Consumer interest in hummus made from different pulses Effects of information about origin and variety seeking tendency

Official dietary guidelines emphasize the importance of consuming pulses due to the advantageous nutritional composition and based on aspects of public health and sustainable production. Along with a global trend of increased focus on sustainability, launching locally produced plant-based food products have thus gained commercial interest. The purpose of this study is to investigate various pulses and how information about their origin affects the willingness to try and willingness to pay for hummus made from these pulses. In addition, variation seeking tendencies and attitude towards the environment were measured. The research design is a quantitative study including data obtained via online surveys. Respondents, (N = 363 Danish speaking and resident consumers) were presented with images of one traditional chickpea and four experimental versions of pulse-based hummus. Participants were presented with representative photos of the hummus samples and respective pulses and randomly selected to be given information regarding origin of the pulses (n = 164) or presented blind (n = 199). To investigate modulation of variation seeking tendency and individual attitude towards the environment, the VARSEEK scale and EAI-24 scale were applied respectively. Results suggest that providing information regarding origin has a positive effect on willingness to try and willingness to pay. Increase in variety seeking tendency significantly increased both willingness to try, willingness to pay, and ratings for appetizing appearance. Increased attitude towards the environment showed only a slightly significant increase on appetizing appearance. Furthermore, providing information about the pulses to respondents, particularly the segment with low variation seeking tendency had a significantly positive effect on willingness to try and ratings of appetizing appearance.

Official dietary guidelines emphasize the importance of consuming pulses due to the advantageous nutritional composition and based on aspects of public health and sustainable production. Along with a global trend of increased focus on sustainability, launching locally produced plant-based food products have thus gained commercial interest. The purpose of this study is to investigate various pulses and how information about their origin affects the willingness to try and willingness to pay for hummus made from these pulses. In addition, variation seeking tendencies and attitude towards the environment were measured. The research design is a quantitative study including data obtained via online surveys. Respondents, (N = 363 Danish speaking and resident consumers) were presented with images of one traditional chickpea and four experimental versions of pulse-based hummus. Participants were presented with representative photos of the hummus samples and respective pulses and randomly selected to be given information regarding origin of the pulses (n = 164) or presented blind (n = 199). To investigate modulation of variation seeking tendency and individual attitude towards the environment, the VARSEEK scale and EAI-24 scale were applied respectively. Results suggest that providing information regarding origin has a positive effect on willingness to try and willingness to pay. Increase in variety seeking tendency significantly increased both willingness to try, willingness to pay, and ratings for appetizing appearance. Increased attitude towards the environment showed only a slightly significant increase on appetizing appearance. Furthermore, providing information about the pulses to respondents, particularly the segment with low variation seeking tendency had a significantly positive effect on willingness to try and ratings of appetizing appearance.

