Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Jun 1.
Published in final edited form as: Brain Cogn. 2023 Apr 19;168:105985. doi: 10.1016/j.bandc.2023.105985

Intraindividual variability in brain activation—a novel correlate of obesity risk among female college students

Lance O Bauer 1,*
PMCID: PMC10175168  NIHMSID: NIHMS1894899  PMID: 37084591

Abstract

There are published data describing impairments in the brain function of adolescents or young adults who have a genetic or familial predisposition for obesity. From these descriptions, it is often assumed that the impairments are appropriately captured by a central tendency estimate and therefore consistently detectable. The present study questions this assumption and shows that the variability in brain function over the time course of a cognitive task is a better predictor of familial risk than its central tendency. Sixty-nine female young adults lacking an obese parent and 24 female young adults with an obese parent were compared on the average amplitude and inter-trial variability (ITV) in amplitude of their P300 electroencephalographic responses to rarely-occurring stimuli during a selective attention task. Simple group comparisons revealed statistically significant findings with effect sizes that were markedly greater for analyses of P300 ITV versus P300 average amplitude. It is suggested that the elevation in P300 ITV among young adults with familial risk indicates temporal instability in systems responsible for the maintenance of attention. These fluctuations may episodically disrupt their attention to satiety cues as well as other cues that influence behavior regulation.

Keywords: Obesity, family history, genetic, P300, event related potential, intraindividual variability

1. Introduction

It is well known that parental obesity is related to offspring obesity (Rath et al., 2016). In comparison to normal-weight mothers, mothers who are obese before pregnancy are more likely by a factor of 6.96 to give birth to a child who becomes obese before age 22. For obese versus normal-weight fathers, the odds-ratio for the same outcome is 3.75.

The present investigation was designed to measure a personal characteristic that could contribute to this parent-offspring association. The characteristic is a subtle difference in brain function that may be heritable and adversely affect one’s ability to attend to visual and interoceptive cues that inform decisions about food quantity and quality, as well as cues to satiety, reward, or behavior regulation. Several investigations have searched for it.

One example is an investigation by Sheearrer and colleagues (Shearrer et al., 2018) who measured the brain activation patterns of adolescents with normal weight parents and obese or overweight parents. They compared the groups on their response to the ingestion of a highly palatable and rewarding beverage—a milkshake. Their functional magnetic resonance imaging (fMRI) data showed that the adolescent children of overweight/obese parents showed relatively greater activation in the striatum—a brain region often implicated in reward and behavior reinforcement.

Similar investigations have been conducted with a focus on genetic differences. Rapuano and colleagues (Rapuano et al., 2017) examined groups of 9–12 year-old children differing in the presence of the fat mass and obesity-associated (FTO) gene polymorphism. In comparison to non-carriers, FTO carriers exhibited greater tissue volume and fMRI responses to food cues within another putative reward processing structure--the nucleus accumbens.

A final example is the work of Opel and colleagues (Opel et al., 2017) who analyzed multiple obesity-associated genes among middle-aged adults and computed a polygenic risk score. They found that the association between the genetic risk score and body mass index was mediated by a smaller volume of gray matter in the medial prefrontal cortex. Because the medial prefrontal cortex is often described as subserving a response inhibition function, the investigators suggested that the smaller volume may limit the cognitive control of these genetically at-risk adults over maladaptive behaviors including overeating.

The approach adopted in the present investigation is different from the approaches used by these three research groups. The first difference in approach is our focus on impairments in sustained attention versus reward processing or response inhibition. Sustained attention to the internal and external environment is a critical function underlying the self-monitoring of behavior. It is the most frequently identified and proximal cognitive domain implicated in obesity risk (Burke et al., 2011).

The second difference in approach is a focus on college students—a population that possesses greater-than-average resources and lower-than-average obesity risk. Epidemiological studies find that 16% of students are obese (Manchester, 2020) in comparison to 32.7% of adults of a similar age (Ellison-Barnes et al., 2021) and 42.4% of adults generally (Hales et al., 2020). Demonstrating a statistically significant association in this generally resilient population between parental obesity and impairments in sustained attention via an analysis of brain activation would be convincing evidence of the power of this dispositional characteristic.

The third and most notable difference in approach is a rejection of a prevailing tenent in most studies of the brain function of at-risk populations. The tenent is an assumption that decrements in function are validly estimated by a central tendency estimate computed over time or trials of a task. The problem in logic is the failure to recognize that most individuals with obesity risk do not consistently show large deficits, impairments, or abnormalities in reward processing, response inhibition, or sustained attention. Their functional impairments—if any—are very likely sporadic and brief.

