The Psychological Record https://doi.org/10.1007/s40732-018-0287-y

ORIGINAL ARTICLE

The Relationship Between Delay and Social Discounting, and Body Mass Index in University Students Katherine Wainwright 1 & Breanna E. Green 1 & Paul Romanowich 1

# Association for Behavior Analysis International 2018

Abstract Delay discounting is associated with body mass index (BMI), with individuals who discount more having higher BMIs. This same relationship was found when a social component was added during a social temporal discounting procedure. However, no study has tested whether the social component by itself is related to BMI. The current study tested the relationships between social discounting, delay discounting, and BMI. Eight hundred seven undergraduates completed delay and social discounting measures, and a demographic questionnaire from which BMI was calculated. Social discounting was not significantly related to age-adjusted or non-age-adjusted BMI. However, using non-age-adjusted BMI scores and dichotomizing individuals as either non-obese/obese resulted in a significant relationship between social discounting and BMI in the predicted direction, whereby participants with a higher BMI shared significantly more than non-obese individuals. Consistent with previous research, there was a significant relationship between delay discounting and BMI, and a significant correlation between delay and social discounting. Additionally, participants with low BMI (< 18.5) appeared to substantially contribute to the significant results. The current results replicated a previous delay discounting and BMI relationship, but also demonstrate the need to standardize how BMI is calculated and focus on participants with low BMI in relation to discounting measures. Keywords Altruism . BMI . Discounting . Self-control . Sharing

Food and eating are an important aspect of our everyday lives. Beyond providing the essential nutrients for sustaining life, eating and food sharing are incorporated within a larger social system that allows us to connect interpersonally and culturally (Rozin, 2005). Research suggests that individuals model their food consumption after those individuals with whom they eat (Kaisari & Higgs, 2015; de Castro, King, Duarte-Gardea, Gonzalez-Ayala & Kooshian, 2012). For example, Howland, Hunger, and Mann (2012) examined eating norms wherein individuals decreased food intake when their friends restricted eating, and later self-restricted their food intake when alone. When compared to eating with other companions such as coworkers, eating with family and/or spouse correlates to increased and quicker food consumption (de Castro, 1994). Eating with others can also involve a food-sharing component, where individuals offer their own food to others, ask to * Paul Romanowich [email protected] 1

Department of Psychology, University of Texas at San Antonio, One UTSA Circle, San Antonio, TX 78249-0652, USA

consume foods belonging to their eating partner, or eat meals where the food is to be divided among each of the eating companions. Koh and Pliner (2009) conducted a study examining plate size, degree acquaintance, and sharing among pairs of women. The participants served and ate more food with friends than with strangers. Participants eating from large plates with friends reported more additional helpings when sharing than when not sharing, and eating with a stranger had the opposite effect (Koh & Pliner, 2009). The results suggest that eating behavior can be significantly influenced by social and environmental factors. Unfortunately, these changes in eating behavior can also lead to increased caloric intake, which contributes to weight gain and an increased body mass index (BMI). Research has also shown a positive relationship between BMI and age (Bickel, Wilson, Franck, Mueller, Jarmolowicz, Koffarnus, et al., 2014; Dahl, Reynolds, Fall, Magnusson, & Pedersen, 2014; Jackson, Janssen, Sui, Church, & Blair, 2012; Nooyens, Visscher, Verschuren, Schuit, Boshuizen, van Mechelen, & Seidell, 2009) due to an increase in caloric intake, a sedentary lifestyle, or a combination of the two (Reas, Nygård, Svensson, Sørensen, & Sandanger, 2007).

Psychol Rec

Social discounting is a quantitative description of resource allocation (i.e., sharing) as a function of social proximity. In social situations, reward values change depending on the number of people one is sharing with (Rachlin & Raineri, 1992) and the social distance between those people (Jones & Rachlin, 2006, 2009; Rachlin & Jones, 2008a, b). Increasing either the number of people or the social distance with whom an individual can share decreases sharing. Delay discounting, whereby the current value of reward hyperbolically decreases as the time until receiving that reward increases (Kirby & Marakovic, 1996), is closely linked to social discounting. Delay discounting is significantly correlated with social discounting (Jones & Rachlin, 2009). That is, as delay discounting increases, sharing decreases. Importantly, delay discounting has been linked to BMI, whereby individuals who discount more steeply have higher BMIs (Bickel et al. 2014; Bruce, Black, Bruce, Daldalian, Martin, & Davis, 2011; Epstein, Jankowiak, Fletcher, Carr, Nederkoorn, Raynor, & Finkelstein, 2014; Fields, Sabet, Peal, & Reynolds, 2011; Nederkoorn, Smulders, Havermans, Roefs, & Jansen, 2006; Reimers, Maylor, Stewart, & Chater, 2009; Thamotharan, Lange, Ramos, & Fields, 2016; Weller, Cook, Avsar, and Cox, 2008) and are more likely to overeat (Yeomans, Leitch, & Mobini, 2008). Because many different studies have shown this delay discounting and BMI relationship, training methods have been applied to obesity interventions in adolescents (Kulendran, Vlaev, Sugden, King, Ashrafian, Gately, & Darzi, 2014). Given the social nature of eating and the previous link between delay discounting and BMI, one would expect a relationship between social discounting and BMI, as well. One study has examined the relationship between BMI and social temporal discounting (Bickel et al., 2014). Social temporal discounting is similar to delay discounting but includes a social aspect where a participant can share their reward. Individuals were asked to choose between a smaller reward now that would be evenly distributed among a group (or kept for themselves), and a larger reward later shared with the same group. Bickel et al. (2014) compared three different types of discounting including temporal (i.e., delay), social temporal, and probability discounting to determine if any of these measures could differentiate between obese and non-obese individuals. Participants were classified as either normal weight (BMI < 25), overweight (BMI ≥ 25, and < 30), or obese (BMI ≥ 30). Results indicated no significant differences between these three groups, so the overweight and obese participants were collapsed into one group (i.e., overweight/obese group). After this modification, results indicated that the overweight/obese group discounted more than the non-obese group in temporal discounting and social temporal discounting, but not probability discounting. That is, obese individuals discounted more than non-obese individuals. However, given that the social temporal discounting measure

