CHILDHOOD OBESITY Month 2016 j Volume X, Number X Mary Ann Liebert, Inc. DOI: 10.1089/chi.2016.0122 ORIGINAL ARTICLE An Experimental Approach to Study Individual Differences in Infants' Intake and Satiation Behaviors during Bottle-Feeding Alison K. Ventura, PhD1,2 and Julie A. Mennella, PhD1 Abstract Background: As a group, bottle-fed infants are at higher risk for rapid weight gain compared with breast-fed infants. However, little is known about individual differences in feeding behaviors of bottle-feeding infants, as well as maternal and infant characteristics associated with bottle-feeding outcomes. Methods: We conducted a 2-day, within-subject study of 21 formula-feeding dyads; the within-subject factor was feeding condition: mother-led (ML; mothers were given the instruction to feed their infants as they typically would) vs. infant-led (IL; the experimenter ensured feeding began when infants signaled hunger and ended when they rejected the bottle on three consecutive occasions). Intake was determined by bottle weight; feedings were video-recorded and later analyzed to determine feeding duration and types of satiation behaviors displayed. Percent difference scores were calculated for each outcome as [((ML - IL)/IL) 100] to standardize differences among dyads. Mothers completed questionnaires of feeding styles and infant temperament. Results: On average, infants consumed *42% more formula during the ML- than IL-condition ( p = 0.03). However, notable variation existed in difference scores for intake (range = -52.8% to 268.9%; higher scores reflect greater intake during ML than IL). Stepwise regression illustrated that greater intakes during the ML-condition were predicted by the combination of: (1) higher infant age; (2) lower levels of infant rhythmicity and adaptability; (3) higher levels of infant positive mood; and (4) lower levels of maternal restrictive and responsive feeding styles. Conclusions: This objective, experimental approach illustrated that variation in bottle-feeding outcomes is associated with characteristics of both members of the dyad. Keywords: bottle-feeding; infant feeding behavior; infant temperament; maternal feeding practices; over-feeding; satiation excess weight gain.13 From the perspective of the infants, those who feed from a bottle are sensitive to what is in the bottle (e.g., the composition of the formula), and will modify intake accordingly.15,16 Further, bottle-fed infants can communicate hunger and satiation, but individual differences exist in the extent to which and the ways in which they communicate during feeding.14 Likewise, from the perspective of the mothers, some are more responsive to signals of hunger and satiation by their infants than others,17,18 and individual differences in infant feeding beliefs and practices exist and are associated with maternal characteristics, such as race, ethnicity, and obesity.19,20 To further our understanding of the dynamics of the mother-infant dyad during bottle-feeding and to identify which infants may be at risk for overfeeding, the present study used an experimental paradigm that measured Introduction nfants who feed from a bottle are more likely than those who breastfeed to exhibit rapid rates of growth during the first year postpartum,1-3 a known risk factor for childhood and adult obesity and other comorbidities.4-8 One likely mechanism underlying rapid rates of growth is that formula-fed infants consume more per feeding and over the course of a day than breast-fed infants.9 Such overfeeding may occur, because infants exert less control when feeding from a bottle, because their mothers are more responsive to contextual cues (e.g., amount in the bottle) than to their infants' hunger and satiation cues, or both.10-12 Despite these widely held hypotheses, research, although limited, suggests that bottle-feeding does not uniformly place infants at risk for overfeeding11,13,14 and I 1 Monell Chemical Senses Center, Philadelphia, PA. Department of Kinesiology, California Polytechnic State University, San Luis Obispo, CA. 2 1 2 infants' feeding and satiation behaviors during two conditions: (1) a ''mother-led'' (ML) feeding condition, wherein mothers were instructed to feed their infants as they typically would; and (2) an ''infant-led'' (IL) feeding condition, wherein maternal influences were minimized and the experimenter ensured that feeding began when the infant signaled hunger and ended when the infant signaled satiation by rejecting the bottle on three consecutive occasions.16,21,22 Using this within-subject and experimental approach, we aimed at directly assessing how the conditions of feeding and individual differences in characteristics of the infants (e.g., temperament; behavioral displays during feeding) and their mothers (e.g., weight status, feeding styles) relate to bottle-feeding outcomes. Subjects and Methods Participants Twenty-one mothers, whose healthy infants were formulafeeding, were recruited through ads in local newspapers, websites, and Philadelphia Women, Infants, and Children (WIC) offices. Inclusion criteria included that infants were born full term and did not have any medical conditions that interfered with feeding. Six additional