As evidence of the problem, we have conducted several studies that compare the sensitivity to group differences of central tendency versus intraindividual variability estimates of dysfunction in sustained attention. Most of our studies have used the P300 event-related electroencephalographic potential (ERP). P300 is of interest because its amplitude is an objective and sensitive metric of the attentional resources directed toward a stimulus. It is largest in amplitude when it follows a stimulus that is both rare and task relevant and intermediate in amplitude when it follows a rare stimulus that has no instructed relevance. It is smallest when it follows a frequently occurring stimulus that also has no relevance. Rare, task irrelevant stimuli evoke a P300a subcomponent that has a frontocentral scalp distribution. Rare, task relevant stimuli evoke a P300a/P300b complex which is more neuroanatomically diffuse (Clark et al., 2000) and largest over the parietal scalp (Polich, 2007).

The P300 ERP offers several advantages in the study of inter-trial variability (ITV) in brain activation. For example, unlike fMRI (Elliott et al., 2020), the P300 ERP has excellent signal-to-noise characteristics and test-retest reliability (Hall et al., 2006). Accordingly, it can be measured on individual trials. In previous studies, we have shown that P300 ITV is elevated in groups possessing phenotypes related to obesity risk, including an obese body mass (Bauer, 2022) as well as conduct problems and borderline personality disorder features (Bauer, 2020). In most studies, we find that P300 ITV outperforms P300 average amplitude in differentiating these groups from low risk peers.

The present investigation is a logical extension of our prior work. It hypothesizes that college students with a parental history of obesity differ from negative parental history peers in P300 ITV and other indicators of disruptions in the maintenance of attention—elevated levels of impulsivity and slower and more variable reaction times during a simple visual oddball task. It also hypothesizes that ITV is superior to the average level of P300 activation in differentiating the parental history positive and negative groups.

2. Methods

2.1. Participants and Procedures

One-hundred-and-four female students were recruited from the freshman and sophomore classes at the University of Hartford, St. Joseph College, Trinity University, Wesleyan University, Central Connecticut State University, or the University of Connecticut. They were contacted through newspaper and radio advertisements and posters placed around the college campus. Interested students were invited to call the study office for information and eligibility screening. Those respondents who initially appeared eligible were invited to visit the medical school campus on a subsequent day for further screening and evaluations. At the time of this in-person visit, they reviewed and signed an IRB-approved informed consent and HIPAA agreements.

To describe their characteristics and inform the final decision about eligibility, the students completed questionnaires assessing problems from alcohol [Michigan Alcoholism Screening Test, MAST (Saunders & Lee, 2000)] and drug [Drug Abuse Screening Test, DAST (McCabe et al., 2006) abuse as well as impulsivity in attention, motor, and non-planning domains [Barratt Impulsiveness Scale Version 11, BIS-11 (Patton et al., 1995)]. Also, they answered questions from the EDDS [Eating Disorder Diagnostic Scale (Stice et al., 2004)] which assessed binge eating disorder features, including the weekly frequency during the past 6 months of excessive overeating with a loss of control. In addition, they completed a questionnaire describing their own medical history as well as the medical and psychiatric histories of their biological parents. Their body mass indices were calculated from directly-measured height and weight.

Students who reported no past year pregnancy, psychosis, or major medical disorders that would complicate body weight (HIV, thyroid disease, Type 1 diabetes) or evoked electroencephalographic responses (head injury, heart disease, neurological disorders) were included. To ensure consistency of outside influence, they were required to have full-time college enrollment status, live on campus, and participate in the school’s food service meal plan. All students were between 18 and 20 years of age.

To assess the sustained attention skills of these students, we presented a simple visual selective attention task consisting of 300 trials (Bauer, 1997). Before the task, they were instructed to focus attention on a computer monitor, which would display three different categories of stimuli: the rare target letter “X” (p=0.1), a novel non-target and distracting letter “C” (p=0.1), and a frequent non-target letter “T” (p=0.8). The stimuli were presented individually at a rate of 1 stimulus per 2–2.5 s at a visual angle of 4.7°. The students were instructed to press a response key quickly after the onset of the target letter and ignore the other letters.

During this task, electroencephalographic (EEG) activity was recorded from 64 Ag/AgCl electrodes referenced to linked earlobe electrodes. Electrodes placed above and below the right eye were used for eye movement and eyeblink detection. All of the signals were digitized at 500 Hz, amplified, and processed by a Compumedics® USA Incorporated Neuroscan® System (Charlotte, NC) controlled by SCAN® 4.4 software. EEG and task performance data were stored for off-line reduction and analysis.

2.2. Data Reduction

The recordings were submitted to a montage tool that selected activity at the Fz, Cz, Pz, and orbital sites for further analysis. The data were subsequently processed by a bandpass digital filter (0.5–30.0 Hz, 12 db/octave roll-off) to remove artifacts associated with voltage offsets, lead sway, electromyographic, and line frequency artifacts. To further discount the contribution of artifacts, a linear regression algorithm (Semlitsch et al., 1986) mathematically removed eye movements and eyeblinks from the Fz, Cz, and Pz channels. The ensuing data reduction steps included routines that digitally removed frequencies outside of the bandwidth of the P300 (bandpass=0.5–8 Hz, filter roll-offs=12 and 24 db/octave, respectively), re-referenced the EEG channels to the linked ears, extracted EEG epochs surrounding the onset of the rare target and novel distractor stimuli (−250 ms to +600 ms), excluded epochs with omission or commission errors, and aligned each epoch relative to the average voltage during the 250 ms period prior to stimulus onset.