Introduction
There is increasing scientific research that proves the nutritional benefits of plant-based diets whether being totally vegan or partial like vegetarian or flexitarian diets (Williams and Patel, 2017). Decreases in meat consumption have proven beneficial in reducing life-style related diseases like cancer, diabetes, and hypertension (e. g. Chi et al., 2016;Cremonini et al., 2016;Van Dijk et al., 2009). Despite reductions of cardiovascular disease related mortality rates, the global health issue remains extensive and encounters more than 2 billion obese and overweight adults (World Health Organization, 2014). Diets low in legumes including pulses have been linked directly related to the annual cases of deaths caused by cardiovascular diseases, with around 534 thousands deaths, and around 11 million disease-adjusted loss of years (DALYs) (Afshin et al., 2019;Foyer et al., 2016). Besides the health issues related to meat consumption, intensive livestock production disrupts the environmental balance by requiring large amounts of water and increasing the emission of greenhouse gasses (GHG). In comparison pulses require approximately 350 L of water per kg. produced pulses whereas beef cattle require approximately 15,400 L of water per kg. produced meat (Hoekstra and Chapagain, 2008;Mekonnen and Hoekstra, 2010). Additionally, the GHG emission per kg of product for beef production for even the most efficient production systems is at least 15-50 times that of pulses (Poore and Nemecek, 2018). Production of pulses is feasible in most climates and cultivation of a wide variety of different legumes and diverse strains within each variety provide a foundation for increasing biodiversity.
The extent of both the health and environmental related problems has become so comprehensive that a mega trend of sustainable awareness has emerged particularly among an aspiring generation Z (Aschemann- Witzel et al., 2020;Faber et al., 2020). A dual focus on personal health and minimal environmental impact has created an extensive consumer demand for sustainable food solutions (Dabija et al., 2019). Among the Danish population, consumption of pulses and legumes alone nearly doubled from approximately 0.8 g/day in 2005 to approximately 1.6 g/day in 2013 supporting the changes in consumer demands for alternative protein sources (Fagt, 2019). The most recent survey of Danish legume consumption reports a mean of 6.5 g/day, although the median is only 2.2 g/day, indicating a large spread in consumption (Mertens et al., 2019). The EAT-Lancet commission is recommending a consumption of legumes, pulses, and soy products of approximately 75 g/day as part of a healthy and sustainable diet (Willett et al., 2019). The official dietary guidelines issued from the Danish government has recently emphasized the importance of pulses in a healthy diet and recommends an intake of approximately 100 g/day as a supplement to a low intake of meat (Fødevarestyrelsen, 2021), so a very large increase in consumption is desirable. Besides being protein-rich and containing large amounts of dietary fiber, pulses compose a wide variety of beneficial micronutrients like vitamin A, E, and C and folate, magnesium, potassium, and iron.
Historically legumes, in particular local varieties of peas and beans, have been a stable in Danish agriculture and diet, but during the 19th century the consumption gradually declined, as potatoes and grains increased (see e.g. the trilogy of the development of Danish food consumption from 1800 to 1914, Hyldtoft, 2012Hyldtoft, , 2016Hyldtoft, , 2019. In addition, in the latter part of 20th century the increase in use of chemical fertilizers lead to a large decline in local production of legumes, with the majority produced for feed, such that in 2008, only 0.33% of the total farmed land was used for peas and beans (Statistikbanken.dk, n.d).
Consequently, there are very few varieties to purchase in todays' supermarkets. The only widely available Danish pulses are yellow peas often prepared and consumed as a traditional stew of the same name while other traditional pulses like grey peas have been reduced to hobby scale productions and almost forgotten (Linding, 2015).
Today traditional dishes from various cultures are being spread rapidly around the globe with mega trends in consumption patterns. One such example is the Middle Eastern side dish hummus made from chickpeas, that has become popular in the US and Europe (Feifer, 2014). In the period from 2014 to 2017, the sales of hummus in Coop, one of the largest food retailers in Denmark has increased by a factor 10 (COOP Analyse, 2017). Assessment of Danish supermarkets offerings in bricks and mortar shops made by the authors at the time starting the experimental work, revealed a comprehensive selection of ready to eat hummus. Availability from the three largest online retail channels sum op to more than 35 different products (Coop Danmark, n.d.; Føtex, n.d.; Nemlig, n.d.). In the development of hummus for the Danish food culture, diverse food trends and cultural influences have contributed to a variety of hummus or hummus like products based on other peas and beans, versions topped with pesto, herbs or chili, and mixed with beetroot, squash, olives, and peppers. Danish food producers have been experimenting with a commercial version of hummus made from yellow peas as an alternative to chickpeas due to the similarity in taste, texture, and color. The overall sensory profile strongly depends on the species, growing conditions, and the post-harvesting process. Additionally, the distinctive flavor of beans originates from lipoxygenase induced hexanol, hexanal, and octenol while the characteristic flavor of peas originates from an aroma compound related to methoxy-isobutyl pyrazine (see e.g. McGEE, 2004). The texture of pulses depends on the seed coat and cotyledon structure and the composition of starch and protein. Also, preservation processes like canning or drying and subsequent cooking methods influence the texture and color (Alvarez et al., 2017). By selecting different types of pulses, there is an opportunity to create a large variety in the sensory properties of hummus, including appearance.
Another contributing factor to the increasing emission of GHG is the world-wide transportation of goods, leading food activists and governments to advocate for locally produced foods (Skallerud and Wien, 2019). Feldmann and Hamm (2015) reviewed the emerging trend of consuming locally produced foods and found it well-supported by several aspects of sustainability, but that the definition of local remains nebulous and depends upon product category in question. No previous research has investigated consumer perception of local pulses. This present study takes into consideration pulses which can all be produced within Europe and some of them are grown in Denmark. Additionally, this study considers consumer preferences and acceptance which highly depends on individual ability to conceptualize, categorize and evaluate the different hummuses from visual assessment (van Trijp and van Kleef, 2008). The accept of a novel food can be supported by using well-considered and appropriately presented sensory attributes (Tuorila and Hartmann, 2020). It is expected that hummus made from lesser known pulses are considered novel by Danish consumers. In the experimental design both unknown, lesser known, and well-known pulses are included.
The origin and the information about the ingredients in a food can play a large role in our acceptance of foods (Deliza, 2018). Even though this type of information may not manifest itself in tangible sensory differences in a food product such as hummus, the information can easily affect both our hedonic and sensory evaluation of the product (Piqueras-Fiszman and Spence, 2015). Mielby and Frøst (2010) found that the relationship between types of descriptions of dishes in a restaurant setting and perception of the dishes is not straightforward. However, the effect on hedonic response was larger for lesser liked dishes. Similarly Fibri and Frøst (2020), demonstrated that for the traditional Indonesian food tempeh, information about the origin of the raw materials affected both Indonesian consumers' acceptance as well as many other aspects of their appraisal of the tempeh. This can be used as a lever to increase acceptance of novel foods, or use of less traditional ingredients in product innovations. Although consumers generally are open towards innovations in traditional food products that highlight innovations that are extrinsic to the product, such as packaging, convenience, and improved shelf life, they may reject innovations that affect the intrinsic (sensory) properties of the product (Vanhonacker et al., 2013). Thus, innovation in established food categories needs to be performed carefully, particularly regarding the intrinsic characteristics of the product.
Connections between visual assessment only and willingness to try (WTT) have been established and further addition of explanations and descriptions positively influences the mental imagery representation (Adaval, 2001;Chaudhuri and Micu, 2014). The present study controls for the individual variation seeking tendency by additionally measuring ratings of appetizing appearance (APP) which can provide perspectives on consumers' conceptualization. In addition to WTT, willingness to pay (WTP) provides further indications of successful marketing in consumer acceptance being a key element in value generating marketing strategies (Breidert et al., 2006).
The present study specifically aims to investigate the relationship between Danish consumers' environmental attitude (measured by EAI-24) and variation seeking tendency (measured by VARSEEK) as psychographic background characterization on one hand, and how these two measures may predict WTT, WTP, and APP towards five hummus samples. Simultaneously, the effects of providing information about origin, history, and common use of the selected pulses was investigated. By this, the effects of the visual appeal and the extrinsic product information about the main raw material of the hummus can be separated.