dyads were recruited but then excluded because they either did not complete both testing days or did not comply with study procedures. All study procedures were approved by the Office of Regulatory Affairs at the University of Pennsylvania, and informed consent was obtained from each mother at study entry. This trial was registered at clinicaltrials.gov as NCT02284152. Methods Each mother-infant dyad was tested at the same time of day on 2 test days separated by 2.1 - 0.4 days. The 2 testing days were scheduled to occur at a time of day that each infant typically consumed formula. After acclimating to the room and personnel for *1 hour (average = 59.6 - 34.3 minutes), infants were video-recorded while their mothers fed them. Videos were captured using a video camera that was placed *12 feet in front of the dyad. The researcher remained in the room but was hidden behind a partition, out of view of the dyad, during the feeding. The researcher monitored the feeding via a monitor that was placed behind the partition and also out of view of the dyad. On both days, infants were fed the formula that they were currently feeding (e.g., Similac\u0002, Good Start Gentle Plus\u0002, Good Start Soy\u0002) and from the type of bottle with which they were familiar. So that mothers would be blind to the hypothesis and not be influenced by experiencing the IL feeding paradigm, the first day of testing was always the ML-condition in which mothers were only given the instruction to ''Please feed your infant as you normally would at home.'' The second day of testing was the ILcondition, which used a protocol that enables infants to control the pace and duration of feeding16,21,22 and minimizes maternal influences on infant feeding behaviors.23,24 In brief, during the IL-condition: (1) feeding sessions be- VENTURA AND MENNELLA gan when infants displayed signs of hunger (e.g., mouthing, rooting, fussing) and mothers verified that their infants were hungry; (2) mothers were instructed not to talk to their infants and to remain as neutral as possible; (3) mothers were instructed to feed at an IL pace (i.e., guided by infant feeding cues); and (4) feeding sessions ended when infants displayed signs of fullness (e.g., turning head and/or body away from the bottle, biting or chewing on the nipple, spitting the nipple) on at least three consecutive occasions. Formula intake (g) was determined by weighing the bottle both before and after each feeding, which was then converted to volume (mL) assuming a formula density of 1.03 g/mL.25-27 Video-Record Analyses Each video-recorded feeding (N = 42) was later analyzed frame by frame using an event recorder program (Observer XT, version 10.5, Wageningen, The Netherlands) to determine the duration of feeding and the frequency, timing, and type of behaviors infants used to signal satiation. For each video, we focused on the second half of each meal. Raters (N = 2) coded 11 mutually exclusive behavior types (see Ventura et al.14 for more detail): (1) waves arm(s); (2) displays negative facial expressions; (3) leans away or arches back; (4) turns head and/or body away from bottle; (5) bites or chews nipple; (6) pushes bottle away; (7) spits out nipple; (8) gags, coughs, or chokes; (9) spits out formula or spits up; (10) crying bout; and (11) sleeping bout. Inter-coder reliability was established by the common scoring of a total of 10 feedings by both raters. The mean Pearson's rho for the correspondence between the frequency, timing, and type of behaviors coded in each of the double-coded videos was q(8df) >0.80, indicating good reliability between raters. Characteristics of Mothers and Infants Mothers were queried on sociodemographic characteristics and completed questionnaires to obtain measures of infant temperament (the Infant Temperament Questionnaire28,29) and maternal feeding styles (the Infant Feeding Styles Questionnaire30). Because we aimed to identify characteristics of mothers and infants that explained individual differences in intake and feeding behaviors of the infant during the two conditions, our analyses focused on the temperament subscales that describe infant behaviors during feeding: activity level, rhythmicity, approach, adaptability, and mood28,29; and on the three maternal feeding styles that are the most applicable to bottle-feeding infants: (1) pressuring, defined as the extent to which the mother encourages her infant to finish the bottle and uses food as a soothing method; (2) restrictive, defined as the extent to which the mother limits the infant's intake; and (3) responsive, defined as the extent to which the mother recognizes and feeds in response to her infant's hunger and satiation cues.30 CHILDHOOD OBESITY Month 2016 Infants' weight and recumbent length were measured in triplicate by using standard anthropometric techniques,31 a scale accurate to 0.1 kg (Scale Tronix, White Plains, NY), and an infantometer accurate to 0.1 cm (Harpenden Infantometer 702, Crymych; Dyfed). Mothers' weight and height were also measured in triplicate31 (Health-O-Meter; Sunbeam Products, Inc., Boca Raton, FL). Infant weight and length measures were normalized to age- and sex-specific zscores [i.e., weight-for-age