assesses both delay and social factors simultaneously, there was no way for Bickel et al. to assess the social factor independent of the delay factor. Thus, it is still unknown whether social discounting is related to BMI, without a delay component. In social temporal discounting, participants are asked to choose between a smaller reward to be split between the participant and a group of 9 strangers (or kept for themselves), and a larger but delayed reward to be split between the same group. The number of people in the group and the social distance (i.e., stranger) remained constant for each choice. By comparison, social discounting does not include a time delay, but rather asks participants to choose between an amount of money to keep for themselves and an amount of money to immediately share with a person of a given level of social proximity (Jones & Rachlin, 2006, 2009; Rachlin & Jones, 2008a, 2008b). For example, they may choose to keep $135 for themselves instead of keeping $75 and giving $75 to the other person. Given the existing literature on the social aspects of eating (i.e., Koh & Pliner, 2009) and the relationship between social temporal discounting and BMI found by Bickel et al. (2014), the present study aimed to assess the relationship between social discounting and BMI, which has not yet been examined without a delay component. If social discounting is related to BMI in a similar manner as both delay discounting and social temporal discounting, participants who share more will also have higher BMIs. Thus, we hypothesize that individuals with higher BMI will have higher self-reported social discounting (i.e., share more). In addition, we sought to replicate the findings of Bickel et al. (2014) and others who have found a positive relationship between delay discounting and BMI. Our second hypothesis is that delay discounting and BMI will be significantly related, with individuals with higher BMI’s discounting more. Lastly, we hypothesize that, consistent with Jones & Rachlin’s (2009) findings, there will be a significant correlation between delay and social discounting, with individuals who discount more on a delay discounting measure sharing less on a social discounting measure.

Method Participants Participants included 801 undergraduate students at a large university in the southwestern USA. Participants were recruited through their Introduction to Psychology course and completed the online survey, accessible from their personal computers, through SurveyMonkey for partial credit towards an experimental requirement. The study was approved by the local Institutional Review Board (IRB) before collecting data, and informed consent was obtained before survey initiation by

Psychol Rec

all participants. Participants’ age ranged between 17 and 41 years (M = 18.99, SD = 2.42) with 68.3% female. Most participants identified as Latino (47.3%), followed by Caucasian (27.6%), African American (12.3%), Asian (9.7%), and other (3.1%).

Measures Body Mass Index Participants completed a basic demographic questionnaire including self-reported height and weight. BMI, a measure of body fat, was calculated using the following equation: BMI = (Weight (lbs) / Height (in)2) × 703 (Centers for Disease Control and Prevention [CDC], 2015a). Adults aged 20 years old or older are considered underweight with a BMI of less than 18.5, normal weight with a BMI between 18.5 and 24.9, overweight between 25.0 and 29.9, and obese above 30.0 (CDC, 2015a; World Health Organization [WHO], 2017a). Following best practices of both the WHO (2017b) and the CDC (2015b), for participants younger than 20, BMI categories were adjusted for age and gender. Using percentile growth charts to compare each participant’s BMI relative to other adolescents of the same gender and age, participants aged 17 to 19 years were classified as underweight at less than the 5th percentile, normal weight between the 5th and 85th percentile, overweight between the 85th and 95th percentile, and obese at the 95th percentile or above (CDC, 2015b). Social Discounting Participants completed a social discounting task (Jones & Rachlin, 2006) in which they were asked to make choices between a hypothetical amount of money to keep for themselves and a hypothetical amount of money to share with a person of a given level of social proximity. Participants were first asked to mentally make a list of 100 people from a friend or relative at 1 to a mere acquaintance at 100. Social discounting was assessed at the following levels of social proximity: 1, 2, 5, 10, 20, 50, and 100, and was always in ascending order. Up to 9 choices were made for each of the 7 levels of social proximity. Choices always started with the largest amount available for the participant alone ($155) and decreased in intervals of $10 to $75 for the participant alone. The alternative choice was a constant $75 for the participant and $75 for the other person. The point at which a participant switched between keeping all the money for themselves and giving up money to share with the other person at a given social distance ended the trial for that particular social distance and was considered the indifference point. For example, a participant may choose to keep $135 for themselves instead of keeping $75 for themselves and giving $75 to the other person. However, for the next choice between $125 for themselves and $75 for each person, they choose to forgo the $125 and allocate $75 for each person. In this case, they did not forgo $60 (135–75 = 60), but did forgo $50 (125–75 = 50).

