13% proteins; 1.19% lipids; 0.95% fiber), decorticated green pea (59.00% carbohydrates; 25.00% protein; 1.97% lipids; 9.50% fiber), and decorticated chickpea (56.00% carbohydrates; 24.00% proteins; 6.60% lipids; 10.10% fiber) were kindly provided by Molino Peila S.p.A. (Valperga, Italy), as well as the bran obtained from green pea (92.00% fiber; 3.30% proteins; 0.22% lipids) and chickpea (78.00% fiber; 11.20% proteins; 5.30% lipids). All values are expressed on dry basis.

Co-extruded snacks were prepared from rice (R), green pea (P), and chickpea (C). Moreover, bran from both green pea and chickpea were included in rice-based snack formulation at 15% and 30% levels, obtaining four different bran-enriched samples: C15, C30, P15, and P30. Overall, seven formulations were tested. Co-extruded snacks were produced at an industrial level by Fudex Group S.p.A. (Settimo Torinese, Italy) in the shape of bars. Extrusion was performed using a co-rotating

twin-screw extruder (model 2FB90; screw speed: 150 rpm; temperature: 110 ◦C; pressure: 70 bar; Settimo Torinese, Italy).

#### *2.2. Instrumental Texture Analysis*

The textural properties of the snacks were determined by a three-point bend method using a TA.XT plus texture analyzer (Stable Micro Systems Ltd., Godalming, UK) equipped with a 100 N load cell. Snack bars were compressed with the Heavy, Duty Platform/Three Point Bending (HDP/3PB) probe at a crosshead speed of 1 mm/s to 5 mm of the original diameter of the snack. The compression generated a curve with the force over distance. The highest value of force was taken as a measurement for hardness. The test was carried out on 35 pieces for each sample, and the average value was considered.

#### *2.3. Sensory Evaluations*

#### 2.3.1. Subjects

Seventy-two subjects (42 women; mean age: 29.6 ± 9.3 years) were recruited among students and employees of the Faculty of Agriculture and Food Sciences of the University of Milan. The exclusion criteria were as follows: subjects who did not like rice and legumes, subjects suffering from food intolerances and allergies, as well as those who were on medical treatments that could modify taste perception. This study, approved by the Ethics Committee of the University of Milan, was conducted in compliance with the principles laid down in the Declaration of Helsinki. All subjects provided informed, written consent prior to participation.

#### 2.3.2. Hedonic Evaluation

Subjects were asked to taste the products and to express their liking using a labeled affective magnitude (LAM) scale, anchored by the extremes "greatest imaginable dislike" (score 0) and "greatest imaginable like" (score 100) [23]. Prior to tasting, the experimenters provided to the participants instructions for the use of the scale.

#### 2.3.3. Sensory Descriptive Evaluation

A separate group of 12 untrained subjects (mean age: 22.0 ± 4.1 years) were involved in a focus group, wherein they used a free listing questionnaire to define the appropriate sensory attributes to describe the extruded snacks [24]. Subjects had to evaluate the sensory characteristics of the snacks and identify all attributes for describing their color, appearance, odor, taste, flavor, and texture. After the development of the individual lexicon, an open discussion was made, and sensory attributes were selected by the experimenters considering the most commonly mentioned (frequency of terms selection at least of 40%) words in order to avoid synonyms [25]. Finally, the check-all-that-apply questionnaire consisted of a list of 23 sensory attributes: 3 for the appearance (dark yellow, light yellow, and green), 6 for the odor (strong, mild, toasted, rice, whole-meal, and legume), 3 for the taste (sweet, bitter, and salty), 6 for the flavor (strong, mild, rice, peas, chickpeas, and spicy), and 5 for the texture (crumbly, sticky, hard, porous, and dry). Subjects were asked to select the terms best describing each sample. Attributes' positions were randomized using the "to assessor" list order allocation scheme [26].

#### 2.3.4. Questionnaires

#### Food Neophobia Scale

Neophobic traits were investigated through the Food Neophobia Scale (FNS) developed by Pliner and Hobden (1992) [21]. The FNS consists of 10 statements each offering 7 graded response alternatives, from "strongly disagree" (score 1) to "strongly agree" (score 7). After reversing the negatively worded statements, the FNS score was calculated as a sum of the responses, yielding a range of 10–70.

#### Food Technology Neophobia Scale

In order to investigate individual attitudes toward new food technologies, the Food Technology Neophobia Scale (FTNS) [22], consisting of 13 items, was used. Each statement offers 7 graded alternative responses, from "strongly disagree" (score 1) to "strongly agree" (score 7). Four of the 13 items reflect food neophilia, so responses had to be reversed in order to calculate the final neophobia score. The FTNS score was calculated as a sum of the participant's answers for each statement, yielding a range from 13 to 91. Higher scores indicate a higher food technology neophobia level.

#### 2.3.5. Experimental Procedure

Subjects attended one online session and one laboratory session. During the online session, they were asked to complete a questionnaire including demographic variables and the food neophobia and the food technology neophobia scales. Subsequently, they were invited at the sensory and consumer science laboratory designed according to ISO guidelines (ISO 8589 2007) and were asked to refrain from consuming anything but water for 2 h before the test.

Samples were provided to the participants following a monadic presentation (one at a time) in a serving portion of approximately 30 g. The experimental samples were presented to the participants in plastic plates labeled with three-digit codes. Water was available for rinsing the palate between the samples. For each sample, subjects had to evaluate their overall liking and perform a sensory descriptive analysis by means of the check-all-that-apply (CATA) methodology. The entire session took approximately 30 min. Data were collected using the Fizz v2.47 software program (Biosystemes; Couternon, France).

#### *2.4. Data Analysis*

One-way analysis of variance (ANOVA) was applied to the data obtained from instrumental texture analysis, and the least significant differences were calculated by the Tukey's HSD test.

ANOVA model was performed on overall liking scores considering samples (R, P, C, C15, C30, P15, and P30), gender (women and men), age (≤26 years old; >26 years old) and their interactions as factors. When a significant difference (*p* < 0.05) was found, the LSD post hoc test was performed as a multiple comparison test.

The frequency of mention for each term of the CATA questionnaire was determined by counting the number of consumers who used that term to describe each sample. Cochran's *Q* test was applied to identify which sensory attributes were discriminating among samples. The relationship between samples and sensory attributes was evaluated by means of correspondence analysis (CA). The influence of sensory attributes' perception on hedonic scores was also investigated by means of penalty-lift analysis. This analysis suggests which sensory attributes are significantly (*p* < 0.05) positively or negatively associated with hedonic responses [27].

Correlations between instrumental and sensory texture data were examined using Pearson's correlation coefficient with a minimum significance level defined as *p* < 0.05.

The internal consistency reliability of the food technology and food technology neophobia scale was explored by Cronbach's alpha. ANOVAs were performed on FTNS and FNS scores considering age, gender, and their interactions as factors. To investigate the relationship between food neophobic traits and snack liking, subjects were categorized according to their neophobia scores into the following three groups: adults with scores in the lower 25th percentile of FNS scores, score <14 (Neophilic\_FNS); adults with scores between the 25th and 75th percentiles, 14 ≤ FNS score ≤ 31 (Neutral\_FNS); and adults with scores >31 (Neophobic\_FNS). The same approach was used to identify subjects showing a lower (score < 31; Neophilic\_FTNS), medium (31 ≤ FNTS score ≤ 46; Neutral\_FTNS), or higher (score > 46; Neophobic\_FTNS) level of food technology neophobia.

