Choice-induced tasting. Evaluating the effect of choice on children ’ s acceptance of an unfamiliar vegetable

Children ’ s vegetable intake is in general below recommendations. It is known that self-selection of vegetables out of multiple options can increase intake in children. However, it is not clear if this effect is driven by a pre-existing preference for the selected food, or purely by having a choice. Allowing children to choose could create a positive bias and endorse feelings of autonomy, which could increase acceptance of the food and promote intake. The aim of the present pre-registered study was to investigate the effect of choice in promoting the acceptance of an unfamiliar vegetable during a blind-choice task. Children aged 4 – 5 years old ( n = 161) were randomly assigned to the choice or no-choice condition. Each child was presented with three closed cups, containing a vegetable. The children were unaware that the cups contained the same vegetable, a piece of raw celeriac. In the choice condition, the children were able to choose a cup themselves whereas in the no-choice condition, the experimenter made the decision. We hypothesized that children in the choice condition would show more acceptance of the vegetable compared to children in the no-choice condition and that higher levels of picky eating would lead to less vegetable acceptance. The results demonstrated that choice indeed significantly increased vegetable acceptance (4.7 ± 1.7 versus 4.0 ± 1.9 on a 6-point scale), irrespective of pickiness in eating. In addition, picky children showed less acceptance of the vegetable compared to non-picky children. The results of this study imply that choice is an important factor in promoting unfamiliar vegetable intake in young children. Including children in the decision process may be a practical strategy for educators and caregivers to increase children ’ s vegetable intake.