z scores (WAZ)],32 and maternal BMI (BMI = weight in kg/height in m2) was determined. Statistical Analysis Analyses were conducted using SAS, version 9.4 (SAS Institute, Inc., Cary, NC). Data were first assessed for normality. Distributions for the number of satiation behaviors displayed during the second half of the feedings were positively skewed; thus, they were normalized by using log transformations before analyses. After analyses, data were back-transformed by calculating the antilog or square of the estimate.33 For both conditions, outcome measures included infant intake (mL), intake per kilogram of body weight (mL/kg), meal duration (minutes), rate of feeding (mL/min), number of satiation behaviors displayed during the second half of the feedings, and latency to first satiation behavior during the second half of the feedings. We then conducted repeated-measures analysis of variance (ANOVA) with condition (ML vs. IL) as the within-subject factor to determine whether these measures differed by feeding condition. Time elapsed since the last feeding and infant age were included as time-varying covariates in all repeated measures ANOVA models. Second, we examined individual differences in the discrepancy between the ML- and IL-conditions by calculating, for each infant, a percent difference score for each outcome measure [((ML - IL)/IL) 100]; this allowed us to standardize differences between the ML- and ILconditions across infants. Higher percent difference scores reflect greater response (e.g., greater intake or number of satiation behaviors) during the ML- when compared with the IL-condition. Binomial distribution tests were used to determine whether the proportion of infants who showed greater response during the ML- than IL-condition was greater than chance. t-Tests were used to determine whether each percent difference score was greater than zero. Third, to explore which infants were at higher risk for feeding more during the ML-condition, stepwise regression was used to determine the combination of infant and maternal characteristics that best predicted percent difference scores for intake (% DIFF-intake). The infant and maternal variables entered into the stepwise regression analysis were those previously associated with infant risk for overfeeding and rapid weight gain.17-20,34-39 The order of entry for each model variable was determined by the strength of the association between each predictor and the outcome and, with each iteration of the analysis, the most significant variable was added or the least significant var- 3 iable was removed.40 Results are presented as means or least squared means - standard errors (SEs). p-Values <0.05 indicated significant effects, and p-values <0.10 indicated trends. Results Sample Characteristics Table 1 presents the dyads' characteristics. The infants (13 girls, 8 boys) were, on average, 21 weeks (4.8 - 0.5 months) of age and had an average WAZ of 0.4 - 0.2. The mothers (61.1% multiparous) were, on average, 28.0 - 1.4 years of age and had a mean BMI of 29.0 - 1.1; 24% were classified as overweight, and 48% were classified as obese. The majority of mothers was non-Hispanic Black (62%) and had a high school degree or higher (81%). Approximately half (52%) reported an annual family income >$35,000. Table 1. Participant Characteristics (N = 21) Characteristics Mean (SE) or percent (n) Infants Age at study entry (months) 4.8 (0.5) Infant sex (% female) 61.9 (13) WAZ at study entry 0.4 (0.2) Mothers Age at study entry (years) 28.0 (1.4) Percent primiparous 38.9 (7) BMI 29.0 (1.1) Percent overweight (BMI = 25-29.9) 23.8 (5) Percent obese (BMI 30) 47.6 (10) Race/ethnicity Non-Hispanic White 38.1 (8) Non-Hispanic Black 61.9 (13) Education level Did not complete high school 19.1 (4) High school degree 28.6 (6) Some college or vocational degree 38.1 (8) Bachelor or graduate degree 14.3 (3) Family income <$15,000/year 33.3 (7) $15,000 to <$35,000/year 14.2 (3) $35,000 to <$75,000/year 47.6 (10) >$75,000/year 4.8 (1) SE, standard error; WAZ, weight-for-age z-score; DWAZ, change in weight-for-age z-score. 4 VENTURA AND MENNELLA Infant Intake and Rate of Feeding Infant Satiation Behaviors There was no difference between conditions for time elapsed since the last feeding (ML: 2.6 - 0.1 hours vs. IL: 2.8 - 0.2 hours; p = 0.47); testing occurred, on average, 2.7 - 0.1 hours since the previous feeding. When we compared how much formula infants fed and their rates of feeding during the ML- and IL-conditions at the level of the group, differences between conditions for formula intake (mL), intake per kilogram of body weight (mL/kg), and meal duration (minutes) did not reach significance (Table 2). However, infants trended toward exhibiting a greater rate of feeding (mL/min) during the ML- compared with IL-condition ( p = 0.09). On average, infants consumed 139.9 - 9.6 mL of formula per feed (20.3 - 1.5 mL/ kg) at a rate 16.7 - 2.1 mL/min; the formula meal lasted 10.0 - 0.8 minutes. When differences between the ML- and IL-conditions were standardized across infants, there was a wide variation in how condition (ML vs. IL) modified infant feeding outcomes. As shown in Figure 1, the percent difference scores for