ANOVA models were performed on liking data considering FNS level, FTNS level, gender, age, and their interactions as factors. All analyses were performed using IBM SPSS Statistics for Windows, Version 24.0 (IBM Corp., Armonk, NY, USA) and XLSTAT (Version 2019.2.2, Addinsoft™, Boston, MA, USA).

#### **3. Results**

#### *3.1. Hardness*

Snack hardness is shown in Table 1. Snacks based on chickpea showed the highest value, almost two-fold higher than that of rice snack, which was used as control. On the other hand, snacks from green pea showed the least resistance to breakage. The addition of 15% bran from either green pea or chickpea did not significantly affect the snack texture in comparison with the 100% rice snack. Conversely, the type of bran was relevant when the milling by-product was included at 30% level. Specifically, adding 30% chickpea bran significantly decreased the force necessary to break the snack. On the contrary, in the case of 30% green pea bran, an increase (although not significant) in hardness values was recorded.

**Table 1.** Hardness values of co-extruded snacks. Mean (*n* = 35) ± SEM. Different letters in the column correspond to significant differences (Tukey's HSD test; *p* < 0.05).


R = snacks from 100% rice; C = snacks from 100% chickpea; P = snacks from 100% green pea; C15 = snacks from 85% rice + 15% chickpea bran; C30 = snacks from 70% rice + 30% chickpea bran; P15 = snacks from 85% rice + 15% green pea bran; P30 = snacks from 70% rice + 30% green pea bran.

#### *3.2. Hedonic Evaluation*

Hedonic evaluation results are provided in Table 2. A significant sample effect was found for liking scores. The rice sample obtained the lowest liking score and was not considered acceptable by the consumers (mean hedonic score lower than middle of the scale = 50), while the samples with 100% pea and C15 were the preferred. Comparable liking scores were also provided for samples made with 100% C as well as formulations with 15% pea bran. These two last formulations were in turn comparable to the snacks made with legume bran at 30% (C30 and P30).

**Table 2.** Mean hedonic ratings ±SEM by samples, gender, and age groups. Hedonic scale range 0–100. Different letters show significant differences (*p* < 0.05) according to post hoc test.


R = snacks from 100% rice; C = snacks from 100% chickpea; P = snacks from 100% green pea; C15 = snacks from 85% rice + 15% chickpea bran; C30 = snacks from 70% rice + 30% chickpea bran; P15 = snacks from 85% rice + 15% green pea bran; P30 = snacks from 70% rice + 30% green pea bran. Significant *p*-values are reported in bold

A significant gender effect on liking scores was also found, with men providing generally higher scores compared to women. Moreover, younger subjects gave generally higher scores compared to older subjects. The two- and three-way interactions were not significant.

#### *3.3. Sensory Descriptive Evaluation*

The frequency table of terms checked by consumers to describe snack samples is reported in Table 3.

**Table 3.** Frequency counts (%) of check-all-that-apply (CATA) terms used to describe the extruded snacks and results of Cochran's Q test for comparison among the samples.


R = snacks from 100% rice; C = snacks from 100% chickpea; P = snacks from 100% green pea; C15 = snacks from 85% rice + 15% chickpea bran; C30 = snacks from 70% rice + 30% chickpea bran; P15 = snacks from 85% rice + 15% green pea bran; P30 = snacks from 70% rice + 30% green pea bran. Different letters show significant differences (*p* < 0.05) according to post hoc test. N.s.= not significant; \*\* *p* < 0.01; \*\*\* *p* < 0.001.

Cochran's Q test yielded both discriminating and non-discriminating sensory attributes. Significant differences were found in the frequency of mention for 19 out of 23 terms for the five categories considered, suggesting that consumers perceived differences between samples in terms of their sensory characteristics. The sensory attributes that were not useful in order to discriminate samples were: mild odor, salty taste, bitter taste, and dry. In fact, snacks samples were generally characterized by a mild odor and low salty and bitter tastes.

A bi-plot of the products based on sensory descriptive analysis was obtained by means of a correspondence analysis (CA). The CA performed on the total frequency of participants' counts for each attribute resulted in two dimensions accounting for 79.09% of variance of data. As shown in Figure 1, samples were discriminated according to bran percentages, with all samples containing bran (C15, C30, P15, and P30) positioned in the upper left side of the map well separated from the other samples not containing bran. In the other three quadrants, sample with 100% legumes (C and P) and 100% rice (R) were positioned.

**Figure 1.** Correspondence analysis from check-all-that-apply data. Snack samples are reported in blue; the sensory attributes in black (O = odor, F = flavor).

The main sensory attributes that significantly (*p* < 0.05) influenced consumer hedonic perception are reported in Figure 2. Penalty analysis results revealed that two sensory attributes played a positive influence (drivers of liking: mild flavor and crumbly), and five attributes had a negative influence (drivers of disliking: light-yellow color, hard, dry, and sticky).

**Figure 2.** Penalty-lift analysis of sensory attributes across all snack samples. Only attributes that resulted in significant increase or decrease in overall liking are presented. (F = flavor).

Looking to Pearson's correlation coefficient in Table 4, significant correlation was found between texture results obtained by instrumental measurement and sensory data. In particular, positive correlations (*p* = 0.07) were highlighted between hardness (N) and "hard" attribute, while significant negative correlations were found between hardness (N) and crumbly, porous attributes.


**Table 4.** Pearson correlation coefficients among texture attribute perception and hardness measured by instrumental analysis (N).

(\*) *p* < 0.10; \* *p* < 0.05; \*\* *p* < 0.01.

#### *3.4. Food Neophobia*

Satisfactory internal consistency of food neophobia scale, as calculated by Cronbach's alpha test (Cronbach's alpha = 0.92), was observed among items. The mean food neophobia value of subjects involved was 23.4 ± 12.5. No significant differences in neophobic traits could be attributed to gender and age (F(1,68) = 0.44, *p* = 0.51; F(1,68) = 1.22, *p* = 0.27, respectively). A significant effect of food neophobia on liking scores was found (F(2420) = 3.46, *p* = 0.03). As reported in Figure 3, neophilic and neutral subjects gave generally significant higher liking scores (53.2 ± 1.6; 51.5 ± 0.9, respectively) compared with neophobic subjects (46.8 ± 1.9).

**Figure 3.** Mean liking scores ± SEM according to food neophobia levels. \* *p* < 0.05. FNS = Food Neophobia Scale.

The food neophobia level × gender interaction was also significant (F(2420) = 4.58, *p* = 0.01). As reported in Figure 4, no significant differences were found in hedonic scores according to gender in neophilic subjects (women: 55.1 ± 2.0; men 51.3 ± 2.5), while significant higher scores were provided by men with neutral food neophobia level (53.8 ± 1.4) and neophobic FNS (52.2 ± 3.0) compared with women (Neutral\_FNS: 49.2 ± 1.3; Neophobic\_FNS: 41.1 ± 2.4). No significant food neophobia level × sample effect was found.

**Figure 4.** Mean liking scores ± SEM according to food neophobia levels and gender. \* *p* < 0.05; \*\* *p* < 0.01. FNS = Food Neophobia Scale.

#### *3.5. Food Technology Neophobia*

Cronbach's alpha for the 13 items in the FTNS assessment showed a satisfactory internal consistency (Cronbach's alpha = 0.82). The mean food technology neophobia value was 40.3 ± 1.3. Significant differences (F(1,68) = 4.42, *p* = 0.03) in neophobic traits according to age were found, with higher scores provided by subjects >26 years old (42.9 ± 1.7) compared with younger subjects (37.5 ± 1.9). No gender and gender × age effects were found on FTNS scores (F(1,68) = 0.22, *p* = 0.64; F(1,68) = 0.62, *p* = 0.43, respectively).