Therefore, their indifference point at that social distance was estimated to be $55 (60 + 50/2). Social discounting was estimated by calculating the area under the curve (AUC) for each participant using the seven indifference points with an ordinal scaling transformation (i.e., AUCord; Borges, Kuang, Milhorn, & Yi, 2016). Calculated AUC scores range from 0 to 1 with scores closer to 1 indicating a shallow discounting function and more sharing. Participants with an AUC score closer to 1 share larger amounts of money at larger social distances than those participants with AUC scores closer to 0. Previous research has shown that people discount real and hypothetical rewards at similar rates and hypothetical rewards are a valid proxy for social discounting research (Locey, Jones, & Rachlin, 2011). Delay Discounting A monetary gains questionnaire was used to assess delay discounting rates for each participant. Participants were presented with a list of 27 questions developed by Kirby, Petry, and Bickel (1999) and asked to make a choice between smaller, immediate rewards or larger, delayed rewards. For example, they were asked to choose between $25 right now and $60 in 14 days. Similar to social discounting research, previous delay discounting research has also shown that people discount real and hypothetical rewards at similar rates and hypothetical rewards are a valid proxy (Johnson & Bickel, 2002; Madden, Begotka, Raiff, & Kastern, 2003). For each participant, indifference points were calculated and plotted as a function of time. A free parameter k was used to indicate the steepness of the discounting curve that corresponded with the geometric midpoint of the ranges (Kaplan, Amlung, Reed, Jarmolowicz, McKerchar, & Lemley, 2016). Values ranged from 0.00016 to 0.25, with higher k-values indicating more discounting.

Analysis Survey completion time was analyzed to determine if participants were actively engaged in the survey. Completion time ranged from 4.6 to 53.3 min with a mean of 11.2 min (SD = 6 min). Univariate outliers for survey completion time (n = 29) and BMI (n = 11) were excluded. Additionally, while there were no univariate outliers for delay discounting given the bounded nature of the measure, if a participant responded inconsistently (i.e., less than 75% consistency score based on Kaplan et al., 2016, n = 64), they were also not included in analysis. These quality checks resulted in the exclusion of 104 participants. Using the algorithm for determining systematic discounting described by Johnson and Bickel (2008), 246 participants were also excluded for non-systematic social discounting scores. These 260 participants responded with at least one data point that increased, rather than decreased, as social distance increased (e.g., a participant was unwilling to share with social distances 1 through

Psychol Rec

50, but did share with distance 100). Finally, 14 participants were excluded for SurveyMonkey error, resulting in uninterpretable social discounting data. The resulting sample size included 437 participants. Delay discounting scores were not normally distributed. Therefore, delay discounting scores were normalized using a logarithmic transformation prior to conducting a withinsubjects ANOVA to test for a magnitude effect. The magnitude effect test was conducted to provide further evidence for delay discounting validity in the current sample. Pearson correlations were used to assess the correlations between social and delay discounting score and BMI. Pearson correlations between social discounting, BMI, and the three delay discounting magnitudes were also calculated. As mentioned in the introduction, Bickel et al. (2014) first compared discounting rates in normal, overweight, and obese individuals and then dichotomized BMI to compare discounting rates between overweight/obese and non-obese individuals. Following that model, multiple one-way between-subjects ANOVAs with Tukey HD post hoc tests were utilized to test for group differences in social and delay discounting between underweight, normal, overweight, and obese individuals. Participants were then dichotomized into two groups: obese or non-obese and independent sample t tests were utilized to compare both social and delay discounting scores between the two groups. Bickel et al. (2014) dichotomized their participants as either obese or not, given a preliminary finding of no discounting score difference between normal, overweight and obese individuals. Previous research suggests both obese and overweight individuals discount at different rates than normal weight individuals (Fields et al., 2011). As such, it may not be appropriate to collapse overweight individuals into the same category as underweight and normal weight individuals. Therefore, we also dichotomized participants into two categories based on being overweight or not, where overweight and obese individuals were collapsed into one higher weight category and compared to a second lower weight category that included underweight and normal weight individuals. Independent samples t tests were then used to compare social and delay discounting scores.

0.03). A within-subjects ANOVA with a Greenhouse-Geisser correction between the three magnitudes (small, medium, and large) of the delay discounting measure provided evidence of a magnitude effect (F(1.92, 835.34) = 218.88, p < 0.001), supporting the validity of delay discounting within the current sample. As shown in Table 1, significant Pearson correlations were found between social discounting and delay discounting with higher rates of social discounting (i.e., less sharing) associated with more delay discounting. There was a marginally significant relationship between social discounting and BMI, with higher rates of social discounting associated with higher BMIs. There was a significant relationship between delay discounting and BMI, with higher rates of delay discounting associated with higher BMIs. The delay discounting small magnitude was significantly correlated with social discounting (r = − 0.119, p = 0.013), but not BMI (r = 0.071, p = 0.136). The medium magnitude was significantly related to both social discounting (r = − 0.131, p = 0.006) and BMI (r = 0.113, p = 0.019), as was the large magnitude (r = − 0.155, p = 0.001 and r = 0.123, p = 0.010, respectively).