Pilot study
In a preliminary pilot study eleven pulses were selected based on two main criteria: (1) origin (DK vs. global) and existing information regarding the pulses and (2) their availability. Additionally, the pulses were sorted based on variety, anticipated familiarity, and immediate hummus-like texture, when using a standard recipe. Based on these initial criteria three types of pulse were removed from the set (Brown Ö land beans, White bean and Fava beans). Subsequently, the remaining eight pulses were evaluated via fast sensory profiling for final selection. For the sensory profiling a panel (n = 12) consisting of experts and students within food science was created. Respondents were presented with eight hummus samples based on a standard recipe including the following additional ingredients: Water, sunflower oil, lemon juice, tahin, salt, cumin, garlic powder, and cayenne pepper. Samples were served randomly as 10g at approximately 7 • C in small clear plastic containers with a lid and marked with a three-digit code. Respondents were asked to describe the visual appearance from a list of CATA attributes comprising color, texture, and associational key words. Additionally, the respondents were asked to rate viscosity, mouth coating, acidity, and bitterness on a 7-point scale from 1 (very low) to 7 (very high). Based on the sensory evaluation presented in Table 1, in particular the appearance, the number of samples were reduced to five samples (eliminated samples shown in grey in table). The selected pulses include borlotti beans (produced in China for Urtekram) originating from Italy, carmencita beans (produced at Svanholm Gods, Denmark) originating from Denmark, chickpeas (produced in Italy for Urtekram) originating from Turkey, lollandske rosiner peas (produced at Knuthenlund, Denmark) originating from Denmark, and yellow peas (produced in Denmark for Vestkorn) originating from Denmark.
Representation of the five samples were created by piping 175g hummus into a clear plastic container identical to the packaging from a commercially produced hummus. In addition to the finished hummus, images of the pulse used as the main ingredient in each hummus sample were presented. All photos were taken with an iPhone X. Color and light corrections were done in Photoshop version CC 2015 and Affinity Photo version 1.8.3. Images of hummus and main ingredients are presented in Fig. 1.