As regards the influence of food technology neophobia level on liking scores, no effect was found (F(1420) = 0.72, *p* = 0.48)

#### **4. Discussion**

In the present study, the use of chickpea and green pea flour and related bran in extruded snack formulations was investigated considering both sensory and texture properties. Sensory attributes influencing consumer preferences were characterized. Moreover, food neophobia and food technology neophobia were considered to define whether these behavioral attitudes could impact on hedonic perception.

Even though the use of legumes as a high-fiber and high-protein ingredient in food formulation has been widely investigated [28], to our knowledge this is one of the first studies that has evaluated consumer responses to extruded snacks containing different percentages of chickpea and green pea bran as sustainable food ingredients.

Samples developed with 100% chickpea and green pea, as well as samples with different percentages of legume bran, obtained significantly higher liking scores compared with the control sample made only with rice. These results suggest that the legume-based formulations developed here have a better market potential compared with the more traditional rice-based snacks. Legume-based snacks represent a promising gluten-free alternative not only for subjects with gluten allergy or intolerance but also for those who follow gluten-restricted diets for health reasons [29]. A gluten-free diet is actually one of the most popular diets, with a greater number of people avoiding gluten for nonmedical reasons than those who are dealing with a gluten-related disorder [30]. Moreover, due to their high-fiber content, the consumption of the legume-based snacks could help consumers reach their daily recommended intake of dietary fiber, which could have a potentially positive health effect. Indeed, despite the proven beneficial effects associated with a fiber-rich diet, the average intake of such components in adults is lower than the recommended daily intake [11]. In this context, food products, such as minimally processed snacks and ready-to-eat foods, with a low fat and salt, high fiber, and high-value proteins could be part of a balanced diet and lead to a consequent good health status [31]. Cereal-based snacks are mainly produced by extrusion-cooking, i.e., a relatively cheap, easy, and versatile technology that allows the production of a variety of textures and shapes that appeal to consumers [32]. The positive effects of extrusion on nutritional traits, including the decrease in antinutritional factors and the increase in soluble dietary fiber and in protein and starch digestibility, have been widely discussed [32–34]. On the other hand, extrusion might cause the loss of heat-labile vitamins and the reduction of the nutritional value of proteins, due to the Maillard reaction between protein and sugars.

The results reported in this paper agreed with the study of Balasubramanian and collaborators (2012) [35] who found that extruded samples made with black gram, green gram, lentil, and peas were well accepted. However, since previous sensory data on extruded snacks with legumes were obtained involving a small number of consumers, and thus not leading to robust and reliable results, the comparison between our hedonic data and previous results is not indicative. Generally, the replacement of cereals with legumes leads to a general trend toward decrease in food acceptability as the percentage of legumes increases [36]; however, it greatly depends on the food matrices used and how the process conditions are adapted with respect to the change in formulation. Indeed, legume flour in some products, such as biscuits and pasta, can enhance food acceptability [37].

It should also be pointed out that encouraging the legume chain represents an important sustainable action that can help to reduce greenhouse gas emissions, break the cycle of pests and diseases with crop diversification in agroecosystems, and contribute to protein production [38]. Moreover, using legume bran as a value-added ingredient for new food formulations reduces the environmental impact of this food chain [39]. Descriptive analyses revealed that some sensory evaluations were more affected by the type of legumes used in snack formulations rather than by the quantity of bran added. This is in line with evidence that changes introduced by the addition of bran are much more significant in wheat flour products than in gluten-free products that do not have such a complex and functional matrix as a gluten network has. To corroborate this hypothesis, it has been reported that fiber addition generally reduces acceptability in terms of consistency, flavor, and appearance, although when initial acceptability is low, as for gluten-free products, fiber addition can improve consumer preference [40–42].

Among sensory attributes, texture was found to be the most interesting. Indeed, sensory descriptive analysis revealed that, besides mild flavor, the other sensory attributes that positively affected overall hedonic responses were related to texture properties. Although texture has been referred to as the "forgotten attribute" due to the little attention it has received for several years [43], it is a complex sensory dimension including tactile, visual, and auditory sensations, playing an important role in defining consumer responses [44]. The present findings indicated that crumbliness of the products was an important driver of consumer preference. Interestingly, three out of four attributes responsible for the negative scores were related to texture. Our results are in line with previous research that found texture to be a critical factor for consumer acceptance of many kinds of food products [45].

In accordance with the sensory data, instrumental data showed chickpea snacks to have the greatest hardness, whereas green pea products were found to be less hard. Sensory and instrumental texture parameters were related to each other with the term "hard" being the most often mentioned when describing the chickpea sample. Differences in chemical composition might account for differences in texture. Specifically, the higher lipid content in chickpea might have favored amylose–lipid complex formation during extrusion, thus limiting starch swelling and gelatinization and accounting for a firmer structure. The addition of different percentages of legume bran led to an increase in hardness values. These results are corroborated by evidence reporting that the integration of fiber- and protein-rich plant by-products generally results in dense, hard extrudates due to several factors. Apart from starch dilution, fiber can interrupt the starch matrix and disrupt the bubble cells, leading to poorer texture (i.e., great hardness) [34]. Moreover, both proteins and fiber may compete with starch for free water, thus decreasing the occurrence of starch gelatinization. Our results suggest that reformulating snacks with 15% of legume bran will have no effect in limiting starch gelatinization, leading to products with textural features similar to those of rice snacks.

The type of bran seems to play a role only at high enrichment levels (i.e., 30%), with green pea bran and chickpea bran impacting the snack texture in an opposite way. Pea bran—being higher in fiber—could absorb water during processing, limiting its availability for starch gelatinization, thus resulting in a more compact structure with high hardness values. Besides differences in the chemical composition and, eventually, in the structural and functional characteristics of fiber, interactions between rice starch and legume fiber might also be considered.

Although the impact of either legume or plant-food processing by-products has been investigated [34,46], a direct comparison of our data with those found by other researchers is difficult. Indeed, snack features depend on several factors including moisture content, temperature, screw speed, die dimension, and screw profile. Adjusting such processing parameters would enable the creation of a wide variety of extruded food products with different structure–texture properties. Specifically, the snack produced in the present study is a co-extruded snack characterized by an extrusion-cooked outer shell that is later filled with either a savory or sweet filling. To the best of our knowledge, there are no studies dealing with the textural features (measured by instrumental analysis) of this kind of product.

Moving on to the behavioral attitudes that could have a role in the acceptance of high-fiber products, food neophobia data indicated that subjects scoring low to medium for neophobia gave higher liking scores to all samples compared to subjects scoring high. Generally, it is widely reported that neophobic subjects prefer less vegetable-based foods, with high fiber amount, compared with neophilic ones [47]. The present data are in line with a previous finding that demonstrated—in a large sample of children from five different European countries—that subjects more prone to try and eat new/unfamiliar food appreciated more experimental samples enriched in fiber [48]. Accordingly, it is well established that food-neophobic subjects are diffident in trying and buying novel foods, while neophilic ones tend to have a wide and varied diet [49]. No food technology neophobia effect was found regarding sample acceptability, while recent findings showed that most adolescents with a low level of food technology neophobia appreciated a flat bread with mushroom powder rich in ß-glucans compared with a control sample containing only wheat flour [50]. These contrasting results could be associated with the consumer sample involved in the study. Indeed, in the present research, people with knowledge about food science and technology were recruited as subjects, and this should be mentioned as a limitation of the study, whereas the adolescents involved in the previous mentioned research were more naïve consumers.