Results

Table 1

After univariate time outliers and non-systematic discounting data were removed, participant’s (N = 437) adjusted BMI ranged from 15.55 to 39.53 with an average adjusted BMI of 23.86 (SD = 4.54). Most participants were normal weight (59.7%; n = 261), followed by overweight (19%; n = 83), obese (11%; n = 48), and underweight (10.3%; n = 45). Consistency scores on the social discounting measure based on Kaplan et al. (2016) ranged from 0.85 to 1 (M = 0.98, SD =

Age/Gender-Adjusted BMI Results A one-way between-subjects ANOVA comparing underweight, normal, overweight, and obese individuals found no significant differences on social discounting scores (F(3, 433) = 1.10, p = 0.347, η2 = 0.008). Figure 1 shows the median amount forgone across social proximity as a function of each different BMI category. As shown in Fig. 1, sharing decreased for individuals in the underweight category (open circles). Starting at social proximity 5, the median amount forgone for the underweight group was below the normal, overweight, and obese median values. This difference continued for social proximities 10, 20, and 50, before converging at social proximity 100. However, obese median values were equivalent to the normal weight values at all social proximities except 10 and 50. A one-way between-subjects ANOVA comparing underweight, normal, overweight, and obese individuals found significant differences between groups on delay discounting log k scores (F(3, 433) = 3.05, p = 0.028, η2 = 0.021). As shown in Pearson correlations between discounting and BMI

Social discounting Delay discounting BMI *Significant

Delay discounting

BMI

rs = − 0.144 p = 0.003* 1

rs = 0.093 p = 0.053 rs = 0.110 p = 0.022* 1

Psychol Rec

Fig. 1 Median amount of money forgone across social proximity as a function of adjusted BMI category. Open circles and diamonds indicate underweight and normal BMI median values, respectively. Open triangles and squares indicate overweight and obese median values, respectively. Median indifference points are shown due to the non-normalcy of the indifference points during the social discounting task

the top panel of Fig. 2, participants who were underweight (M = − 2.15, SD = 0.60) had significantly smaller delay discounting log k scores than participants who were either overweight (M = − 1.82, SD = 0.61, p = 0.032, d = 0.36) or obese (M = − 1.80, SD = 0.57, p = 0.046, d = 0.32), but not participants who were normal weight (M = − 1.91, SD = 0.70). All other pairwise comparisons were non-significant. The bottom panel of Fig. 2 shows AUCord data across each BMI category. Participants were reclassified based on BMI score into two categories: obese or non-obese. Two independent sample t tests found no significant differences between groups on delay discounting log k scores (t(435) = − 1.21, p = 0.23, d = 0.21) or social discounting AUCord scores (t(435) = − 0.50, p = 0.62, d = 0.08).

Non-adjusted BMI Results Analysis was also completed without age and gender adjusted BMI categories. That is, all participants were classified based on the adult BMI cut-off scores. Most participants were normal weight (61.6%; n = 269), followed by overweight (22%; n = 96), obese (10.8%; n = 47), and underweight (5.7%; n = 25). A one-way between-subjects ANOVA comparing underweight, normal, overweight, and obese individuals for nonadjusted BMI category found similar results to adjusted categories; there were significant differences for delay (F(3, 433) = 2.04, p = 0.042, η2 = 0.02) but not social discounting scores (F(3, 433) = 2.04, p = 0.11, η2 = 0.01). Participants were also reclassified and dichotomized as either obese or non-obese based on the adult BMI cut-off scores. Two independent sample t tests using non-adjusted BMI scores and obese/non-obese categories found no significant difference

Fig. 2 Box-and-whisker plots for delay discounting log k-values (top figure) and social discounting area under the curve (AUCord) values (bottom figure) as a function of adjusted BMI category. Gray boxes represent the interquartile range (IQR), with medians shown by the horizontal lines bisecting each box. Vertical lines above the boxes indicate either the IQR × 1.5 (delay discounting), or the distance between the third quartile and maximum obtained AUC value (social discounting). Vertical lines below the boxes represent either the distance between the first quartile and the minimum obtained log k-value (delay discounting), or the IQR × 1.5 (social discounting) for that BMI category

in delay discounting (t(435) = − 1.56, p = 0.12, d = 0.25) or social discounting scores (t(435) = − 0.75, p = 0.46, d = 0.01).

Age/Gender-Adjusted Collapsed Group Results After collapsing the four BMI categories into either lower (underweight + normal BMI) or higher (overweight + obese BMI) groups, two independent sample t tests found no significant difference on social discounting scores between the lower and higher weight categories (t(435) = − 1.63, p = 0.11, d = 0.17) but a significant difference on delay discounting (t(435) = − 1.96, p = 0.05, d = 0.22) with participants of lower weight (M = − 1.95, SD = 0.69) discounting less steeply than participants of higher weight (M = − 1.81, SD = 0.59).

Psychol Rec

Non-adjusted Collapsed Group Results Finally, analyses were also completed without age-adjusted BMI categories (see Table 2). An independent sample t test replicated differences in delay discounting (t(435) = − 2.39, p = 0.02, d = 0.25) with participants of lower weight (M = − 1.96, SD = 0.70) discounting less steeply than participants of higher weight (M = − 1.80, SD = 0.58) when BMI category was not adjusted for age. Contrary to age-adjusted results, independent sample t tests using non-age-adjusted BMI categories found a significant difference (t(435) = − 2.29, p = 0.02, d = 0.22) in social discounting scores with participants in the lower weight category (M = 0.47, SD = 0.28) discounting less steeply than those in the higher weight category (M = 0.53, SD = 0.27).