intake (% DIFF-intake) during the ML- compared with the IL-condition ranged from -52.8% to 268.9%; recall positive scores indicate that the infant consumed more during the ML-condition compared with the IL-condition, whereas negative scores indicate that the infant consumed less. Sixty-seven percent of infants (14 of 21) consumed more during the ML- than IL-condition; this proportion trended toward being significantly different from chance ( p = 0.09). The average % DIFF-intake score (41.8% - 17.8%) was significantly greater than zero [t(20) = 2.35, p = 0.03; Table 2]. Results did not change when we accounted for the body weight of the child (i.e., intake/kg; p = 0.04). Percent difference scores for rate of feeding were also significantly greater than zero ( p = 0.01). Infants were consistent in the overall number, frequency, and timing of satiation behaviors displayed during the second half of the ML- and IL-conditions (Table 2). In addition, the latency to display the first satiation behavior did not differ between the two conditions ( p = 0.62). When considering each satiation behavior individually, there were no significant differences in the frequency of 10 of 11 behaviors (i.e., waving arms; negative facial expressions; leans away or arches back; turns away from bottle; bites or chews nipple; pushes bottle away; spits out nipple; gags, coughs, and chokes; spits out formula or spits up; crying bout; or sleeping bout; all p > 0.05). Displays of leaning away or arching back was displayed significantly fewer times during ML- than IL-condition (0.2 - 0.1 vs. 1.7 - 0.5 displays, respectively; p = 0.01). On average, infants displayed 13.7 - 1.1 satiation behaviors, which began 6.1 - 3.3 minutes after the start of the feeding. The number of satiation behaviors displayed during MLcondition was significantly and positively associated with the number of satiation behaviors displayed during ILcondition [r(19) = 0.42, p = 0.05], suggesting a consistency in the behavioral phenotype of the infant. The number of satiation behaviors displayed during each condition was not associated with how much formula the infant consumed during that particular condition (ML: p = 0.47 and IL: p = 0.64). When differences between the ML- and IL-conditions were standardized across infants, the percent difference scores for each infant's satiation behaviors displayed during the second half of ML- compared with IL-condition (% DIFF-behaviors) ranged from -94.3% to 320.0% (mean = 6.9 - 21.8%); recall positive scores indicate that the infant displayed more satiation behaviors during the second half of the ML- compared with the IL-condition, whereas negative Table 2. Least-Squared Means (SE) for Infant Feeding Behaviors during Mother-Led and Infant-Led Conditions (n = 21) ML-Condition IL-Condition pa 148.1 (11.9) 131.6 (16.0) 0.32 41.8 (17.8) 0.03 21.2 (1.9) 19.4 (2.5) 0.45 40.5 (18.1) 0.04 Meal duration (minutes) 9.2 (0.9) 10.9 (0.9) 0.15 -6.3 (10.5) 0.56 Rate of feeding (mL/min) 18.9 (2.5) 14.4 (1.7) 0.09 50.6 (18.5) 0.01 Number of satiation behaviors displayed during the second half of the meal 11.6 (1.3) 16.1 (1.2) 0.16 6.9 (21.8) 0.75 Latency to first satiation behavior (minutes) during the second half of the meal 5.9 (0.8) 6.3 (0.7) 0.62 -20.9 (18.1) 0.27 Duration of the second half of the meal (minutes) 5.3 (0.5) 6.1 (0.7) 0.38 -6.3 (10.5) 0.56 Intake (mL) Intake per kg body weight (mL/kg) a % DIFFb For the mean difference between the IL and the ML feeding; all models were controlled for time since last feeding and infant age. Percent difference score, calculated as [(ML - IL)/IL] 100. c For t-tests to determine whether the percent difference score was significantly different than zero. % DIFF, percent difference score; IL, infant-led; ML, mother-led. b pc CHILDHOOD OBESITY Month 2016 5 Figure 1. Percent difference scores for infants' intakes during ML vs. IL feeding conditions. Percent difference is defined as [(ML - IL)/ IL] 3 100. ML, mother-led; IL, infant-led. scores indicate that the infant displayed fewer. Sixty-two percent (13 of 21) of infants displayed more behaviors (i.e., had a positive % DIFF-behaviors scores) during the ML- than the IL-condition; this proportion was not significantly different from chance ( p = 0.19). In addition, the average % DIFF-behaviors score was not significantly greater than zero [t(20) = 0.32, p = 0.75]. Predictors of Percent Differences in Intake During ML- vs. IL-Feeding Conditions Stepwise multivariate regression was used to explore which combination of mother and infant characteristics best predicted infants' % DIFF-intake. Predictor variables included % DIFF-behaviors; infants' age and WAZ at testing, and temperament scores; and mothers' age and BMI at study entry, and levels of pressuring, restrictive, and responsive feeding styles. Table 3 illustrates the best-fit model for prediction of % DIFF-intake scores. The final model explained 75.4% of the variance in infants' % DIFF-intake and was statistically significant [F(6,19) = 6.66, p = 0.0021]. The variables that remained in the final model included infants' age, mothers' perceptions of infants' level of rhythmicity, adaptability, and mood, and mothers' level of