#### **5. Conclusions**

Snack products with legume flour and bran represent an interesting food formulation for two reasons. From a nutritional standpoint, these products incorporating milling by-products at a high percentage represent an interesting source of fiber, as well as a valuable food alternative in the worldwide increasing demand for high-fiber and gluten-free products. From a sustainable point of view, the exploitation of milling by-products could reduce the environmental impact of this food category. All legume-based products containing bran in our study were accepted by the consumers involved, even if the hedonic scores were rather low. Crumbliness and mild flavor attributes positively influenced hedonic scores, whereas stickiness, dryness, hardness, and to a lesser extent, visual aspect affected them negatively. As future perspectives, it could be interesting to involve a larger sample population to obtain more representative data about consumer responses to pulse snacks. Moreover, it could be useful to compare the present snacks with a savory or sweet filling and to involve a commercially available product type (e.g., rice-based snack with filling) to better understand the acceptance level of the prototype products

**Author Contributions:** Conceptualization, C.P., A.M. and E.P.; Formal analysis, C.P. and A.B.; Funding acquisition, A.M.; Investigation, C.P. and A.B.; Supervision, A.M. and E.P.; Writing—original draft, C.P., A.B. and A.M.; Writing—review & editing, C.P., A.B., A.M. and E.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was partially supported by the Regione Piemonte (POR FESR 2014–2020), as a part of the "Prodotti e processi innovative per la produzione di estrusi e pasta gluten free" EXFREE Project.

**Acknowledgments:** The authors would like to thank Massimo Blandino (Università degli Studi di Torino, Italy) for determining the chemical composition of raw materials and coordinating sample preparation. The authors would like also to thanks students and collaborators for their help in the experimental phases.

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

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 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* **Consumer Acceptance of Brown and White Rice Varieties**

**Tanweer Aslam Gondal 1,2 , Russell S. J. Keast 1 , Robert A. Shellie 1 , Snehal R. Jadhav 1 , Shirani Gamlath 1 , Mohammadreza Mohebbi <sup>3</sup> and Djin Gie Liem 1, \***


<sup>3</sup> Biostatistics Unit, Faculty of Health, Deakin University, Geelong, VIC 3125, Australia; m.mohebbi@deakin.edu.au

**\*** Correspondence: gie.liem@deakin.edu.au

**Abstract:** Rice is consumed as a staple food by more than half of the world's population. Due to a higher fibre and micronutrient content, brown rice is more nutritious than white rice, but the consumption of brown rice is significantly lower than that of white rice, primarily due to sensory attributes. Therefore, the present research aimed to identify the sensory attributes which drive liking of Australian-grown brown and white rice varieties. Participants (*n* = 139) tasted and scored (9-point hedonic scale) their liking (i.e., overall liking, aroma, colour and texture) of brown and white rice types of Jasmine (Kyeema), Low GI (Doongara), and Medium grain rice (Amaroo). In addition, participants scored aroma, colour, hardness, fluffiness, stickiness, and chewiness, on Just About Right Scales. A within-subjects crossover design with randomised order (William's Latin Square design) was used with six repeated samples for liking and Just About Right scales. Penalty analyses were applied to determine the relative influence of perception of sensory attributes on consumer liking of the rice varieties. Across all varieties, white rice was liked more than brown rice due to the texture and colour, and Jasmine rice was preferred over Low GI and Medium Grain. Rice texture (hardness and chewiness) was the most important sensory attribute among all rice varieties and aroma was important for driving of liking between white rice varieties.

**Keywords:** brown rice; white rice; sensory; consumer acceptance; Just About Right scale; JAR; penalty analysis

#### **1. Introduction**

Rice is consumed as a staple food by more than 4 billion people around the globe [1–3]. Rice is a significant source of dietary nutrients such as carbohydrates, vitamins, and minerals [4,5]. For populations that rely on rice as a staple food, it delivers approximately 21% of the consumed energy and 15% of the consumed protein [6].

Australia produces high quality rice from different varieties, which are categorised as aromatic Thai jasmine origin and non-aromatic rice [7]. Aromatic rice varieties have distinctive popcorn like flavour notes due to the presence of 2-acetyl-1-pyrroline [7–10]. Furthermore, rice can be classified based on the milling process. The milling of the whole grain results in brown rice, and a further removal of bran and germ results in white rice. [11]. Although white rice is more commonly consumed, brown rice is considered healthier due to nutritional components such as lipids, proteins, dietary fibre, and polyphenols [12,13].

The sensory profile of rice is an important driver of consumer acceptance. Sensory attributes have a strong influence on product selection, consumption, and purchase decisions [14,15]. Sensory attributes such as physical appearance (i.e., uniformity, cleanliness, brightness, glossiness and translucency of the rice grain) [16], taste (e.g., sweetness, bitter-

**Citation:** Gondal, T.A.; Keast, R.S.J.; Shellie, R.A.; Jadhav, S.R.; Gamlath, S.; Mohebbi, M.; Liem, D.G. Consumer Acceptance of Brown and White Rice Varieties. *Foods* **2021**, *10*, 1950. https://doi.org/10.3390/ foods10081950

Academic Editor: Antti Knaapila

Received: 7 July 2021 Accepted: 19 August 2021 Published: 22 August 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 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 (https:// creativecommons.org/licenses/by/ 4.0/).

ness), and aroma (e.g., floral notes) are drivers of liking [17] that affect consumer acceptance of rice.

Furthermore, rice texture (i.e., cohesiveness, softness) has been suggested to be of high importance for consumer acceptance of rice. A previous study reported that brown rice texture was less liked compared to white rice and there was variation in liking of the various textures of different brown rice varieties [18]. Along the same lines, Suwansri et al. suggested that an increase in the hardness of rice is associated with a lower consumer acceptability [19]. The importance of texture has also been emphasised by Maleki et al., who suggests that consumers can be segmented based on their preference for different rice textures [20]. In their study, fluffiness was a driver of liking for the majority of consumers (44%), whereas for smaller segments of consumers, liking was mainly driven by flavour attributes.

Within each rice variety, the milling process (e.g., white vs. brown rice) alters the nutrient composition and sensory attributes [21]. For example, brown rice has a higher lipid content compared with white rice. The lipid context affects the sensory profile due to lipid oxidation in the bran layer of brown rice [22]. Lipid oxidation leads to the development of off flavours [23], which potentially impact consumer perception and acceptance. In short, differences in the acceptance of white and brown rice are likely caused by differences in sensory profiles, which are related to differences in nutrient composition [24].

In Australia, 90% of rice is consumed as white rice, whereas only 10% is consumed as brown rice [25], which is similar to global rice consumption patterns [25,26]. Brown rice is considered a healthier option than white rice [27]. To understand what drives the difference in consumption of brown and white rice, it is important to investigate the sensory differences of brown and white rice.

The objective of this study was to identify the drivers of liking of Australian grown brown and white rice varieties. It will provide important information for rice industry and breeding programmes for the development of new rice varieties to meet consumer needs.

#### **2. Participants, Materials, and Methods**

#### *2.1. Study Design*

A within-subjects crossover design with randomised order (William's Latin Square design) for liking and Just About Right scales with six repeated samples was used in the present study. To determine the required participant sample size, G\*power [Version 3.1.9.2, Franz Faul, Universitat Kiel, Kiel, Germany] was used. Based on six measurements (six rice samples) comparisons within subjects with alpha level 0.05, power of 0.8, and a small effect size (f = 0.10), the minimum sample size was 109. To account for potential dropouts, 140 participants from Consumer Analytical Safety Sensory (CASS) Food Research Centre database were recruited. Participants were excluded if they had food allergies, dietary restrictions, and/or were pregnant or lactating. Participants were asked to refrain from eating, drinking, or brushing their teeth one hour prior to testing. The rice consumer study was approved by the research ethics committee Deakin University (HEAG-H 29\_2018).