Discussion Given the relationship observed between BMI and social temporal discounting (Bickel et al., 2014) and the social aspects of eating behavior (de Castro, 1994; de Castro et al., 2012; Kaisari & Higgs, 2015; Koh & Pliner, 2009), the current study examined the relationship between social discounting and BMI without a delay component in college students. Social discounting was not significantly related to age-adjusted BMI, which did not support our first hypothesis. However, the relationship between delay discounting and BMI was consistent with previous research, and our hypothesis, in that increased BMI, was associated with more discounting. This significant relationship was confirmed with both ways of calculating BMI category (e.g., adjusted and non-adjusted), and without collapsing individuals into obese/non-obese groups (i.e., Bickel et al.). Lastly, consistent with our hypothesis and Jones & Rachlin (2009), there was a significant correlation between social and delay discounting, whereby delay discounting increased when sharing decreased. There was no relationship between social discounting and BMI when participants under 20 years of age had their BMI’s age-adjusted (i.e., adjusting weight category relative to other adolescents of the same gender and age), which is currently the best practice for calculating BMI (CDC, 2015b; WHO, 2017b). However, without age-adjusting BMI scores and dichotomizing individuals as either overweight/not (similar to Bickel et al. 2014), there was a significant relationship between social discounting and BMI in the hypothesized direction. That is, participants with a higher BMI (i.e., overweight and obese) shared significantly more than non-obese individuals. These results indicated a relationship between BMI and social discounting score, without a delay component. There are some noteworthy differences between the current study and previous discounting and BMI research. While WHO and CDC best practices encourage the use of age-

adjusted BMI scores, the current delay discounting–BMI literature is mixed in categorization of BMI class for individuals less than 20 years old. Some researchers examining delay discounting and BMI in a university sample have used the age-adjusted BMI categories (e.g., Bruce et al., 2011; Fields et al., 2011; Thamotharan et al., 2016), while others have not (e.g., Bickel et al., 2014; Epstein et al., 2014; Weller et al., 2008). In addition, much of the research examining social aspects of eating and its relationship with BMI has not used an age-adjusted BMI categorization (de Castro et al., 2012; Kaisari & Higgs, 2015; Koh & Pliner, 2009). This mixed methodology can make comparisons across studies difficult. In the current study, we used both BMI calculations and showed that the delay discounting–BMI relationship was unaffected by the different calculation, suggesting that this is a robust phenomenon. However, the relationship between social discounting and BMI was affected by the different ways of calculating BMI. Future studies should examine the relationship in social discounting and BMI in an adult (20 years or older) population to determine if the non-age-adjusted results found in the current study can be replicated and extended to age-adjusted BMI measures. Bickel et al. (2014) compared discounting rates in normal (BMI < 25), overweight, and obese individuals. The current study also included underweight individuals as a separate category. Although the effect sizes were small (d = between 0.01 and 0.36), the differences in delay discounting scores found in the current study were only between the underweight category and overweight and obese classifications. This may be a result of a relatively small sample size of underweight individuals. In addition, the range of delay discounting scores was smaller for underweight participants, relative to the other three BMI groups (Fig. 2, top panel). The interquartile range for underweight participants (0.011) was approximately one-third that of the other three BMI groups (0.031–0.038). Overall, these results are interesting given that decreased discounting is typically thought of as a desirable pattern of behaviors (Bickel, Jarmolowicz, Mueller, Koffarnus, & Gatchalian, 2012). However, individuals with an underweight BMI are at risk for a large number of health problems, such as anorexia nervosa, heart issues, and muscle weakness (Fairburn, Cooper, Shafran, Bohn, Hawker, Murphy, & Straebler, 2008). There is also evidence that shows an increased probability of dysfunctional behaviors, such as perfectionism, for college students who discounting the least in terms of delay discounting (Wainwright & Romanowich, 2016). Thus, while discounting too steeply may be detrimental, not discounting enough may also be maladaptive (Dickman, 1990; Mobini, Grant, Kass, & Yeomans, 2007; Siegel, 2012). However, because the effect sizes for the delay discounting–BMI relationship were small, replication of this finding with a larger sample of underweight individuals would strengthen the current findings. Regardless, the difference in delay

Psychol Rec Table 2 Means and standard deviations of BMI collapsed groups for log k (delay discounting) and AUCord (social discounting)

Age/gender adjusted Non-adjusted

1,2,3

BMI group

Delay discounting

Social discounting

Lower weight (underweight + normal BMI) Higher weight (overweight + obese BMI) Lower weight (underweight + normal BMI) Higher weight (overweight + obese BMI)

− 1.95 (0.69)1 − 1.81 (0.59)1 − 1.96 (0.70)2 − 1.80 (0.58)2

0.47 (0.29) 0.52 (0.26) 0.47 (0.28)3 0.53 (0.27)3

Superscript of same number indicates significant difference between groups

discounting scores highlights the need for underweight individuals to be the target of future discounting interventions, not just obese individuals. In the current study, there was a significant correlation between social and delay discounting, replicating previous research (Jones & Rachlin, 2009). This finding is additional evidence that these two discounting processes are significantly related, even when assessed through different formats. The current study used the Kirby et al. (1999) questionnaire for delay discounting, whereas Jones and Rachlin (2009) used delay discounting formats similar to the social discounting measure used in the current study. This is noteworthy given that the literature assessing the relationship between delay discounting and various health outcomes often use different discounting measures. Finding a similar relationship between social and delay discounting with two different measures of delay discounting provides additional evidence of the strength of this relationship. Additionally, there were significant correlations between delay discounting and both social discounting and BMI, but not across all three magnitudes. While all three magnitudes were negatively related to social discounting, only the medium and large magnitudes were positively related to BMI. Previous research comparing discounting rates of different commodities has also found differences within magnitudes. For example, Lemley, Kaplan, Reed, Darden, and Jarmolowicz (2016) found that men discounted beer more than money at small magnitudes, but not at medium or large magnitudes. No differences between commodities were found for women regardless of magnitude. The current study adds to the literature by replicating previous studies (i.e., Bickel et al., 2014; Jones & Rachlin, 2009) and adding more research on social aspects of eating behavior (e.g., Koh & Pliner, 2009). The current study has several strengths, including a large sample size and a relatively ethnically diverse sample, which suggests this finding is generalizable to multiple ethnic groups in the USA. However, there are also some weaknesses in the current study. The first weakness is the large number of participants excluded for nonsystematic discounting. After initial screening for time, BMI, and delay discounting, roughly 13% of participants were eliminated. However, after screening for non-systematic social