Introduction
Many young children show difficulties with eating such as picky eating or food neophobia, which can lead to insufficient vegetable intake (Dovey et al., 2008).Only 23.5% of European children meet the World Health Organization (WHO) recommendation for vegetable intake (Lynch et al., 2014), which may lead to an increased risk of several diseases.In addition, (eating) behaviors appear difficult to change later in life and it is, therefore, important to stimulate healthy eating habits in young children (Aune et al., 2017;Fadnes et al., 2022;McDermott et al., 2010;Nicklaus, Boggio, Chabanet, & Issanchou, 2005;Skinner et al., 2002).Sufficient healthy food intake starts with food acceptance, something that appears to be no problem when children are in their first year (Birch et al., 1998).However, around the age of 2-4 years, children start avoiding unfamiliar foods and rejecting foods they previously liked (Jacobi et al., 2003).Even though most children grow out of this, individual differences remain and show that children who accept more vegetables at an early age, continue having healthier diets later in life (Nicklaus, 2009;te Velde et al., 2007).
There is a need for effective and easy-to-implement strategies to encourage children to meet the recommended amount of vegetable intake.Offering children a choice of vegetable might be such a strategy.There is substantial evidence of the influence of preference on food choice and acceptance.Studies show that children from age 2 tend to make food choices based on their preferences (Nicklaus, Boggio, & Issanchou, 2005), and children eat more vegetables when they can pick the vegetables they prefer (Blanchette & Brug, 2005;Gibson et al., 1998).However, the effectiveness of choice as a strategy to enhance food acceptance in children has not been conclusively shown in the literature.King et al. (2008) demonstrated the enhancing effect of providing choice to adult consumers in food evaluations and intake.They found that having respondents choose a product, was associated with higher hedonic ratings and acceptability scores, compared to a no-choice test format (King et al., 2007(King et al., , 2008)).A study with children found that in a school canteen setting choice-offering had a positive effect on children's vegetable intake (Domínguez et al., 2013).In addition, a study by de Wild et al. (2015) showed that choice-offering is an effective strategy to increase children's intake of familiar vegetables during family dinners at home.Altintzoglou et al. (2015) also observed a positive effect of choice on liking when children were able to choose a type of fish themselves compared to a no-choice situation in which one of two types of fish was assigned to them.Thus, providing children with a choice possibility led to a more positive evaluation.However, two studies found no effect of free choice on vegetable intake (Olsen et al., 2019;Zeinstra et al., 2010).Yet overall, it seems that involving children in the decision-making process might lead to an increase in food acceptance and consumption.
Although choice-offering appears to be a promising strategy to increase food intake, several factors seem to be able to explain the enhancing effect.Since participants do not choose blindly, it is plausible that initial preference caused the increased food intake.Furthermore, studies have shown that serving a variety can also be effective in increasing intake (Bucher et al., 2014;Hendriks et al., 2021;Roe et al., 2013).Because offering choice implies that more than one vegetable is provided, hence more variety, it can be difficult to distinguish between variety and choice as an intervention (de Wild et al., 2015).Also, as reported by most studies, it is possible that allowing participants to choose increased liking and acceptance of the vegetables.Studies conducting a blind paradigm have previously shown that choosing can increase liking, a phenomenon called choice-induced preference.The theory of choice-induced preference states that people prefer the things they choose and that this results from the effort to reduce cognitive dissonance by aligning our attitudes with our behavior (Brehm, 1956;Enisman et al., 2021;Sharot et al., 2010).Children around two years of age already show choice-induced preference when choosing blind between toys (Egan et al., 2010;Silver et al., 2020;Wiesmann et al., 2022).Due to the choice-dissonance bias, the initial dislike of a vegetable might be in conflict with the fact that they choose it themselves, and to solve this conflict they might increase the liking of the vegetable.Another reason why choice might increase liking is because it gives children more autonomy and control.According to self-determination theory, increased feelings of autonomy and personal control when choosing enhances intrinsic motivation (Deci et al., 1994).Intrinsic motivation, in turn, has been related to improved performance, higher liking, interest, and greater satisfaction (Patall et al., 2008).
However, the existing studies on the effect of choice on vegetable intake do not allow strong conclusions about the relative importance of these factors (Altintzoglou et al., 2015;de Wild et al., 2015;Domínguez et al., 2013).Was there an increase in liking and intake of the vegetables because the children were able to choose, or did they choose the food they already preferred?The objective of the present study was therefore to disentangle these possibilities and investigate the sole effect of choice on young children's acceptance of an unfamiliar vegetable in a blind-choice paradigm (Chen & Risen, 2010;Izuma & Murayama, 2013).In this paradigm preferences and variety cannot play a role in choice behavior because the food samples to choose from are hidden and are all the same.Therefore, we can measure the sole effect of making a choice.We expect that when children have the option to choose, they will be more willing to accept the food.However, it is important to consider the influence of picky eating and the inter-related construct of food neophobia, as they contribute to the rejection or acceptance of foods, particularly fruits and vegetables (Dovey et al., 2008;Taylor & Emmett, 2019).Previous research investigating the effect of choice on the acceptability of novel flavors has shown that under choice test conditions, 'neophobics' select novel flavors less often compared to 'neophilics' (King et al., 2008).
We hypothesized that: first, children who have the opportunity to choose would show greater acceptance of the vegetable compared to children without a choice.Second, higher levels of picky eating would be associated with lower vegetable acceptance.Third, we hypothesized that picky eating would moderate the effect of choice on vegetable acceptance, specifically that if picky eating is higher, the effect of choice on acceptance would be smaller.

Study design
The study employed a randomized trial using a between-subjects design with two conditions.Data were collected from each child during a single individual session at their elementary school, either in a morning or afternoon session.The children were randomly assigned to either the choice condition, where they could select one of three closed cups containing a hidden vegetable, or the no-choice condition, where one of the cups was chosen by a researcher.After the cup was chosen and the food was revealed, the children were invited to taste the celeriac, and their acceptance was rated on a scale ranging from 1 (refusal to taste) to 6 (swallowing the food).In addition, before the choice test, the child's pickiness was assessed by a vegetable photo task.Children were presented with photos of 12 vegetables and rated their willingness to try each vegetable using a scale from 1 (not at all willing to try) to 4 (very willing to try).An overall score was calculated as the mean rating across all 12 vegetables; lower scores indicated higher levels of pickiness, while higher scores indicated low pickiness.This pickiness score served as the moderator variable tested in this study.
The present study was approved by the Ethical Review Committee of the Faculty of Psychology and Neuroscience of Maastricht University with reference number ERCPN-OZL_233_11_02_2021_S7.The study was pre-registered via As Predicted (https://aspredicted.org/VPT_WBB) and the protocol followed the pre-registered study design.