Questionnaire design
Two questionnaires were designed to compare the effect of origin information versus no information given hereafter referred to as 'With information' and 'Without information'. Respondents were asked to answer sociodemographic questions regarding gender, age, educational level, family relationship, place of residence, and nationality.
Regarding instruments to measure individual's awareness of sustainability, recent studies conducted a structural confirmation to validate the Environmental Attitude Inventory (EAI) 24 scale in a European country and found a satisfactory predictive power of general attitude (Ajdukovic et al., 2019;Moussaoui et al., 2016). The question battery comprises 12 straightforward statements paired with reversed questions covering subdimensions of preservation and utilization to increase the validity (Gifford and Sussman, 2012;Milfont and Duckitt, 2010). The original version of EAI composed 72 questions requiring a certain amount of time to answer, giving reason for investigating the effect of shorter versions. There is an established connection between WTT and variation seeking tendency. van Trijp and Steenkamp (1992) developed the variation seeking tendency scale (VARSEEK) to provide a short and valid assessment of an individual's variety seeking. In a comparison to the Food Neophobia Scale (Pliner and Hobden, 1992), the VARSEEK scale has been tested just as valid and preferable in measuring the relationship with novel and unfamiliar foods (Lenglet, 2018). The methods are widely accepted and applied in multiple studies investigating aspects of food neophobia.
In assessing the general environmental relationship, both questionnaires included the Environmental Attitude Inventory (EAI) 24-version (Milfont and Duckitt, 2010). To support the aspect of familiarity and overall willingness to try new foods, the Variation Seeking Tendency (VARSEEK) scale was included in both questionnaires (van Trijp and Steenkamp, 1992). All questions were translated from English to Danish via back-translation and reversed questions appear in both scales to ensure reliability.
To substantiate the assumption of hummus being popular on the Danish market, respondents were asked to indicate their uses of pulses with a CATA list of dishes and ways to consume pulses. Table 2 shows the top 10 results from this part of the questionnaire and the full list is shown in appendix 1.
The experimental condition for the evaluation of the samples differed in one central manner. All respondents were presented with a photo of each sample and associated pulse (Fig. 1). Respondents in the 'With information' condition, received information about the origin of the pulses, in addition to the photos, whereas the respondents in the 'Without information' only received the photos. Subsequently to each presented sample, the participants were asked to evaluate how much they would be willing to pay (WTP) for a portion of 175g from the following categories: 10-15 Danish kroner (1.34-2.00 EUR), 15-20 Danish kroner (2.00-2.70 EUR), 20-25 Danish kroner (2.70-3.36 EUR), 25-30 Danish kroner (3.36-4.00 EUR), or more than 30 Danish kroner (4.00 EUR). Due to WTP categories being defined with an equal value span and the last category remaining without limit, data is converted into a 1 (10-15 Danish kroner) to 5 (+30 Danish kroner) scale for analytical simplicity. Participants were also asked to evaluate the (un)appetizing appearance (APP) on a 7-point scale from 1 (very unappetizing) to 7 (very appetizing) and their willingness to try (WTT) also on a 7-point scale from 1 (not at all) to 7 (very much).