restrictive and responsive feeding style. This model suggests that greater infant intakes during the ML- than the IL-condition were predicted by the combination of: (1) older age of infant; (2) lower levels of infant rhythmicity and adaptability; (3) higher levels of infant positive mood; and (4) lower levels of maternal restrictive and responsive feeding styles. Discussion The present research illustrated a novel approach to study infant behaviors and satiation during bottle-feeding, Table 3. Final Regression Model Predicting Percent Difference for Infant Intakes (% DIFF-Intake)a during Mother-Led vs. Infant-Led Conditions b (SE) Predictor F Infant characteristics Age 43.47 (12.78) 11.58* Rhythmicityb 80.31 (19.50) 16.95** Adaptabilityc 51.62 (19.03) 7.36* -66.73 (19.21) 12.07** -43.87 (13.03) 11.35** -76.78 (15.90) 23.30** Mood d Maternal characteristics Restrictive feedinge Responsive feeding e Model F-value = 6.66, p = 0.0021. Model R2 = 0.75. a Percent difference scores for infant's formula intake during ML vs. IL feeding = [(ML - IL)/IL] 100. b Higher scores = more arrhythmic; from the Infant Temperament Questionnaire.26,27 c Higher scores = more nonadaptable; from the Infant Temperament Questionnaire.26,27 d Higher scores = more negative mood; from the Infant Temperament Questionnaire.26,27 e Higher scores = greater levels of restrictive or responsive feeding style; from the Infant Feeding Styles Questionnaire.28 *p < 0.05; **p < 0.01. ML, mother-led; IL, infant-led. 6 which can be used to identify sources of individual differences in bottle-feeding outcomes. We illustrated that infants can and do signal satiation during bottle-feeding and are consistent in their display of satiation behaviors during different feeding conditions. Although how much formula infants consumed and how long they spent feeding in the present study were characteristic of infants in this age range,9,14,16,41 we noted wide variation in infant feeding outcomes when comparing ML with IL feeding conditions. This variation was associated with characteristics of both members of the dyad. Our exploration of the maternal and infant characteristics that was associated with variation in formula feeding outcomes provides some insights into our understanding of which dyads are at higher vs. lower risk of consuming more during typical, ML feeding conditions compared with IL-conditions. With respect to maternal characteristics, we noted significant associations between mothers' self-reported feeding styles and infants' relative intake during ML- vs. IL-conditions. Specifically, higher levels of restrictive feeding style were associated with lower intakes, whereas lower levels of responsive feeding style were associated with higher intakes during the ML- compared with the IL-condition. These findings are consistent with previous research illustrating that restrictive feeding styles are associated with lower daily energy intake for infants, whereas feeding styles that are less responsive to infant hunger and satiation cues are associated with greater daily energy intake.39 In addition, although it is commonly reported that responsive feeding is characteristic of breastfeeding mothers and controlling or restrictive feeding is more characteristic of bottle-feeding mothers;10,42-45 our study illustrates that this dichotomy may oversimplify associations between feeding mode and feeding styles. Rather, among our sample of bottle-feeding mothers, we noted variation in mothers' level of restrictive and responsive feeding styles and that greater levels of responsive feeding and lower levels of restrictive feeding were associated with a closer match between ML and IL feeding conditions for feeding outcomesin other words, for some bottle-feeding mothers, the ML feeding was essentially an IL feeding, likely because these mothers fed in response to their infants' cues. Thus, bottle-feeding and responsive feeding are not mutually exclusive, and future studies should aim at understanding how to promote responsive feeding styles for all mothers who choose to bottle-feed.46 When considering infant characteristics, we found that older age, lower levels of rhythmicity and adaptability, and more positive mood were predictive of greater intakes during ML- compared with IL-conditions. These associations are consistent with previous research illustrating that what the child ''brings to the table'' may influence the dynamics of feeding and be determined, in part, by the children's past experiences with their caregivers during feeding.22,36,47,48 For example, it is well documented that children learn to overeat in response to parent feeding VENTURA AND MENNELLA practices that are overly restrictive or prompt children to eat in the absence of hunger (see Rollins et al.49 for a review); similarly, our finding that older infants had greater discrepancies between ML compared with IL feeding conditions may represent the greater experience that older infants have with their mothers' feeding styles, resulting in a learned tendency to consume more during ML feeding. A growing body of research has also recognized associations between child temperament and both children's eating behaviors34-38 and parents' feeding