discounting using Johnson and Bickel’s (2008) criteria, an additional 32.2% of the data was excluded. Given the initial large sample, these reductions resulted in a sample size still adequate to power the tests in this study.1 However, the large number of non-systematic discounters may be indicative of the validity of the social discounting task format in an online study with college-aged students. Another potential weakness of the present study is the reliance on self-report for height and weight. While not feasible for a sample this large, measuring height and weight may have provided a more accurate measure of BMI, as both men and women tend to under-report weight and over-report height (Nyholm, Gullberg, Merlo, Lundqvist-Persson, Råstam, & Lindblad, 2007). Finally, in the current study, there were small effect sizes found throughout the results for both delay and social discounting which may be a result of the large sample size and have implications for the relevance of targeting delay discounting in obesity treatment. There is evidence that behavioral and cognitive interventions are successful in decreasing delay discounting rates (see Koffarnus, Jarmolowicz, Mueller, & Bickel, 2013 for review), and these methods have also been applied to obesity interventions in adolescents (Kulendran et al., 2014). However, given the small effect sizes found here, strategies that include reducing delay discounting may be less effective for decreasing obesity as they are for other issues related to delay discounting, such as drug addiction (Bickel, Yi, Landes, Hill, & Baxter, 2011) and smoking (Yi, Johnson, Giordano, Landes, Badger, & Bickel, 2008).

1 Analyses were re-run including all participants that provided a delay and social discounting score (i.e., including all inconsistent and non-systematic discounters), but without the 14 participants with invalid social discounting SurveyMonkey data (n = 787). This analyses showed one different outcome. Unlike the results reported with the exclusion criteria in place (n = 437), there was not a statistically significant difference for the non-adjusted BMI delay discounting outcome F(3, 783) = 1.64, p = 0.179. However, when collapsing BMI groups into either lower or higher weight, this difference was now statistically significant t(785) = − 2.25, p = 0.025. Consistent with the results using the exclusion criteria, there was not a significant difference for the non-adjusted BMI social discounting outcome F(3, 783) = 2.166, p = 0.091. Collapsing the BMI groups into either lower or higher weight still resulted in a significant social discounting difference t(785) = − 1.449, p = 0.015.

Psychol Rec

In summary, social discounting and BMI were not significantly related when BMI was age-adjusted. The current results, combined with Bickel et al. (2014) and Jones and Rachlin (2009), suggest that both BMI and social discounting are related to delay discounting. However, social discounting was only significantly related to BMI when non-age-adjusted BMI was used and participants were dichotomized as either normal or overweight during analyses. Future research should use a wider age range of participants and a larger sample of underweight individuals to replicate and strengthen the current results, especially as it relates to clinical populations. Finally, standardizing BMI calculations would allow for greater crossstudy comparisons in the health and discounting literature.

Compliance with Ethical Standards Conflict of Interest The authors declare that they have no conflict of interest. Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed Consent Informed consent was obtained from all individual participants included in the study.

References Andrade, L. F., & Petry, N. M. (2014). White problem gamblers discount delayed rewards less steeply than their African American and Hispanic counterparts. Psychology of Addictive Behaviors, 28(2), 599–606. Bickel, W. K., Jarmolowicz, D. P., Mueller, E. T., Koffarnus, M. N., & Gatchalian, K. M. (2012). Excessive discounting of delayed reinforcers as a trans-disease process contributing to addiction and other disease-related vulnerabilities: emerging evidence. Pharmacology & Therapeutics, 134, 287–297. Bickel, W. K., Wilson, A. G., Franck, C. T., Mueller, E. T., Jarmolowicz, D. P., Koffarnus, M. N., et al. (2014). Using crowdsourcing to compare temporal, social temporal, and probability discounting among obese and non-obese individuals. Appetite, 75, 82–89. Bickel, W. K., Yi, R., Landes, R. D., Hill, P. F., & Baxter, C. (2011). Remember the future: working memory training decreases delay discounting among stimulant addicts. Biological Psychiatry, 69(3), 260–265. https://doi.org/10.1016/j.biopsych.2010.08.017. Borges, A. M., Kuang, J., Milhorn, H., & Yi, R. (2016). An alternative approach to calculating area-under-the-curve (AUC) in delay discounting research. Journal of the Experimental Analysis of Behavior, 106, 145–155. Bruce, A. S., Black, W. R., Bruce, J. M., Daldalian, M., Martin, L. E., & Davis, A. M. (2011). Ability to delay gratification and BMI in preadolescence. Obesity, 19(5), 1101–1102. Centers for Disease Control and Prevention. (2015a). About adult BMI. Retrieved from https://www.cdc.gov/healthyweight/assessing/bmi/ adult_bmi/index.html Centers for Disease Control and Prevention. (2015b). About child and teen BMI. Retrieved from https://www.cdc.gov/healthyweight/ assessing/bmi/childrens_bmi/about_childrens_ bmi.html.