Respondent recruitment
Respondents were recruited via the standing consumer panel at the Department of Food Science, University of Copenhagen. The panel was segmented to exclude consumers younger than 18 years. Additional recruitment via the author's personal network was done to ensure a large sample size consisting of a minimum of 100 females and 100 males in total (Saracino et al., 2013). Consumers that did not consume legumes were excluded. There was no specific requirement for hummus consumption frequency, but questions regarding legume consumption was included. The recruitment message instructed respondents to check their clock and note themselves the last minute digit of time. Respondents with equal numbers were to choose the first link and participants with unequal numbers to choose the second link for a random division of respondents. Questionnaire completion time approximately 15-20 min was expected, and participants were given the opportunity to enter their email address after submitting their answers and enter a lottery for two gift certificates to a national department store with a value of 500 Danish kroner (67 EUR).

Ethics approval and GDPR
Data handling complied with the General Data Protection Regulation (GDPR) 2016/679. The Science Ethics Committee of the University of Copenhagen reviewed and approved it (case: 504-0154/20-5000). Informed consent was collected from participants before data collection.

Data analysis
To investigate the effect of EAI and VARSEEK, respondents were segmented into three groups based on their scores in each of the measures. In line with the technique utilized by Frøst (2019, 2020), cut-off points were defined by EAI and VARSEEK scores quartiles, with the lowest quartile classified as low, the middle two quartiles classified as medium, and the highest quartile as high. For EAI the maximum possible range thus spans from 1 x 24 to 7 x 24 = 168. While for VAR-SEEK The maximum possible range spans from 1 x 8 to 7 x 8 = 56. Multiple iterative ANOVA analysis were conducted for WTT, WTP, and APP. Non-significant higher order interaction effects were stepwise eliminated until only significant interactions remained. The outcomes for main, two-way, and higher order interactions are presented in Table 7. Post hoc Fisher's LSD tests were used to analyze comparisons when the omnibus test was significant. For all analyses a significance level of p < 0.05 was used.  Table 2 Top ten popular dishes for use of pulses among respondents.

Respondents
After 10 days of posting the survey, the number of new respondents decreased to a low level. At the time of stopping the survey access, 458 respondents had attempted to answer the survey. Of these, 95 respondents were excluded due to partially complete answers or not fulfilling the requirements of age exceeding 18 years, or never consuming legumes. Of the participants 363 were kept for further analysis. 45.3% respondents completed the questionnaire with information (n with information = 164) and 54.7% respondents completed the questionnaire without information (n without information = 199) about the samples.
The mean age of the sample is 41 years and predominantly characterized by women (72%) and respondents with a medium (28%) to high educational level (49%). Nearly half of the respondents are in a relationship without children living at home (44%) and an additional 29% are single without children living at home. Almost three quarters have residence in the Capital Region (73%) and the 98% of the respondents had Danish citizenship. A full overview of the sample characteristics is represented in Table 3.
A segmentation of the respondents' environmental attitude (EAI) is based on the summed score. The lowest observed summed score is 63 and highest is 134, indicating that none of the respondents only used the extreme scale and to indicate their EAI. The lowest quartile represents the 'low' segment (range 63-96), the two middle quartiles represent the 'medium' segment (range 97-109), and the highest quartile represents the 'high' segment (range 110-134). ANOVA analysis with EAI as dependent variable shows no significant differences between the 'with information'-group and the 'without information'-group (p-value = 0.066). The number of respondents, percentiles, and the respective summed score ranges for each segment is presented in Table 4.
A similar segmentation of the respondents based on the variaty seeking tendency is presented in Table 5. The observed summed scores span the whole possible range from 8 to 56. Segments are based on approximate quartiles, 'low' being the lowest quartile (range 8-38), 'medium' being the two middle quartiles (range 39-51), and 'high' being the highest quartile (range 52-56). ANOVA analysis shows nonsignificant differences between the 'with information'-group and the 'without information'-group (p = 0.584). Number of respondents and the percentiles of the segments are presented in the table. Both lower and upper limit of mean range include extremes at both ends of the 7point scale indicating that some respondents have extremely high food variety seeking, while others absolutely avoid variety seeking. The extremities of the 'medium' segment are concentrated in a narrow range of 38-51 indicating a slightly skewed distribution. Additionally, the number of females and males with and without information provided is presented. In all three VARSEEK segments, males in the without information condition are overrepresented by 50% or more than those with information, leaving only 10 and 9 respondents in the 'low' and 'high' segment. Similarly, for EAI we find that only 12 and 6 males are in the low and high segments (Table 4). We will return to this in the analysis of interaction effects.