practices.50,51 This research has highlighted that children who score higher on measures of temperamental surgency or impulsivity also show greater food responsiveness and risk for overeating,34,35,37,38 whereas children who score higher on measures of temperamental difficulty or negativity have parents who report greater use of food to soothe and are at greater risk for overfeeding.50,51 Consistent with these findings, our analyses revealed that dimensions of surgency/impulsivity (i.e., greater levels of positive mood52) and difficult/negative temperament (i.e., lower levels of rhythmicity and adaptability28) were predictive of greater intakes during ML compared with IL feeding conditions. However, given that our measure of temperament was self-reported by mothers, it is unclear whether infant temperament is a driver of mothers' feeding practices and infant intake, or whether mothers' perceptions of their infants' temperament are shaped by feeding interactions and outcomes.22 Further research that appreciates what both mothers and infants contribute to feeding interactions is needed to better understand causal mechanisms linking maternal and infant characteristics and feeding outcomes. It is likely that additional factors not assessed in the present study also contributed to individual differences in feeding outcomes, and some aspects of our methodology (e.g., the order of conditions and conditions of testing) may have contributed to variation in feeding outcomes. First, we intentionally designed our study such that the MLcondition preceded the IL-condition in an attempt to minimize mothers' expectations regarding how they should feed their infants. However, it is possible that some of the infant behaviors observed on the first day of testing (e.g., faster suck rates) may have been in response to the novel feeding setting, rather than in response to the mothers' feeding behaviors. Second, we used an established protocol to create an IL feeding condition, wherein the feeding occurred in response to infant hunger and fullness cues and the influence of the mother was minimized, in an effort to allow the infant to express behaviors that typical ML feeding conditions may inhibit or modify. However, it is possible that mothers' lack of talking and maintained neutrality during the IL-condition caused infants to behave differently in response to the novelty of this condition.53,54 In line with this possibility, Lumeng et al.23 illustrated that when 7- to 14-week-old infants were bottle-fed in an IL manner by trained research nurses, infants consumed 43% more CHILDHOOD OBESITY Month 2016 formula when the nurses interacted with them compared with when the nurses were instructed to remain neutral,23 indicating that social interaction was a driver of intake for young infants. Whether these findings translate to mother- infant interactions was not directly tested, but it is plausible that social interaction from mothers would similarly influence infant intakefor example, as a driver of faster sucking rates seen during the ML feedingand that greater social interaction during the ML feeding conditions or variation in the degree to which mothers interacted with their infants may have led, in part, to variation in how much formula infants consumed. It should also be noted that, although our small sample size allowed for in-depth coding of infant behaviors, it may have also led to Type II errors due to low power. Thus, further research that uses the approach illustrated in the present study, but also includes a larger sample size and observational measures of maternal behaviors, is needed to further understand the effects of mother-infant social interaction, mothers' feeding practices, and infants' characteristics on bottle-feeding outcomes in the short term, and infants' weight gain trajectories and developmental outcomes in the long term. Conclusions The experimental paradigm used here allows for the observational assessment of independent contributions of the infant and mother to short-term bottle-feeding outcomes. Our approach complements the research of others,1,2,23,55-57 in that we objectively assessed infant intake during a typical feeding, wherein mothers were given no instruction regarding how or how much to feed, and then compared feeding outcomes with an IL feeding paradigm during which infants determined the pace and duration of feeding and the influence of their mothers was minimized.16,21,22 This within-subject approach allows for observational measurements of the behaviors of infants and could be applied to similarly assess the behaviors of caregivers to better understand the unique contributions of each member of the dyad during bottlefeeding interactions.22,47,58-60 Further research using this paradigm has the potential to increase knowledge on how to improve feeding interactions and outcomes for bottlefeeding dyads. Acknowledgments The project described was supported by NIH grants R01HD072307 and R01HD37119; American Recovery and Reinvestment Act supplement 3R01HD037119-10S1; and National Research Service Award F32HD063343 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The funding agencies had no role in the design and 7 conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the article. The authors thank Loma Inamdar, Mariya Keselman, Sehris Khawaja, Jillian Fink, Shawna Comalli, and Nicole Halfin (whose position was created by ARRA supplement 3R01HD037119-10S1) for technical assistance. They also thank Dr. Linda Kilby and the staff at the Philadelphia WIC Program for their assistance with subject recruitment. Author Disclosure Statement No competing financial interests exist. References 1. Li R, Fein SB, Grummer-Strawn LM. Association of breastfeeding intensity and bottle-emptying behaviors at early infancy with infants' risk for excess weight at late infancy. Pediatrics 2008; 122(Suppl):S77-S84. 