Participants
To estimate the number of participants needed in this study to detect a significant difference between the two groups, a sample size calculation was performed using G*power 3.1.For an ANCOVA analysis, to observe a medium effect size (f = 0.25), with 80% power and alpha set at 0.05, a minimum of 128 participants was required.Participants were recruited per class of schoolchildren, and it was decided not to recruit only part of a class.Four public elementary schools in Maastricht, the Netherlands, were approached in September 2022 based on their geographic proximity.Two schools, which collectively provided seven classrooms with in total 181 children, agreed to participate in the study.
After receiving active consent from the elementary school department head, passive consent was obtained from the children's parents and caregivers.The caretakers were informed about the study via a letter distributed via the weekly online school newsletter and provided with the option to withdraw their child from the study.The children provided verbal consent at the onset of the study, and they were excluded if the child indicated that they did not want to participate.Due to the opt-out approach employed in the present study (complete inclusion of eligible children), the selection bias was minimized.All children in the first two grades, who attended the preschool class on the test day, were considered eligible to participate.The parents or caregivers of 13 children decided to exclude their child from testing and another 7 children were either sick on the testing day or not willing to participate.No further participants were excluded from the data analysis.A remaining total of 161 children, 81 boys and 80 girls aged between 48 and 74 months old (M = 59.32,SD = 6.74) participated in the study.

Vegetable Photo Task
The vegetable photo task was an adjusted version of the test from Pliner (1994) to assess picky eating.The photographs were a mixture of R. Mourmans et al. commonly eaten vegetables in the Netherlands (tomato, carrot, bell pepper, cucumber, mushroom, broccoli) and more uncommon vegetables (fennel, celeriac, okra, lotus root, romanesco, artichoke) (Ocké et al., 2008).Studies have shown that food photos are efficient in obtaining information about food preferences and food choices in children of various ages (Olsen et al., 2012).
We presented the twelve photographs one by one in a random order on a tablet device (iPad; Apple Inc., Cupertino, California, USA) using the Qualtrics software (Qualtrics LLC, Utah, USA).The randomization of questions was done by the software, and the researcher presented the photos to the child on the screen and asked the questions verbally.First, we assessed the child's experience with each vegetable by asking them to indicate (i) whether they recognized the vegetable (yes = 1, no = 0) and (ii) if they had ever eaten the vegetable before (yes = 1, no = 0).The scores were averaged for the 6 common and uncommon vegetables separately, with higher scores (up to a maximum of 1) indicating better recognition and higher frequency of vegetables ever eaten before respectively.
In addition, for each photograph, children indicated their willingness to try the vegetable.The child's response was evaluated on a 4-point smiley face scale ranging from 1 (not at all willing to try) to 4 (very willing to try).A 4-point forced-choice scale was used so the participants could not adopt a neutral attitude (Rubio et al., 2008).Furthermore, smileys were used because they make it possible to introduce the task in a game-like situation, which is known to increase children's motivation and attention (Proserpio et al., 2020).The researcher entered the responses for the first two questions, while for the smiley scale, the child occasionally interacted directly by clicking on the chosen smiley positioned beneath the vegetable photo on the iPad.The mean score of the 12 ratings was computed, resulting in a score between 1 and 4, which served as an index of picky eating.Higher scores indicated less picky eating.

Choice Task
The choice task was a variation of the blind-choice paradigm used in Wiesmann et al. (2022), which includes the dimension of food acceptance.In this task, children were presented with three closed cups containing a hidden vegetable.Participants were told that underneath each of the cups, a piece of vegetable was hidden that they had seen earlier during the vegetable photo task.This way, children had some idea of the pool they were choosing from, but they were unaware of which specific vegetable was hidden.However, in both the choice and no-choice condition, all three cups contained a piece of raw celeriac.When cut into small cubes, raw celeriac becomes almost unrecognizable, ensuring that visual cues could not influence food acceptance.This is important since the focus of the study was to investigate the sole effect of choice on food acceptance while preventing familiarity or preference from intervening.
In the choice condition, children were allowed to choose one of the three cups themselves, while in the no-choice condition, the experimenter chose a cup for them.Then, similar to the vegetable photo task, the participants indicated their willingness to try the chosen (still hidden) vegetable using the 4-point smiley scale.This scale ranged from 1 (not at all willing to try) to 4 (very willing to try), measuring the child's willingness to try the food before it was revealed.This measurement was included to assess the effect of choice on willingness to try before the visual cues of seeing the food could potentially influence food acceptance.After the cup was chosen and the food was revealed, the children were invited to taste the celeriac, and their acceptance was rated according to their interaction with it on a scale ranging from 1 (physical refusal) to 2 (verbal refusal), 3 (smelling), 4 (licking), 5 (placing in the mouth), and 6 (swallowing the food) (Blissett et al., 2016).This score, ranging from 1 to 6 serves as the outcome measurement of vegetable acceptance.