Dahl, A. K., Reynolds, C. A., Fall, T., Magnusson, P. K. E., & Pedersen, N. L. (2014). Multifactorial analysis of changes in body mass index across the adult life course: a study with 65 years of follow-up. International Journal of Obesity, 38(8), 1133–1141. de Castro, J. M. (1994). Family and friends produce greater social facilitation of food intake than other companions. Physiology & Behavior, 56(3), 445–455. de Castro, J. M., King, G. A., Duarte-Gardea, M., Gonzalez-Ayala, S., & Kooshian, C. H. (2012). Overweight and obese humans overeat away from home. Appetite, 59(2), 204–211. de Water, E., Cillessen, A. H. N., & Scheres, A. (2014). Distinct agerelated differences in temporal discounting and risk taking in adolescents and young adults. Child Development, 85(5), 1881–1897. de Wit, H., Flory, J. D., Acheson, A., McCloskey, M., & Manuck, S. B. (2007). IQ and nonplanning impulsivity are independently associated with delay discounting in middle-aged adults. Personality and Individual Differences, 42(1), 111–121. Dickman, S. J. (1990). Functional and dysfunctional impulsivity: personality and cognitive correlates. Journal of Personality and Social Psychology, 58(1), 95–102. Du, W., Green, L., & Myerson, J. (2002). Cross-cultural comparisons of discounting delayed and probabilistic rewards. The Psychological Record, 52(4), 479–492. Epstein, L. H., Jankowiak, N., Fletcher, K. D., Carr, K. A., Nederkoorn, C., Raynor, H. A., & Finkelstein, E. (2014). Women who are motivated to eat and discount the future are more obese. Obesity, 22(6), 1394–1399. Fairburn, C. G., Cooper, Z., Shafran, R., Bohn, K., Hawker, D. M., Murphy, R., & Straebler, S. (2008). Underweight and undereating. Cognitive behavior therapy and eating disorders; cognitive behavior therapy and eating disorders (pp. 147–182, Chapter xii, 324 Pages) Guilford Press, New York. Fields, S. A., Sabet, M., Peal, A., & Reynolds, B. (2011). Relationship between weight status and delay discounting in a sample of adolescent cigarette smokers. Behavioural Pharmacology, 22(3), 266– 268. Hendrickson, K. L., & Rasmussen, E. B. (2017). Mindful eating reduces impulsive food choice in adolescents and adults. Health Psychology, 36(3), 226–235. Howland, M., Hunger, J. M., & Mann, T. (2012). Friends don’t let friends eat cookies: effects of restrictive eating norms on consumption among friends. Appetite, 59(2), 505–509. Jackson, A. S., Janssen, I., Sui, X., Church, T. S., & Blair, S. N. (2012). Longitudinal changes in body composition associated with healthy ageing: men, aged 20-96 years. The British Journal of Nutrition, 107(7), 1085–1091. Johnson, M. W., & Bickel, W. K. (2002). Within-subject comparison of real and hypothetical money rewards in delay discounting. Journal of the Experimental Analysis of Behavior, 77(2), 129–146. https:// doi.org/10.1901/jeab.2002.77-129. Johnson, M. W., & Bickel, W. K. (2008). An algorithm for identifying nonsystematic delay-discounting data. Experimental and Clinical Psychopharmacology, 16, 264–274. Jones, B. A., & Rachlin, H. (2006). Social discounting. Psychological Science, 17, 283–286. Jones, B. A., & Rachlin, H. (2009). Delay, probability, and social discounting in a public goods game. Journal of the Experimental Analysis of Behavior, 91, 61–73. Kaisari, P., & Higgs, S. (2015). Social modelling of food intake. The role of familiarity of the dining partners and food type. Appetite, 86, 19– 24. Kaplan, B. A., Amlung, M., Reed, D. D., Jarmolowicz, D. P., McKerchar, T. L., & Lemley, S. M. (2016). Automating scoring of delay discounting for the 21- and 27-item monetary choice questionnaires. The Behavior Analyst, 39(2), 293–304.