2. Li R, Magadia J, Fein SB, Grummer-Strawn LM. Risk of bottlefeeding for rapid weight gain during the first year of life. Arch Pediatr Adolesc Med 2012;166:431-436. 3. Mihrshahi S, Battistutta D, Magarey A, Daniels LA. Determinants of rapid weight gain during infancy: Baseline results from the NOURISH randomised controlled trial. BMC Pediatr 2011; 11:99. 4. Evelein AM, Visseren FL, van der Ent CK, et al. Excess early postnatal weight gain leads to thicker and stiffer arteries in young children. J Clin Endocrinol Metab 2013;98:794-801. 5. Stettler N, Iotova V. Early growth patterns and long-term obesity risk. Curr Opin Clin Nutr Metab Care 2010;13:294-299. 6. Skilton MR, Marks GB, Ayer JG, et al. Weight gain in infancy and vascular risk factors in later childhood. Pediatrics 2013;131: e1821-e1828. 7. Baird J, Fisher D, Lucas P, et al. Being big or growing fast: Systematic review of size and growth in infancy and later obesity. Br Med J 2005;331:929. 8. Ong KK, Emmett P, Northstone K, et al. Infancy weight gain predicts childhood body fat and age at menarche in girls. J Clin Endocrinol Metab 2009;94:1527-1532. 9. Heinig MJ, Nommsen LA, Peerson JM, et al. Energy and protein intakes of breast-fed and formula-fed infants during the first year of life and their association with growth velocity: The DARLING Study. Am J Clin Nutr 1993;58:152-161. 10. Crow RA, Fawcett J, Wright P. Maternal behavior during breastand bottle-feeding. J Behav Med 1980;3:259-277. 11. Ventura AK, Golen RP. A pilot study comparing opaque, weighted bottles with conventional, clear bottles for infant feeding. Appetite 2015;85:178-185. 12. Dunn J. Feeding and sleeping. In: Rutter M (ed), The Scientific Foundations of Developmental Psychiatry. London: Heinemann Medical Books, 1980, pp. 427-455. 13. Mennella JA, Ventura AK, Beauchamp GK. Differential growth patterns among healthy infants fed protein hydrolysate or cowmilk formulas. Pediatrics 2011;127:110-118. 14. Ventura AK, Inamdar LB, Mennella JA. Consistency in infants' behavioural signalling of satiation during bottle-feeding. Pediatr Obes 2015;10:180-187. 8 15. Fomon SJ, Filer LJ, Jr, Thomas LN, et al. Relationship between formula concentration and rate of growth of normal infants. J Nutr 1969;98:241-254. 16. Ventura AK, Beauchamp GK, Mennella JA. Infant regulation of intake: The effect of free glutamate content in infant formulas. Am J Clin Nutr 2012;95:875-881. 17. Hodges EA, Hughes SO, Hopkinson J, Fisher JO. Maternal decisions about the initiation and termination of infant feeding. Appetite 2008;50:333-339. 18. Hodges EA, Johnson SL, Hughes SO, et al. Development of the responsiveness to child feeding cues scale. Appetite 2013;65:210- 219. 19. Gross RS, Mendelsohn AL, Fierman AH, et al. Maternal infant feeding behaviors and disparities in early child obesity. Child Obes 2014;10:145-152. 20. Thompson AL. Intergenerational impact of maternal obesity and postnatal feeding practices on pediatric obesity. Nutr Rev 2013;71 Suppl 1:S55-S61. 21. Mennella JA, Beauchamp GK. Developmental changes in the acceptance of protein hydrolysate formula. J Dev Behav Pediatr 1996;17:386-391. 22. Forestell CA, Mennella JA. More than just a pretty face. The relationship between infant's temperament, food acceptance, and mothers' perceptions of their enjoyment of food. Appetite 2012; 58:1136-1142. 23. Lumeng JC, Patil N, Blass EM. Social influences on formula intake via suckling in 7 to 14-week-old-infants. Dev Psychobiol 2007; 49:351-361. 24. Gunnar MR, Stone C. The effects of positive maternal affect on infant responses to pleasant, ambiguous, and fear provoking toys. Child Dev 1984;55:1231-1236. 25. Abbot Nutrition. Similac advance nutrition information. 2016. Available at http://abbottnutrition.com/brands/products/similacadvance Last accessed February 19, 2016. 26. Enfamil. Enfamil infant nutrition information. 2016. Available at www.enfamil.com/products/routine-feeding/enfamil-infant Last accessed February 19, 2016. 27. Gerber. Gerber good start nutrition information. 2016. Available at http://medical.gerber.com/products/formulas/good-start-gentle Last accessed February 19, 2016. 28. Carey WB, McDevitt SC. Revision of the infant temperament questionnaire. Pediatrics 1978;61:735-739. 29. Medoff-Cooper B, Carey WB, McDevitt SC. The early infancy temperament questionnaire. J Dev Behav Pediatr 1993;14:230- 235. 30. Thompson AL, Mendez MA, Borja JB, et al. Development and validation of the Infant Feeding Style Questionnaire. Appetite 2009;53:210-221. 