Hummus differences
The samples were designed to be visually different, and that the visual differences should convey differences in associated texture properties at an overall level. The results from CATA part of the sensory pilot experiments showed that there were apparent visual differences (Table 1, last column). The results from the consumers show that the visual representation of the samples were successful in conveying these differences. The five hummus samples were rated significantly different for all three measured response variables: Willingness to try (WTT), Willingness to pay (WTP) and Appetizing (APP) appearance. Table 6 reports the mean for the three response variables. Although all three show significant differences between samples, it is evident that there is a distinct difference in order of discriminating power between the variables. Table 7 reports the p-values for all main and interaction effects. Albeit that all variables have a main effect of sample, the effect for WTP is much lower (p = 0.0005), than WTT (p = 10 − 15 ) and APP (p = 10 − 53 ). Appetizing appearance divides the five samples into 4 different groups, where only chickpeas and lollandske rosiner peas are not separated. Willingness to try separates samples into two non-overlapping groups. Willingness to pay also divides the samples in two groups, but in a very narrow range, and with overlap between groups. The three variables are correlated (data not shown in tables), but for WTP, the correlation to the two others are much lower (Pearson's r = 0.311 for WTT and r = 0.334 for APP). The Pearson's r between APP and WTT is 0.758.

Effects of information, gender, VARSEEK and EAI
Multiple iterative ANOVA analysis were conducted for WTT, WTP, and APP. The outcomes for main, two-way, and higher order interactions are presented in Table 7. For the condition of information provided or not, there is a significant effect on WTT (p = 2 10 − 4 ) and a slightly significant effect on WTP (p = 0.026), whereas there was no effect of information on APP (p = 0.721). WTT and WTP mean ratings were significantly higher when information was provided. Effects of gender are only notable for APP (p = 0.035) where females rated 0.2 higher than males. Across the response variables, the VARSEEK segmentation has a larger effect compared to the EAI segmentation. The effect of EAI segmentations on WTT and WTP is non-significant and only slightly significant on APP (p = 0.039). In that particular case, the middle EAI segment rated Appetizing appearance the highest (4.6) and significantly higher than the both low and the high EAI segment (both at 4.4). However, it appears counterintuitive that the middle segments rate higher. We suggest that the results are caused by a smaller sample of respondents (only 36 in the high EAI segment) among those that were presented with information). So, it may reflect spurious results which should be interpreted with care and therefore the figure is not included.
Results show a significant effect of VARSEEK on WTT (p = 10 − 31 ) and APP (p = 10 − 12 ) and a slightly significant effect on WTP (p = 0.04). As expected, effects of VARSEEK segments are correlated with increases in ratings for both WTT, WTP, and APP. Interactions between VARSEEK segments and the information condition has a significant effect on all three variables which follows the same pattern (WTT p = 0.001; WTP p = 0.0002; and APP p = 10 − 7 , respectively). Fig. 2 presents the mean ratings for WTT, WTP and APP grouped by VARSEEK segments. The provided information about the beans and peas increases the responses of the low VARSEEK towards the level of the high VARSEEK segment. The direction of the effect of information is similar across all three response variables presented in Fig. 2. Note however, that this is largest for APP, where there are no significant differences between VARSEEK segments in the informed condition.
Figs. 3 and 4 shows the three-way interaction between gender, information condition and VARSEEK segmentation for WTT and APP. They follow a very similar pattern. The ratings are significantly different and higher for men in the low VARSEEK segment provided with information, though results are based on a small sample size in the VARlow group of men (10 and 23) and should be interpreted with great care. Nevertheless, it demonstrates that some of the two-way interaction effect of VARSEEK and information is indeed caused by the men reacting more strongly to the provided information. Interestingly, providing    information has the opposite effect in the high VARSEEK segment for men, though only significantly different for APP, not for WTT. The remaining significant three-way and higher-order interactions include either information condition or VARSEEK segmentation, of which the significance relies on this relatively small group of men. Interpretations of results should be careful due to the sample size of segmentation categories and further investigation is warranted. When assessing the main effects for APP, both gender and EAI segmentations respectively are slightly significant (p = 0.035 and p = 0.039, respectively), while VARSEEK segmentation is highly significant (p = 10 − 12 ). The two-way interactions between Gender x VARSEEK (p = 10 − 4 ), and Gender x Information condition (p = 10 − 7 ), are also highly significant. Looking at Fig. 4, the 2-way interaction significance is mainly caused by large differences in the low VARSEEK segment for men. Providing information to men in the low segment increases the rating of APP significantly whereas the medium and high segment decreases the rating of APP though not significantly. The additional multipath interactions rely on this small segment of men, and appear spurious and warrants further research before bold conclusions are made.