Child self-report of food fussiness
The child-adapted version of the Children's Eating Behaviour Questionnaire (CEBQ) food fussiness (FF) subscale was translated into Dutch and used to measure picky eating using self-report by the children (Venkatesh & DeJesus, 2022;Wardle et al., 2001).Self-report is valuable as parents can only report on aspects of their children's behavior that are within their control or view, and there is a potential for bias (Proserpio et al., 2020).In addition, Venkatesh and DeJesus (2022) demonstrated that children can report their own picky eating.However, questionnaires reflect only attitude and intention, and this does not necessarily reflect actual behavior (Damsbo-Svendsen et al., 2017).In the present study, it was therefore used as validation for the vegetable photo task.The child-adapted FF questionnaire consisted of 5 items that measured children's food selectivity including the following items: 1. Do you like to try new foods that you have never eaten before?2. Do you like to eat lots of different kinds of foods? 3. Do you like the meals you eat at home? 4. Do you like to taste new foods that you haven't tasted before?and 5. Do you ever decide that you don't like a food before you have tasted it?Based on the study from Venkatesh and DeJesus (2022), children could first answer 'yes', 'sometimes', or 'no'.If they answered 'yes', they were provided with the additional options 'always' or 'most of the time', and if they answered 'no', they were provided with the options 'not very much' or 'never', yielding a 5-point response scale.The responses were scored accordingly, resulting in scores ranging from 1 to 5, with question 5 using reversed scoring to ensure a consistent measurement.Higher scores on the questionnaire indicate greater food pickiness.To minimize any recency bias, the order of response options was randomized, leading the experimenter to either ask "yes, sometimes, no," or "no, sometimes, yes," as presented by the iPad.

Procedure
The experimenters attended the opening classroom session at the schools and introduced themselves to the children.After a short explanation in the classroom, two children at a time were asked to join the two experimenters in a quiet spot outside the classroom.The children were seated outside each other's hearing distance and tested individually.The test procedure consisted of a total of three tasks.The experiment started with a brief verbal description of the procedure.The children were told that they were going to play some games (a game on the iPad with photographs of vegetables) and that they would be able to taste some vegetables.
The experimenter first familiarized the children with the 4-point smiley rating scale.After the children demonstrated their understanding of the scale, they started with the vegetable photo task.This was followed by the choice test, during which participants were randomly assigned to one of two conditions through software-driven random allocation.The conditions included the choice condition (n = 82), in which children autonomously chose a cup, and the no-choice condition (n = 79), where the researcher chose a cup.Once the food was revealed, children in both conditions were invited to taste the vegetable.Tasting was completely voluntary and if they did put the celeriac cube in their mouth, spitting it out was allowed and the experimenter always handed over some tissues to make sure this option was clear.Then, the children completed the third and final task, the food fussiness subscale, and as a reward for their participation, all children were presented with a diploma.
During all phases, the experimenters carefully watched for possible signs of discomfort in the children, in which case the testing would be stopped, and the child returned to the classroom.This was however never necessary.Furthermore, the researchers documented the children's level of understanding regarding the instructions and questions for the three tasks.