31. Lohman T, Roche AF, Martorell R. Anthropometric Standardization Reference Manual. Human Kinetics Books: Champaign, IL, 1988. 32. World Health Organization. Child Growth Standards: Length/ Height-for-Age, Weight-for-Age, Weight-for-Length, Weight-forHeight, and Body Mass Index-for-Age: Methods and Development. World Health Organization, Geneva, 2006. 33. Bland JM, Altman DG. Transforming data. Br Med J 1996; 312:770. 34. Leung CY, Lumeng JC, Kaciroti NA, et al. Surgency and negative affectivity, but not effortful control, are uniquely associated with obesogenic eating behaviors among low-income preschoolers. Appetite 2014;78:139-146. VENTURA AND MENNELLA 35. Leung CY, Miller AL, Kaciroti NA, et al. Low-income preschoolers with higher temperamental surgency enjoy and respond more to food, mediating the path to higher body mass index. Pediatr Obes 2016;11:181-186. 36. Moding KJ, Birch LL, Stifter CA. Infant temperament and feeding history predict infants' responses to novel foods. Appetite 2014; 83:218-225. 37. Nederkoorn C, Braet C, Van Eijs Y, et al. Why obese children cannot resist food: The role of impulsivity. Eat Behav 2006;7:315- 322. 38. Nederkoorn C, Dassen FC, Franken L, et al. Impulsivity and overeating in children in the absence and presence of hunger. Appetite 2015;93:57-61. 39. Thompson AL, Adair LS, Bentley ME. Pressuring and restrictive feeding styles influence infant feeding and size among a lowincome African-American sample. Obesity (Silver Spring) 2013; 21:562-571. 40. Neter J, Kutner MH, Nachtsheim CJ, Wasserman W. Applied Linear Statistical Models, 3rd ed. McGraw Hill: Boston, MA, 1996. 41. Fomon SJ. Nutrition of Normal Infants. Mosby-Year Book, Inc.: St. Louis, MO, 1993. 42. DiSantis KI, Hodges EA, Fisher JO. The association of breastfeeding duration with later maternal feeding styles in infancy and toddlerhood: A cross-sectional analysis. Int J Behav Nutr Phys Act 2013;10:53. 43. Fisher JO, Birch LL, Smiciklas-Wright H, Picciano MF. Breastfeeding through the first year predicts maternal control in feeding and subsequent toddler energy intakes. J Am Diet Assoc 2000; 100:641-646. 44. Jansen E, Mallan KM, Byrne R, et al. Breastfeeding duration and authoritative feeding practices in first-time mothers. J Hum Lact 2016;32:498-506. 45. Wright P, Fawcett J, Crow R. The development of differences in the feeding behaviour of bottle and breast fed human infants from birth to two months. Behav Processes 1980;5:1-20. 46. Institute of Medicine. Early Childhood Obesity Prevention Policies. National Academies Press, Washington, DC, 2011. 47. Hurley KM, Cross MB, Hughes SO. A systematic review of responsive feeding and child obesity in high-income countries. J Nutr 2011;141:495-501. 48. Birch LL, Anzman SL. Learning to eat in an obesogenic environment: A developmental systems perspective on childhood obesity. Child Dev Perspect 2010;4:138-143. 49. Rollins BY, Savage JS, Fisher JO, Birch LL. Alternatives to restrictive feeding practices to promote self-regulation in childhood: A developmental perspective. Pediatr Obes 2016;11:326-332. 50. Stifter CA, Anzman-Frasca S, Birch LL, Voegtline K. Parent use of food to soothe infant/toddler distress and child weight status. An exploratory study. Appetite 2011;57:693-699. 51. Stifter CA, Moding KJ. Understanding and measuring parent use of food to soothe infant and toddler distress: A longitudinal study from 6 to 18 months of age. Appetite 2015;95:188-196. 52. Rothbart MK, Ahadi SA, Hershey KL, Fisher P. Investigations of temperament at three to seven years: The Children's Behavior Questionnaire. Child Dev 2001;72:1394-1408. 53. Tronick E, Als H, Adamson L, et al. The infant's response to entrapment between contradictory messages in face-to-face interaction. J Am Acad Child Psychiatry 1978;17:1-13. 54. Tronick ED, Als H, Brazelton TB. Mutuality in mother-infant interaction. J Commun 1977;27:74-79. CHILDHOOD OBESITY Month 2016 55. Worobey J, Islas Lopez M, Hoffman DJ. Maternal behavior and infant weight gain in the first year. J Nutr Educ Behav 2009;41: 169-175. 56. Bartok CJ. Babies fed breastmilk by breast versus by bottle: A pilot study evaluating early growth patterns. Breastfeed Med 2011; 6:117-124. 57. Disantis KI, Collins BN, Fisher JO, Davey A. Do infants fed directly from the breast have improved appetite regulation and slower growth during early childhood compared with infants fed from a bottle? Int J Behav Nutr Phys Act 2011;8:89. 58. Sacco LM, Bentley ME, Carby-Shields K, et al. Assessment of infant feeding styles among low-income African-American mothers: Comparing reported and observed behaviors. Appetite 2007; 49:131-140. 59. Wasser HM, Thompson AL, Siega-Riz AM, et al. Who's feeding baby? Non-maternal involvement in feeding and its association 9 with dietary intakes among infants and toddlers. Appetite 2013; 71:7-15. 60. Elford L, Brown A. Exploring child-feeding style in childcare settings: How might nursery practitioners affect child eating style and weight? Eat Behav 2014;15:314-317. Address correspondence to: Alison K. Ventura, PhD California Polytechnic State University, San Luis Obispo Kinesiology Department One Grand Avenue San Luis Obispo, CA 93407 E-mail: akventur@calpoly.edu