Psychol Rec Kirby, K. N., Petry, N. M., & Bickel, W. K. (1999). Heroin addicts have higher discount rates for delayed rewards than non-drug-using controls. Journal of Experimental Psychology: General, 128(1), 78–87. https://doi.org/10.1037/0096-3445.128.1.78. Koffarnus, M. N., Jarmolowicz, D. P., Mueller, E. T., & Bickel, W. K. (2013). Changing delay discounting in the light of the competing neurobehavioral decision systems theory: a review. Journal of the Experimental Analysis of Behavior, 99(1), 32–57. Koh, J., & Pliner, P. (2009). The effects of degree of acquaintance, plate size, and sharing on food intake. Appetite, 52(3), 595–602. Kulendran, M., Vlaev, I., Sugden, C., King, D., Ashrafian, H., Gately, P., & Darzi, A. (2014). Neuropsychological assessment as a predictor of weight loss in obese adolescents. International Journal of Obesity, 38(4), 507–512. https://doi.org/10.1038/ijo. Lemley, S. M., Kaplan, B. A., Reed, D. D., Darden, A. C., & Jarmolowicz, D. P. (2016). Reinforcer pathologies: predicting alcohol related problems in college drinking men and women. Drug and Alcohol Dependence, 167, 57–66. Locey, M. L., Jones, B. A., & Rachlin, H. (2011). Real and hypothetical rewards in self-control and social discounting. Judgment and Decision making, 6(6), 522–564. Madden, G. J., Begotka, A. M., Raiff, B. R., & Kastern, L. L. (2003). Delay discounting of real and hypothetical rewards. Experimental and Clinical Psychopharmacology, 11(2), 139–145. Madden, G. J., & Bickel, W. K. (2010). Impulsivity: the behavioral and neurological science of discounting. Washington, DC: American Psychological Association. Mobini, S., Grant, A., Kass, A. E., & Yeomans, M. R. (2007). Relationships between functional and dysfunctional impulsivity, delay discounting and cognitive distortions. Personality and Individual Differences, 43(6), 1517–1528. Myerson, J., Green, L., & Warusawitharana, M. (2001). Area under the curve as a measure of discounting. Journal of the Experimental Analysis of Behavior, 76, 235–243. Nederkoorn, C., Smulders, F. T. Y., Havermans, R. C., Roefs, A., & Jansen, A. (2006). Impulsivity in obese women. Appetite, 47(2), 253–256. Nooyens, A. C. J., Visscher, T. L. S., Verschuren, W. M., Schuit, A. J., Boshuizen, H. C., van Mechelen, W., & Seidell, J. C. (2009). Age, period and cohort effects on body weight and body mass index in adults: the doetinchem cohort study. Public Health Nutrition, 12(6), 862–870. Nyholm, M., Gullberg, B., Merlo, J., Lundqvist-Persson, C., Råstam, L., & Lindblad, U. (2007). The validity of obesity based on selfreported weight and height: implications for population studies. Obesity, 15(1), 197–208. Olson, E. A., Rosso, I. M., Demers, L. A., Divatia, S., & Killgore, W. D. S. (2016). Sex differences in psychological factors associated with social discounting. Journal of Behavioral Decision Making, 29(1), 60–66. Rachlin, H., & Raineri, A. (1992). Irrationality, impulsiveness, and selfishness as discount reversal effects. In G. Loewenstein & J. Elster (Eds.), Choice over time (pp. 93–118). New York: Russell Sage Foundation. Rachlin, H., & Jones, B. A. (2008a). Altruism among relatives and nonrelatives. Behavioural Processes, 79, 120–123.

Rachlin, H., & Jones, B. A. (2008b). Social discounting and delay discounting. Journal of Behavioral Decision Making, 21, 29–43. Reas, D. L., Nygård, J. F., Svensson, E., Sørensen, T., & Sandanger, I. (2007). Changes in body mass index by age, gender, and socioeconomic status among a cohort of Norwegian men and women (1990-2001). BMC Public Health, 7, 269. Reimers, S., Maylor, E. A., Stewart, N., & Chater, N. (2009). Associations between a one-shot delay discounting measure and age, income, education and real-world impulsive behavior. Personality and Individual Differences, 47(8), 973–978. Romanowich, P., & Igaki, T. (2017). Effects of reward magnitude, tobacco, and alcohol use on social discounting in Japan and United States college students. The Psychological Record, 67, 345–353. Rozin, P. (2005). The meaning of food in our lives: a cross-cultural perspective on eating and well-being. Journal of Nutrition Education and Behavior, 37, S107–S112. Siegel, D. J. (2012). The developing mind: how relationships and the brain interact to shape who we are. New York: Guilford Press. Silverman, I. W. (2003). Gender differences in delay of gratification: a meta- analysis. Sex Roles, 49(9), 451–463. Steinberg, L., Graham, S., O'Brien, L., Woolard, J., Cauffman, E., & Banich, M. (2009). Age differences in future orientation and delay discounting. Child Development, 80(1), 28–44. Thamotharan, S., Lange, K., Ramos, A., & Fields, S. (2016). Examining weight concern and delay discounting in adolescent females. Eating Behaviors, 21, 228–231. Wainwright, K., & Romanowich, P. (2016). The relationship between perfectionism and discounting measures. Personality and Individual Differences, 101, 407–412. Weafer, J., & de Wit, H. (2014). Sex differences in impulsive action and impulsive choice. Addictive Behaviors, 39(11), 1573–1579. Weafer, J., De Arcangelis, J., & de Wit, H. (2015). Sex differences in behavioral impulsivity in at-risk and non-risk drinkers. Frontiers in Psychiatry, 6, 7. Weatherly, J. N., & McDonald, J. D. (2011). Delay discounting on different outcomes in a sample of American Indian and non-Indian college students. American Indian and Alaska Native Mental Health Research (Online), 17(2), 43–55. Weller, R. E., Cook III, E. W., Avsar, K. B., & Cox, J. E. (2008). Obese women show greater delay discounting than healthy-weight women. Appetite, 51(3), 563–569. Whelan, R., & McHugh, L. A. (2009). Temporal discounting of hypothetical monetary rewards by adolescents, adults, and older adults. The Psychological Record, 59(2), 247–258. World Health Organization. (2017a). BMI classification. Retrieved from http://apps.who.int/bmi/index.jsp?introPage=intro_3.html World Health Organization. (2017b). BMI-for-age (5–19 years). Retrieved from http://www.who.int/growthref/who2007_bmi_for_ age/en/ Yeomans, M. R., Leitch, M., & Mobini, S. (2008). Impulsivity is associated with the disinhibition but not restraint factor from the threefactor eating questionnaire. Appetite, 50(2–3), 469–476. Yi, R., Johnson, M. W., Giordano, L. A., Landes, R. D., Badger, G. J., & Bickel, W. K. (2008). The effects of reduced cigarette smoking on discounting future rewards: an initial evaluation. The Psychological Record, 58(2), 163–174.