Data analysis
We checked the equivalence of the two groups in terms of participant age, picky eating, and the score on the child-adapted CEBQ FF subscale, with separate independent t-tests with group (choice vs. no-choice) as a between-subjects factor.We further checked the equivalence of the two groups in terms of sex, school, and testing time, with a Chi-square test.
We employed an Analysis of Covariance (ANCOVA) to assess the impact of condition (choice vs. no-choice) on vegetable acceptance, and to test if picky eating moderated this effect.We interpreted the results in the context of both the main effects and the moderation effect of picky eating on the relationship between choice and vegetable acceptance.Furthermore, we conducted an additional ANCOVA to examine the effect of condition (choice vs. no-choice) on participants' willingness to try the still-hidden food sample.As in the previous analysis, we included picky eating as a covariate to test for its potential moderating effect on the relationship between choice and willingness to try the food.In addition, the correlation was calculated between the willingness to try the food before the children saw the sample and the vegetable acceptance.
To evaluate the experience with the vegetables in the vegetable photo task, two paired-sample t-tests were conducted (on recognition of the vegetables and vegetables eaten before) to compare the common and uncommon vegetables.Additionally, the correlations between these two measures of experience (averaged over all twelve vegetables) and picky eating were examined.Lastly, to assess the validity of the vegetable photo task as an index for picky eating, a correlation between picky eating and the child-adapted CEBQ-FF subscale was calculated.
Analyses were conducted including and excluding the participants whose comprehension level regarding the instructions appeared questionable.In total 11, 0, and 19 children were 'flagged' in the vegetable photo task, the behavioral choice task, and the child-adapted FF questionnaire respectively.The pattern of results remained the same, and excluding participants did not change the interpretation of the analyses.Therefore, we reported results including all participants.
All statistical analyses were performed using IBM SPSS version 28, and statistical significance was determined at an α-value of 0.05.The graphics were generated in R using the software package Rstudio and the packages: 'haven' and 'ggplot2'.The (anonymous) raw data used for the analyses are shared on OSF (https://osf.io/syzbn/).

Participant characteristics
The participant characteristics and differences per condition are summarized in Table 1.In total, eighty girls and eighty-one boys participated in the study.A chi-square test of independence showed that sex was not equally distributed over the two conditions.Moreover, there was no significant group difference in the children's age, and the two schools and the testing time (morning vs. afternoon) were balanced across the two conditions.Furthermore, pickiness did not significantly differ between the conditions according to both the vegetable photo task and the child-adapted CEBQ-FF subscale.

Effect of choice on vegetable acceptance
As hypothesized, vegetable acceptance on a 6-point scale as assessed in the choice task, was significantly higher in the choice condition (M = 4.73; SD = 1.74) compared to the no-choice condition (M = 4.03; SD = 1.85),F(1, 157) = 9.27, p = .003,η 2 = 0.06.For example, in the choice task, 73.2% of children in the choice condition placed the celeriac sample in their mouth or swallowed it, compared to 57% of children in the no-choice condition.In addition, a significant main effect of picky eating, evaluated on a 4-point scale as assessed by the vegetable photo task, on the acceptance of the vegetable was found, F(1, 157) = 50.02,p < .001,η 2 = 0.24.Children with a picky eating score in the lowest tertile (score <2.17, indicating high picky eating) had a mean vegetable acceptance score of 3.59, whereas children with a picky eating score in the highest tertile (score >2.75, indicating low picky eating) had a mean acceptance score of 5.35.
Contrary to the third hypothesis, no significant moderation effect of picky eating on the relationship between choice and vegetable acceptance was found, F(1, 157) = 0.76, p = .39.This suggests that the influence of choice on food acceptance remained consistent regardless of pickiness in eating.These findings confirm the first two hypotheses stating that choice increases vegetable acceptance, while picky eating leads to less acceptance of the vegetable.This is illustrated in Fig. 1, where the slopes of the regression lines for vegetable acceptance scores across picky eating scores are not significantly different for the choice condition and the no-choice condition.Additionally, despite the uneven distribution of sex across the two conditions, no effect of sex on food acceptance was observed, t(159) = 0.59, p = .56.