Conclusions
This paper reports results based on a convenience sample of the Danish population mainly from the Capital region. The respondents have an overrepresentation of females with a higher educational level. They are generally expected to have a different consumption pattern than men. The Capital region is embedded in large socio-material networks which influence human actions and beliefs, intensifying and spreading the awareness of sustainability and therefore complicating the generalization from the sample to other parts of the country (Dillon, 2009;Vihalemm et al., 2015).
Based on the findings, variation seeking tendency and thereby individual interest in new foods, has a greater impact on food appreciation as measured here by WTT, WTP and APP, compared to attitude towards the environment. Studies conducted in Poland, Finland, The Netherlands, and US over the past three decades indicate lower variation seeking tendencies compared to results obtained in this study (Olsen et al., 2015;Schaefer et al., 2018;. Olsen et al. (2015) definition of high, medium, and avoider VARSEEK segments by wider mean ranges indicate a lower variation seeking tendency, giving reason to conclude that the variation seeking tendency of the respondents in this study is generally high. The large effect that variety seeking tendencies have on WTT and APP is confirmed.
In addition, we found a strong influence of the presented origin information. The general influence of visual and verbal information, and combinations hereof, on willingness to try innovations in food is explained by Chaudhuri and Micu (2014) supporting the findings of this study. Unexpectedly, men in the low VARSEEK segment were affected much by the information. However, the results are not conclusive due to the low number of respondents in this segment, and more research is needed to establish such an effect. Although the finding provides some robust results and conclusions, it also raises questions regarding other aspects. Especially the differences between gender, combined with food variety seeking, We suggest following up with more research, before sound conclusions can be made. For the present study, an online presentation of visual stimuli and written information was used. Additional investigations of sensory input from full tasting and ingestion of the hummus samples may well amplify the effects that different levels of variety seeking have.

Implications for gastronomy
The purpose of this study is to investigate how different hummus are perceived. In addition it specifically aims to investigate the relationship between Danish consumers' environmental attitude (measured by EAI-24) and variation seeking tendency (measured by VARSEEK) as psychographic background characterization on one hand, and how these two measures may predict Willingness to try, Willingness to pay and Appetizing appearance. Simultaneously, the effects of providing information about origin, history, and common use of the selected pulses was investigated.
Increase in variety seeking tendency significantly increased both willingness to try, willingness to pay, and ratings for appetizing appearance while increased attitude towards the environment showed only a slightly significant increase on appetizing appearance.
The implication for gastronomy mainly lies in the demonstration that variety seeking affects perception of the products more than environmental attitudes. In addition, providing information to respondents, particularly the segment with low variation seeking tendency had a significantly positive effect on willingness to try and ratings of appetizing appearance. This knowledge shows that it is beneficial to target the information about lesser known plant-based products towards convincing consumers with lower levels of variety seeking.

Declaration of competing interest
No potential conflict of interest was reported by the authors.