#### *2.2. Measurements*

Participants were asked to complete two questionnaires concerning (1) demographics (age, gender, education, and marital status), and (2) rice consumption (type of rice (brown or white), number of times they eat rice daily, weekly or fortnightly, and awareness of brown rice health benefits). To assess the liking and sensory perception of the rice samples before and after tasting the rice samples, participants filled out 9-point hedonic scales (1 = extremely dislike and 9 = extremely like) [28] for overall liking, aroma, colour, and texture. In addition, participants completed Just About Right scales for aroma intensity, colour, hardness, fluffiness, stickiness, and chewiness, similar to previous published research [29]. A Just About Right scale, is a bipolar labelled attribute scale [30], which has an anchored mid-point that corresponded to Just About Right for each attribute [31]. The Just

About Right scales provided the participants with 3 answer options per sensory attribute (1 = not enough, 2 = Just about Right, 3 = too much) [32].

#### *2.3. Materials*

Three most commonly consumed Australian rice varieties (Jasmine rice (Kyeema), Low GI (Doongara) and Medium grain (Amaroo) (Table 1)) with both brown and white rice types were sourced from Sunrice (Ricegrowers Ltd., Leeton, Australia) Australia [33].

**Table 1.** Selected Australian rice varieties.


Rice samples were washed 2 to 3 times in cold running water until the water ran clear. Rice samples were cooked in dedicated rice cookers ("Grain Master" HD4514/72\_ UM\_ US\_v1.0, Philips, China), to avoid cross flavour contamination, according to manufacturer's instructions with specific water to rice ratios (Table 1). Rice samples and water quantities were measured by a measuring cup. Rice was cooked at quick rice cooking mode and kept warm at 600 C (as measured by an infrared thermometer Xintest HT-88A; Dongguan Xintai Instrument Co., Guangdong, China) in the rice cooker for no longer than the duration of the sensory test (approximately 45 min).

#### *2.4. Testing Procedure*

Sensory testing took place in a sensory laboratory, which consisted of partitioned booths and a high capacity air filtration system, of the CASS Food Research Centre, Deakin University, Melbourne, Australia. On arrival, participants were instructed to carefully read the Plain Language Statement and sign the consent form. Ten participants participated in each one hour session. Rice samples were served to the participants in 30 mL clear plastic medicine cups that were labelled with three digit unique codes. Each cup contained 10 g of rice and participants were instructed to consume at least one teaspoon of rice. The rice samples were randomly presented one at a time directly from the rice cooker at a temperature of 55 ± 3 ◦C. The participants were instructed to rinse their mouth with filtered water for five seconds and use crackers between tasting the different rice samples.

The test consisted of two parts (i.e., before tasting, after tasting). In the first part, the participants received the following instruction: "do not eat the rice samples, only look, feel (e.g., hold the rice between your fingers) and smell the rice". Next, participants were asked to rate overall liking and their liking for aroma and colour on a 9-point hedonic scale, and fluffiness, stickiness, hardness, and aroma intensity on Just About Right Scales.

In the second part, the participants were instructed to taste the rice samples (one by one) and rate on 9-point hedonic scales, their overall liking, and texture for each rice sample. In addition, participants rated their perceived intensity of flavour, fluffiness, hardness and chewiness on Just About Right scales. There was a one minute break after the tasting of each sample to avoid tasting fatigue of the participants.

The data were collected on computers using Compusense Software Academic Consortium (Compusense, Inc., Guelph, ON, Canada). Gift vouchers (50AUD) were served to each participant on completion of the rice consumer test.

#### *2.5. Statistical Analysis*

All rice consumer study data were exported from Compusense Cloud into Microsoft Excel version 1708 (Microsoft Corporation) for data cleaning. For the statistical analysis of

liking, the program Stata/IC 15.0 (StataCorp LLC, 4905 Lakeway Drive, College Station, TX 77845, USA) was used. Descriptive statistics (mean, standard deviation and correlation coefficient) were calculated for overall liking scores and all sensory attributes. Box plots and scatter plots were extracted for overall liking and for other sensory attributes. Linear mixed model approach was used to analyse repeated measure Analysis of Variance (ANOVA) data to determine the effect of rice varieties (Jasmine, Low GI and medium grain rice samples) and rice types (brown, white) on overall liking, aroma, colour, and texture linking. This approach accounts for within subject autocorrelation via a random intercept in the model. The combined effect of rice varieties and types of rice was tested through a model that contained the main effects of rice type (brown and white) and varieties (Jasmine, Low GI and Medium Grain) as well as the two-way interaction between varieties, and types of rice. The post-hoc pairwise comparison (Bonferroni adjusted) was conducted to identify the significant difference in sensory attributes among rice varieties and rice types.

The descriptive statistics for Just About Right attributes, overall liking, and penalty analysis (*p* < 0.05) of brown and white rice from the three varieties were conducted in XLSTAT Sensory version 2020.3 (Addinsoft, New York, NY, USA). The penalty was a weighted difference between means (mean liking of JAR category minus the mean of liking for other two levels (too low and too high) taken together) [32]. Spearman's correlation coefficients were calculated. Mean drop plots were extracted to identify the effect of JAR attributes on overall liking of rice. The mean drops were plotted against the percentage of consumers. For penalty analysis and mean drop plots, 20% consumers were considered as the threshold level for each JAR attribute [30].

#### **3. Results**

#### *3.1. Participants*

The participants (*n* = 140, female 52%, male 48%) from different age groups participated in the consumer study, one participant was excluded during data cleaning because of incomplete rice tasting session. The participants were rice consumers and mostly thought they were aware of the health benefits of brown rice. The demographics are shown in Table 2.



#### *3.2. Liking (9-Point Hedonic Scale) of Brown and White Rice Varieties before Tasting*

In the result section, rice variety refers to the different varieties which were tested (i.e., Jasmine white, Jasmine brown, Low GI white, Low GI brown, Medium grain white, and Medium grain brown) and rice type refers to brown and white rice. The results (Table 3) indicate that there was a main effect of rice varieties and their types (i.e., brown vs. white) on overall liking before tasting the rice samples (*p* < 0.05). However, there was no statistically significant interaction between rice variety (i.e., Jasmine, Low GI, Medium Grain) and rice type. This means that white rice was preferred over brown rice, regardless of the rice variety (*p* < 0.05) (see Figure 1). Pairwise comparisons show that Jasmine white rice was more liked than any of the other rice varieties (*p* < 0.05), while liking of Low

GI white and Medium grain was not statistically significantly different (represented with shared letter "C"). Likewise, no difference was observed between the overall liking of brown rice varieties.


**Table 3.** Linear mix model (repeated measures) ANOVA table for brown and white rice varieties before tasting.

<sup>1</sup> Statistically significant (*p* < 0.001) from the reference (ref). <sup>2</sup> rice variety with different letters are statistically significant different (*p* < 0.05).

> For aroma liking, there was a significant difference (Table 3) between rice varieties and their types (i.e., brown vs. white) before tasting the rice samples. The differences in mean values of Low GI and Medium Grain were −0.6, 95% CI (−0.8, −0.5) and −0.7, 95% CI (−0.9, −0.5), respectively, when compared with Jasmine rice (a reference sample). The interaction between rice variety (i.e., Jasmine, Low GI, Medium Grain) and rice type was also statistically significant, meaning that the aroma of white rice was preferred over brown rice, regardless of the rice variety (*p* < 0.05) (see Figure 1). Pairwise comparisons show that the aroma of Jasmine white rice was more liked than any of the other rice varieties (i.e., Jasmine white rice has the highest mean 7.0, 95% CI (6.9, 7.3) and Medium Grain brown has lowest mean 5.8, 95% CI (5.5, 6.1)). On the other hand, liking of Low GI white and Medium grain was not statistically significantly different (represented with shared letter "AB"). Similarly, no difference was observed between the aroma liking of Low GI and Medium Grain brown rice varieties.