Effect of choice on the willingness to try the vegetable (measured when the vegetable was hidden)
The willingness to try the vegetable, evaluated on a 4-point scale when the vegetable was still hidden, was significantly higher in the choice condition (M = 3.15; SD = 1.15) compared to the no-choice condition (M = 2.84; SD = 1.26),F(1, 157) = 4.45, p = .036,η 2 = 0.03.For example, 78.1% of children in the choice condition indicated to be willing or very willing to try the hidden vegetable, compared to 65.8% of children in the no-choice condition.In addition, a significant main effect of picky eating, evaluated on a 4-point scale as assessed by the vegetable photo task, on the self-reported willingness to try the vegetable was found F(1, 157) = 57.40,p < .001,η 2 = 0.27.Children with a picky eating score in the lowest tertile (score <2.17, indicating high picky eating) had a mean willingness-to-try-the-vegetable score of 1.88, whereas children with a picky eating score in the highest tertile (score >2.75, indicating low picky eating) had a mean willingness-to-try score of 3.59.However, no moderation effect of picky eating was found F (1, 157) = 1.12, p = .291.In other words, the presence of picky eating did not appear to influence how the choice condition affected children's willingness to try the hidden vegetable.This is depicted in Fig. 2, where the slopes of the regression lines for willingness to try the hidden vegetable scores across picky eating scores are not significantly different a Mean value and standard deviations per group.b Picky eating is calculated as the mean willingness to try 12 vegetable photographs, with scores ranging from 1 (high picky eating) to 4 (low picky eating).c The scores of the child-adapted version of the Children's Eating Behaviour Questionnaire (CEBQ) Food Fussiness (FF) subscale are calculated as the mean of 5 questions, with scores ranging from 1(low picky eating) to 5 (high picky eating).
between the choice condition and the no-choice condition.However, the two main effects are evident as the line representing the choice condition is positioned above the line for the no-choice condition, and both lines exhibit an upward trend.Furthermore, a strong correlation was found between the willingness to try the hidden vegetable and the eventual vegetable acceptance, r (159) = 0.58, p < .001.This means that the children who indicated willingness to try the food before seeing it, also showed this in their behavior by increased vegetable acceptance.

Results of the vegetable photo task
The results indicate that the six vegetables predefined by the researchers as common, based on national data, were better recognized by the children (M = 0.97; SD = 0.12, yes = 1 and no = 0) than the uncommon vegetables (M = 0.25; SD = 0.32), t(160) = − 28.28, p < .001.The effect of choice and picky eating on vegetable acceptance.The main effects of choice and picky eating are significant.The interaction effect between picky eating and choice condition is not significant (the slopes of the two lines do not significantly differ).Please note that in the figure, all data points have been jittered to avoid overlap (0.05 horizontally and 0.15 vertically).

Fig. 2.
The effect of choice and picky eating on the willingness to try the vegetable sample (measured after choice, but before seeing the vegetable).The main effects of choice and picky eating are significant.The interaction effect between picky eating and choice condition is not significant (the slopes of the two lines do not significantly differ).Please note that in the figure, all data points have been jittered to avoid overlap (0.05 horizontally and 0.15 vertically).
The proportion of children who reported that they had ever eaten the common vegetables (M = 0.84; SD = 0.23) was higher than the proportion who had ever eaten the uncommon vegetables (M = 0.15; SD = 0.25), t(160) = − 29.42, p < .001.In addition, picky eating, as measured with this task, was moderately correlated to both the recognition and having ever eaten the twelve vegetables (respectively r(159) = 0.38, p < .001and r(159) = 0.59, p < .001).More experience with vegetables is therefore related to less picky eating.
Furthermore, 24.1% of the children indicated recognizing the celeriac, and 15.2% had previously consumed it.This demonstrates the fact that celeriac, used as the food sample for the evaluation of the dependent variable of food acceptance, was indeed unfamiliar to most of the children.
The two measures of picky eating, the vegetable photo task, and the child-adapted FF subscale were found to be moderately negatively correlated, r(159) = − 0.42, p < .001.This means that participants who are more willing to try the 12 vegetables in the vegetable photo task, scored low on picky eating on the FF subscale.This supports that the test is indeed an index for picky eating.Furthermore, a moderate negative correlation was found between both the FF subscale and the willingness to try both the uncommon vegetables r(159) = − 0.36, p < .001,and the common vegetables r(159) = − 0.33, p < .001.This indicates that the vegetable photo task measures not just food neophobia, but picky eating in general.

Discussion
This investigation of whether offering children a choice affected their acceptance of an unfamiliar vegetable, showed that children in the choice condition had a higher rating of behavioral acceptance of the celeriac than children in the no-choice condition.Likewise, children in the choice condition indicated a greater willingness to try the food sample when it was still hidden.In addition, children low in pickiness accepted the celeriac better than the children high in pickiness.However, the study showed no significant moderation effect of pickiness on the effect of choice, indicating that choice increased food acceptance irrespective of pickiness in eating.The effect of choice may be explained by cognitive dissonance.Cognitive dissonance can lead to a positive bias towards someone's own choices because liking what one chooses restores the mental balance (Brehm, 1956;Festinger, 1962).In addition, the provision of choice may induce feelings of autonomy and personal control (Katz & Assor, 2007;Patall et al., 2008).According to self-determination theory, the increased sensation of autonomy and personal control increases intrinsic motivation to carry out an activity, hence more vegetable acceptance in children who were able to choose (Ryan & Deci, 2000;Zuckerman et al., 1978).
Choice appears to be a useful technique in motivating children to try a novel food and this can be the first step towards increased vegetable intake.An effective way to enhance food acceptance and intake of novel vegetables in young children is through repeated exposure (Cooke, 2007;Pliner, 1982).Children like what they know, and they eat what they like.Repeated exposure to the taste of a novel vegetable can make it more familiar and thus, more preferred.According to Birch (1979), a major part of the variance in 3-to 4-year old's preferences for a particular food was explained by the degree of familiarity.However, in order to expose children to the taste of a novel food, they first need to put it in their mouth.This is not always easy to accomplish, as young children tend to refuse certain foods (often including vegetables).Therefore, combined strategies could be examined in future research with choice offering as a strategy to achieve the first step of acceptance of a novel food, followed by repeated exposure to promote familiarity.Once the child has accepted an unfamiliar vegetable and tried it, repeated exposure could be implemented as a strategy to further increase vegetable intake until daily recommendations are reached.Moreover, the fact that offering a choice appeared equally effective regardless of the children's levels of pickiness, is a positive finding.It suggests that even the picky eaters can be persuaded to give food a try when more autonomy is provided.Therefore, involving children in the decision process by for example choosing a recipe, purchasing ingredients, and cooking may promote healthy eating habits such as increasing vegetable consumption (Maiz et al., 2021).However, a limitation that should be considered in the context of the result is that the study did not measure vegetable intake based on the ad libitum amount consumed by the participants.Participants in this study were only provided with the opportunity to taste one piece of celeriac; therefore, no conclusions can be drawn regarding intake in terms of the amount consumed.
The vegetable photo task appeared to be a good tool for measuring pickiness in children aged 4-5 years old.Children were able to differentiate between common and uncommon vegetables and the measurement correlates with both the self-report on the CEBQ-FF questionnaire and with the actual behavior in the choice task.When children reported recognizing more vegetables, they also reported that they were more willing to try the vegetables.Perhaps children who are offered a greater variety of vegetables at home have more experience with vegetables and thus like a greater number of vegetables because of the repeated exposure.However, it is also possible that the parents of picky children evolved the habit to offer overall fewer vegetables since the children refuse them anyway.Since this current finding is correlational, causality remains speculative.
A strength of this paradigm is that, even at this young age, the children were capable of providing self-assessments for the measures, avoiding the need for caregiver reports.Additionally, many of the attitude measures were not limited to the specific vegetable used in the choice test.
However, a limitation of this study is that we only measured picky eating using self-report by the children.It would be interesting for future research to include parent reports on the Children's Eating Behaviour Questionnaire (CEBQ) to increase validity (Wardle et al., 2001).Furthermore, the study was conducted in a school setting with unfamiliar test leaders.Children may behave differently in a home setting.Also, children were only provided with the intervention once and the effect may diminish after several exposures when the novelty wears off.Finally, we did not test a clinical sample including children diagnosed with for example ARFID, avoidant/restrictive food intake disorder (Zimmerman & Fisher, 2017).Therefore, in the future, including a clinical sample and research validating the effect of choice in the long term is advised.
In conclusion, this study demonstrated that having a choice does not just increase children's food acceptance because they can choose what they prefer, but choice in itself can have a positive effect on food acceptance.So, involving children in the decision-making process can be a useful addition to an exposure program (Nekitsing et al., 2018), easily integrated within a classroom setting to promote vegetable consumption.

Ethical statement
All study procedures were approved by the Maastricht University Ethical Research Committee of Psychology and Neuroscience (reference number: ERCPN-OZL_233_11_02_2021_S7).

Fig
Fig.1.The effect of choice and picky eating on vegetable acceptance.The main effects of choice and picky eating are significant.The interaction effect between picky eating and choice condition is not significant (the slopes of the two lines do not significantly differ).Please note that in the figure, all data points have been jittered to avoid overlap (0.05 horizontally and 0.15 vertically).

Table 1
Participants' characteristics by condition.