> The rice varieties and their types (i.e., brown vs. white) were significantly associated with colour liking before tasting the rice samples. The differences in mean values of Low GI and Medium Grain for colour liking were −0.4, 95% CI (−0.5, −0.2) and −0.1, 95% CI (−0.3, −0.04) respectively when compared with Jasmine rice (a reference sample). However, the interaction between rice variety (i.e., Jasmine, Low GI, Medium Grain) and rice type was not statistically significant. That is, the colour of white rice was liked more than the colour of brown rice, regardless of the rice variety (see Figure 1). Pairwise comparisons show that there was no difference in colour liking of Jasmine white and Medium Grain white rice (represented with shared letter "C"). Likewise, no difference was observed in colour liking of Jasmine brown, Low GI brown and Medium Grain brown (represented with shared letter "A").

**Figure 1.** Mean liking (9-point hedonic scale, 1 = extremely disliked to 9 = extremely liked) of sensory attributes for rice varieties. \* Different letters, shown as A–D, within attribute are statistically significantly different (*p* < 0.05).

#### Liking (9-Point Hedonic Scale) of Brown and White Rice Varieties after Tasting

*≤ ≤* The results show rice variety and rice type (i.e., brown and white) after tasting significantly affect liking (see Table 4). Jasmine rice was liked more than Low GI and Medium Grain rice. For all rice varieties, white rice was preferred over brown rice (mean difference = 0.8, 95% CI (0.6, 1.1). The significant interaction between rice varieties and rice types (i.e., brown and white) shows that Jasmine white rice was liked more than any of the other brown and white rice varieties (see Figure 1). The pairwise comparisons show that no difference was observed between Low GI white rice, Medium Grain white rice and Jasmine brown rice in overall liking after tasting.

*≤ ≤* There was a significant correlation between rice variety and rice type on texture liking (*p* < 0.05) after tasting rice samples (Table 4). This means that the texture of Jasmine rice was liked more than the texture of Low GI and Medium Grain. The mean liking of Low GI and Medium Grain rice were reduced by −0.6, 95% CI (−0.8, −0.4) and −0.4, 95% CI (−0.7, −0.2), respectively, when compared with Jasmine rice (a reference sample). Likewise, the texture of white rice was preferred over brown rice, regardless of rice varieties (mean difference = 0.91, 95% CI (0.6, 1.2). The significant interaction between rice varieties and rice types also indicates that the texture of Jasmine white rice was liked more than the texture of any of the other brown and white rice varieties (see Figure 1). The pairwise comparisons show that no difference was observed between Low GI white rice, Medium Grain white rice, and Jasmine brown rice in texture liking after tasting. However, the texture liking of brown rice varieties was not statistically different.


**Table 4.** Mix model (repeated measures) ANOVA table for brown and white rice varieties after tasting.

1 statistically significant (*p* < 0.001) from the reference (ref). <sup>2</sup> rice variety with different letters are statistically significant different (*p* < 0.05).

#### *3.3. Just About Right Attributes and Penalty Analysis*

3.3.1. Penalty Analysis of Jasmine Brown and White Rice before Tasting

Penalty analysis shown in Table 5 indicates that the overall penalty is significant (*p* < 0.05) for all attributes of Jasmine brown rice. This means that the rice was not perceived at optimum level for all attributes tested. The Jasmine brown rice was rated as being too low in aroma, too dark in colour, too hard in texture, too low in fluffiness, and/or too low in stickiness. For Jasmine white rice, the overall penalty (Table 5) was not significant for any of the attributes. This means that across all tested attributes, a deviation from JAR did not have a significant influence on overall liking. The mean drop plot against consumers for each attribute of Jasmine brown and white rice is shown in Figure 2A,B, which visually represents the results of the penalty analysis.


**Table 5.** The Penalty analysis and JAR variables (before tasting) for Jasmine brown and white rice.

<sup>a</sup> The impact of JAR variables for Jasmine brown and white rice on the overall liking (Spearman's correlation coefficient with a significance level α = 0.05). The correlation coefficients (between JAR attributes and overall liking) show how much JAR attributes have impacted ("low" or "high") on overall liking for rice samples [34]. When the correlation is positive, the "too little" has a bigger impact than the "too much", and vice versa for the negative correlations. If correlation is "0" for a JAR attribute, then that attribute would have a strong impact on overall liking [35]. <sup>b</sup> Selection % is the percentage of consumers who rate the rice as too low, JAR, or too high on a given attribute. <sup>c</sup> Mean is the mean overall liking (9-point hedonic scale) of consumers who rated a given attribute as too low, JAR, or too high. <sup>d</sup> Mean drop is the decrease in liking compared to the mean liking of those who rated the attribute as JAR. <sup>e</sup> Penalty is a weighted difference between means (mean liking of JAR category minus the mean of liking for other two levels (too low and too high) taken together). \* *p* ≤ 0.001, \*\* *p* ≤ 0.05. ≤ ≤

**Figure 2.** Mean drop plots for Jasmine rice variety before tasting (**A**) Jasmine Brown, and (**B**) Jasmine White. \* Mean drop is the decrease in liking compared to the mean liking of those who rated the attribute as JAR. \*\* Consumer % are the consumers which judged an attribute as not ideal (Just About Right). The attributes with large percentages of consumers and penalties are in top right quadrant of the plot, which illustrates the critical points of the product [34].

#### 3.3.2. Penalty Analysis of Low GI Brown and White Rice before Tasting

The overall penalty analysis for Low GI brown rice was significant (*p* < 0.05) for all attributes except "hardness" (*p* = 0.18) (see Table 6). This means that the hardness of Low GI brown rice was the only attribute which was rated as being optimal. The penalty analysis (Table 6) showed that the overall penalty for Low GI white rice was significant for fluffiness (*p* < 0.05). This means that the rating of liking was significantly negatively influenced when participant rated Low GI white as low in fluffiness. Specific changes in liking due to suboptimal attributes are shown in Figure 3A,B which visually represents the penalty analysis of Low GI rice.


**Table 6.** The Penalty analysis and JAR variables (before tasting) for Low GI brown and white rice.

<sup>a</sup> The impact of JAR variables for Jasmine brown and white rice on the overall liking (Spearman's correlation coefficient with a significance level α = 0.05). The correlation coefficients (between JAR attributes and overall liking) show how much JAR attributes have impacted ("low" or "high") on overall liking for rice samples. When the correlation is positive, the "too little" has a bigger impact than the "too much", and vice versa for the negative correlations. If correlation is "0" for a JAR attribute, then that attribute would have a strong impact on overall liking [35]. <sup>b</sup> Selection % is the percentage of consumers who rate the rice as too low, JAR, or too high on a given attribute. <sup>c</sup> Mean is the mean overall liking (9-point hedonic scale) of consumers who rated a given attribute as too low, JAR, or too high. <sup>d</sup> Mean drop is the decrease in liking compared to the mean liking of those who rated the attribute as JAR. <sup>e</sup> Penalty is a weighted difference between means (mean liking of JAR category minus the mean of liking for other two levels (too low and too high) taken together). \* *p* ≤ 0.001, \*\* *p* ≤ 0.05.

**Figure 3.** Mean drop plots for Low GI rice variety before tasting (**A**) Low GI Brown and (**B**) Low GI White. \* Mean drop is the decrease in liking compared to the mean liking of those who rated the attribute as JAR. \*\* Consumer % are the consumers which judged an attribute as not ideal (Just About Right). The attributes with large percentages of consumers and penalties are in top right quadrant of the plot, which illustrates the critical points of the product [34]. ≤ ≤

#### 3.3.3. Penalty Analysis of Medium Grain Brown and White Rice before Tasting

− − The results of the penalty analysis (Table 7) for Medium Grain brown rice show that the overall liking was significantly (*p* < 0.05) influenced when the majority of the participants considered that aroma, colour, and hardness were not at optimum level, the attributes were too high in aroma, too dark in colour, and too hard in texture. Similarly, the overall penalty (Table 7) for Medium Grain white rice was significant for fluffiness (*p* = 0.02). That means that for fluffiness, the deviations from the Just about right level have a significant impact on overall liking. The impact on liking of each attribute is shown in Figure 4A,B.

−

−

−

−

**Figure 4.** Mean drop plots for Medium Grain rice variety before tasting (**A**) Medium Grain Brown and (**B**) Medium Grain White. \* Mean drop is the decrease in liking compared to the mean liking of those who rated the attribute as JAR. \*\* Consumer % are the consumers which judged an attribute as not ideal (Just About Right). The attributes with large percentages of consumers and penalties are in top right quadrant of the plot, which illustrates the critical points of the product [34].


**Table 7.** The Penalty analysis and JAR variables (before tasting) for Medium grain brown and Medium grain white rice.

<sup>a</sup> The impact of JAR variables for Jasmine brown and white rice on the overall liking (Spearman's correlation coefficient with a significance level α = 0.05). The correlation coefficients (between JAR attributes and overall liking) show how much JAR attributes have impacted ("low" or "high") on overall liking for rice samples. When the correlation is positive, the "too little" has a bigger impact than the "too much", and vice-versa for the negative correlations. If correlation is "0" for a JAR attribute, then that attribute would have a strong impact on overall liking [35]. <sup>b</sup> Selection % is the percentage of consumers who rate the rice as too low, JAR, or too high on a given attribute. <sup>c</sup> Mean is the mean overall liking (9-point hedonic scale) of consumers who rated a given attribute as too low, JAR, or too high. <sup>d</sup> Mean drop is the decrease in liking compared to the mean liking of those who rated the attribute as JAR. <sup>e</sup> Penalty is a weighted difference between means (mean liking of JAR category minus the mean of liking for other two levels (too low and too high) taken together). \* *p* ≤ 0.001, \*\* *p* ≤ 0.05.

3.3.4. Penalty Analysis of Jasmine Brown and Jasmine White Rice after Tasting

Penalty analysis (Table 8) indicate that the overall liking was significantly (*p* < 0.05) influenced when participants rated the Jasmine brown rice as not being ideal for flavour, fluffiness, hardness, or chewiness. For Jasmine white rice, the overall penalty (Table 8) was only significant for hardness and not significant for all other attributes after rice tasting. This means that most of the participants considered Jasmine white rice "not hard enough" in texture. The mean drop plot against participants for each attribute by tasting of Jasmine brown and Jasmine white rice is shown in Figure 5A,B.


**Table 8.** The Penalty analysis and JAR variables (after tasting) for Jasmine brown and Jasmine white rice.

<sup>a</sup> The impact of JAR variables for Jasmine brown and white rice on the overall liking (Spearman's correlation coefficient with a significance level α = 0.05). The correlation coefficients (between JAR attributes and overall liking) show how much JAR attributes have impacted ("low" or "high") on overall liking for rice samples. When the correlation is positive, the "too little" has a bigger impact than the "too much", and vice-versa for the negative correlations. If correlation is "0" for a JAR attribute, then that attribute would have a strong impact on overall liking [35]. <sup>b</sup> Selection % is the percentage of consumers who rate the rice as too low, JAR, or too high on a given attribute. <sup>c</sup> Mean is the mean overall liking (9-point hedonic scale) of consumers who rated a given attribute as too low, JAR, or too high. <sup>d</sup> Mean drop is the decrease in liking compared to the mean liking of those who rated the attribute as JAR. <sup>e</sup> Penalty is a weighted difference between means (mean liking of JAR category minus the mean of liking for other two levels (too low and too high) taken together). \* *p* ≤ 0.001, \*\* *p* ≤ 0.05.

**Figure 5.** Mean drop plots for Jasmine rice variety after tasting (**A**) Jasmine Brown and (**B**) Jasmine White. \* Mean drop is the decrease in liking compared to the mean liking of those who rated the attribute as JAR. \*\* Consumer % are the consumers which judged an attribute as not ideal (Just About Right). The attributes with large percentages of consumers and penalties are in top right quadrant of the plot, which illustrates the critical points of the product [34].

−

−

−

#### 3.3.5. Penalty Analysis of Low GI Brown and Low GI White Rice after Tasting

The penalty analysis (Table 9) of Low GI brown rice by tasting shows that the overall liking was significantly (*p* < 0.05) influenced when most of the participants judged that flavour, hardness, and fluffiness were not optimal in Low GI brown rice. For Low GI white rice, the overall penalty (Table 9) was significant (*p* < 0.05) for all attributes tested. This means that the overall liking was significantly influenced, when the majority of the participants rated Low GI white rice as being not ideal for flavour, fluffiness, hardness, or chewiness. The influence on liking of sensory attributes is shown in Figure 6A,B.

#### 3.3.6. Penalty Analysis of Medium Grain Brown and White Rice after Tasting

For Medium Grain brown rice, the penalty analysis (Table 10) showed that the overall liking of Medium Grain brown rice was significantly (*p* < 0.05) influenced when participant rated flavour, fluffiness, and hardness were not at optimum level. Similarly, the overall penalty of Medium Grain white rice was significant for flavour intensity, fluffiness, and chewiness. This means that significant participants perceived Medium Grain white as too low in flavour and fluffiness, and too high in chewiness. The mean drop plots against participants for each attribute of Medium Grain brown rice shown in Figure 7A,B.


**Table 9.** The Penalty analysis and JAR variables (after tasting) for Low GI brown and Low GI white rice.

<sup>a</sup> The impact of JAR variables for Jasmine brown and white rice on the overall liking (Spearman's correlation coefficient with a significance level α = 0.05). The correlation coefficients (between JAR attributes and overall liking) show how much JAR attributes have impacted ("low" or "high") on overall liking for rice samples. When the correlation is positive, the "too little" has a bigger impact than the "too much", and vice versa for the negative correlations. If correlation is "0" for a JAR attribute, then that attribute would have a strong impact on overall liking [35]. <sup>b</sup> Selection % is the percentage of consumers who rate the rice as too low, JAR, or too high on a given attribute. <sup>c</sup> Mean is the mean overall liking (9-point hedonic scale) of consumers who rated a given attribute as too low, JAR, or too high. <sup>d</sup> Mean drop is the decrease in liking compared to the mean liking of those who rated the attribute as JAR. <sup>e</sup> Penalty is a weighted difference between means (mean liking of JAR category minus the mean of liking for other two levels (too low and too high) taken together). \* *p* ≤ 0.001, \*\* *p* ≤ 0.05.

≤

−

α

−

≤

**Figure 6.** Mean drop plots for Low GI rice variety after tasting (**A**) Low GI Brown and (**B**) Low GI White. \* Mean drop is the decrease in liking compared to the mean liking of those who rated the attribute as JAR. \*\* Consumer % are the consumers which judged an attribute as not ideal (Just About Right). The attributes with large percentages of consumers and penalties are in top right quadrant of the plot, which illustrates the critical points of the product [34].



<sup>a</sup> The impact of JAR variables for Jasmine brown and white rice on the overall liking (Spearman's correlation coefficient with a significance level α = 0.05). The correlation coefficients (between JAR attributes and overall liking) show how much JAR attributes have impacted ("low" or "high") on overall liking for rice samples. When the correlation is positive, the "too little" has a bigger impact than the "too much", and vice versa for the negative correlations. If correlation is "0" for a JAR attribute, then that attribute would have a strong impact on overall liking [35]. <sup>b</sup> Selection % is the percentage of consumers who rate the rice as too low, JAR, or too high on a given attribute. <sup>c</sup> Mean is the mean overall liking (9-point hedonic scale) of consumers who rated a given attribute as too low, JAR, or too high. <sup>d</sup> Mean drop is the decrease in liking compared to the mean liking of those who rated the attribute as JAR. <sup>e</sup> Penalty is a weighted difference between means (mean liking of JAR category minus the mean of liking for other two levels (too low and too high) taken together). \* *p* ≤ 0.001, \*\* *p* ≤ 0.05.

**Figure 7.** Mean drop plots for Medium Grain rice variety after tasting (**A**) Medium Grain Brown and (**B**) Medium Grain White. \* Mean drop is the decrease in liking compared to the mean liking of those who rated the attribute as JAR. \*\* Consumer % are the consumers which judged an attribute as not ideal (Just About Right). The attributes with large percentages of consumers and penalties are in top right quadrant of the plot, which illustrates the critical points of the product [34].

#### **4. Discussion**

This study aimed to identify the consumer liking, sensory attributes, and drivers of liking of brown and white rice varieties. The results suggest that, overall, participants liked Jasmine rice varieties more than Low GI and Medium grain rice varieties. This was also reflected in a higher liking of the aroma, colour, and texture of Jasmine rice, compared to Low GI and Medium grain rice varieties. However, white rice was preferred over brown rice regardless of rice varieties.

The present study suggests, in line with previous studies [19,24,36,37], that texture, colour, and aroma are important drivers of consumer liking for rice. However, these drivers of liking do not seem to equally explain the differences in liking of white and brown rice. Indeed, differences in aroma mainly explain the difference in liking for white rice varieties and the aroma of Jasmine white rice was liked more than any of the other rice varieties. The most liked white rice (Jasmine rice), contains more of the compound 2-acetyle-1 pyrroline [10] which is known to elicit a distinctive popcorn/pandan aroma [3,38–40] that has a strong impact on consumer acceptance of rice [41]. On the other hand, the other white rice (non-fragrant) varieties contain less 2AP [42–44] that may have an impact on liking of non-fragrant white rice varieties. This is also reflected in the sensory data of the present study that aroma of Jasmine white rice is an important sensory attribute in predicting consumer liking and acceptance of white rice varieties. Therefore, the aroma of Jasmine white rice was preferred over all other white and brown rice varieties. In contrast to aroma being able to explain liking differences for white rice varieties, aroma does not fully explain differences in liking for brown rice.

Differences between brown rice varieties can be explained by texture (hardness and chewiness). This means that brown rice is considered as too hard and chewy in texture, which is driving the difference between brown rice varieties, whereas Jasmine brown rice was preferred over Low GI and Medium grain brown rice. The results are in line with a previous study conducted on ready-to-eat rice in Korea which concluded that the brown rice was scored less in overall acceptability due to being high in hardness, chewiness, and yellowness [18]. Brown rice hardness in texture is associated with dietary fibre that is present in bran layer [45] whereas, in white rice, polishing removes bran and germ during rice processing [46]. This significantly improves texture liking and consumer acceptance of

white rice. In contrast to previous studies, which used a combination of descriptive analysis and hedonic scaling [16,18–20], the current study investigated consumer acceptance of rice by utilising 9-Point hedonic scales, JAR scales, and penalty analysis. Penalty analysis is a powerful tool to analyse the decreases in acceptability associated with sensory attributes which are perceived by consumers as being not optional [47,48]. This study also compared a range of brown and white rice varieties which enabled to compare brown and white rice, but also identify the drivers of liking between brown rice varieties as well as the drivers of liking within white. In addition, it is interesting to note that rice texture (hardness) is more important for the consumer acceptance and overall liking of Australian brown rice varieties. This study suggests that the decrease in hardness and chewiness will increase the overall liking of Australian brown rice varieties, which can eventually increase brown rice acceptance and consumption.

Brown rice texture (hardness and chewiness) and colour are the sensory attributes that are driving the difference between white and brown rice varieties. Thus, the texture of brown rice is less liked as compare to white rice regardless of rice varieties, because the majority of participants rated brown rice varieties as too hard and too chewy. However, differences in texture seem to be more important when comparing liking between white and brown rice. This is in line with a study conducted on consumer acceptance of parboiled brown and white rice which reported that white rice was preferred to brown rice because of texture and colour [24]. The results are also in agreement with the study that reported consumer acceptance of white rice varieties in Thailand, in which the participants preferred cooked white rice because of the soft texture [36]. Suwansri and Meullenet (2004) reported that Asian consumers preferred rice with white appearance (colour) and less sticky texture [49]. Similarly, the consumers from South Asia and Middle East did not prefer the brown rice texture [50]. In the present study, the sensory results also suggest that brown rice texture (hardness and chewiness) is the most important sensory attribute that is driving the liking and consumer acceptance of brown rice.

Although this was the first study which investigated consumer acceptance of Australian brown and white rice varieties, there are some limitations which need to be taken into consideration. The participants were mainly living in urban areas and were well educated, with 79% of participants holding undergraduate degree or higher. That may have affected their liking because of their awareness of the brown and white rice varieties which may cause bias in evaluation of rice attributes. For future investigation, the sample (participants) could be recruited from different geographical areas to predict the preference of Australian brown and white rice varieties. It is suggested to conduct future studies with a greater focus on the texture attributes of brown rice. To identify the variability in the texture of brown rice, different cooking methods and water to rice ratios are recommended. In addition, the instrumental analysis (colour and texture analyser) can be considered for the better understanding of texture attributes of brown and white rice varieties.

#### **5. Conclusions**

Texture is the most important sensory attribute which explains the difference in liking between brown and white rice, whereas differences in aroma best explain the variation in liking of white rice. Therefore, to increase the acceptance and consumption of brown rice, development needs to mainly focus on the improvement of the texture acceptance of brown rice. Future research is needed to investigate if an increased water absorption, milling process, packaging, and storage of brown rice can positively improve the texture and subsequently increase consumer acceptance.

**Author Contributions:** T.A.G.: Conceptualization, Methodology, Writing, and original draft preparation. R.S.J.K.: Conceptualization, Supervision, Reviewing and Editing. R.A.S.: Writing—Reviewing and Editing. S.R.J.: Supervision, Writing—Reviewing and Editing. S.G.: Supervision, reviewing and editing. M.M.: statistical analysis. D.G.L.: Conceptualisation, Methodology, Supervision, Writing— Reviewing and Editing. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded by Bahauddin Zakariya University Multan, Pakistan and Deakin University Melbourne, Australia.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board (or Ethics Committee) of Deakin University (HEAG-H29\_2018, date of approval 28 February 2018).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

**Conflicts of Interest:** The authors declare that they have no competing interests.

**Ethics Approval and Consent to Participate:** The rice consumer study was approved by the research ethics committee Deakin University (HEAG-H 29\_2018). The participants were asked to read Plain Language Statement (PLS) and all participants signed their consent forms.

#### **Abbreviations**


#### **References**


#### *Article*