The final stage of ERP data reduction involved the calculation of the across-trial average voltages and standard deviations of voltage every 2 ms during the 850 ms epochs surrounding the rare target stimuli and novel distractors. Averages (Figure 1) and standard deviations were computed separately for the Fz, Cz, and Pz epochs. P300 amplitude was the average voltage computed over an epoch window of 250 to 550 ms. To discount the positive statistical association of this average voltage with its standard deviation, P300 ITV was calculated as the residual of the standard deviation averaged over the 250–550 ms window after a linear correction for the across-trial average amplitude.

Figure 1.

Figure 1.

Open in a new tab

ERP waveforms at Fz, Cz, and Pz electrode sites averaged across trials and across members of the PH-NEG and PH-POS groups. P300 is the large positive-going component centered at approximately 375 ms after stimulus onset. The left and center columns respectively display ERPs elicited by rare target and novel distractor stimuli. Although not the subject of the present analysis because they lack a measurable P300, ERPs elicited by frequent nontarget stimuli are shown in the right column.

2.3. Data Analysis

Ninety-three of the 104 students provided at least 25 artifact-free EEG epochs in each trial category that were not associated with omission or commission errors. These students also generated an across-trial average target P300 amplitude at the Pz electrode that fell within 1.5 times the interquartile range. The 93 students were divided into two mutually exclusive groups. The parental history negative (PH-NEG) control group was composed of 69 students whose parents were never treated for obesity or Type 2 diabetes and reported a body mass index (BMI) less than 30 kg/m2. The parental history positive (PH-POS) group was composed of 24 students with at least 1 biological parent whose BMI, via parental report, exceeded the 30 kg/m2 BMI criterion that defines Type 1 obesity.

The analysis of the data proceeded in several stages. The first stage was a simple comparison of the groups on background features including demographics, Barratt Impulsivity Scale scores, personal and family substance use problems, body mass index, obesity prevalence, and binge eating frequency. Comparisons were formally tested with an independent samples t-test for continuous variables (df=1,91; equal variances not assumed) and a Pearson X2 test (df=1) for categorical variables.

The next stage in data analysis was a comparison of the groups in their P300 ERP and task performance data. Because the analysis included a large number of planned univariate comparisons, it was deemed necessary to conduct a multivariate analysis of variance (MANOVA) to screen for the presence of an overall (i.e., nonrandom) pattern of group differences. Univariate tests were conducted if and only if the group effect in the MANOVA was statistically significant (p<0.05).

A second protection against spurious findings was added to the analysis of the ERP and task performance data. Effectively, this protection adjusted the significance threshold, α, on individual univariate tests while limiting the overall false discovery rate (FDR) to 0.05. Unfortunately, there is a problem with available and computationally-straightforward methods of FDR correction, which also occurs with the use of the Bonferroni correction. The problem is an erroneous assumption of statistical test independence, which clearly does not exist for tests of P300 amplitude and P300 ITV across electrode sites and is equivocal for tests of task performance. While mindful of this problem, we decided to compute the correction. Tests that are significant after FDR correction (erroneously assuming test independence) are noted with an asterisk in Table 2.

Table 2.

ERP and task performance data [M(SE)].

PH-NEG PH-POS ANOVA Effect Size, ήp2
Target Stimulus P300 Amplitude (microvolts)
Fz 7.76(4.58) 5.22(4.16) F=5.69, p=0.019 0.059
Cz 12.06(4.87) 9.09(4.57) F=6.82, p=0.011* 0.070
Pz 13.85(4.48) 11.97(5.41) F=2.81, p=0.097 0.030
Target Stimulus P300 ITV (residual of SD)
Fz −0.22(1.26) 0.69(1.13) F=9.82, p=0.002* 0.097
Cz −0.33(1.18) 0.81(1.43) F=14.66, p<0.001* 0.139
Pz −0.41(1.35) 0.86(2.14) F=11.58, p=0.001* 0.113
Novel Distractor P300 Amplitude (microvolts)
Fz 2.96(2.59) 2.17(2.77) F=1.59, p=0.209 0.017
Cz 5.10(2.57) 4.14(2.52) F=2.49, p=0.118 0.027
Pz 5.84(2.30) 5.01(2.32) F=2.29, p=0.134 0.025
Novel Distractor P300 ITV (residual of SD)
Fz −0.07(1.21) 0.28(1.36) F=1.48, p=0.227 0.016
Cz −0.08(1.08) 0.29(1.07) F=2.15, p=0.145 0.023
Pz −0.20(1.40) 0.37(1.80) F=2.56, p=0.113 0.027
Task Performance
Hit Rate, proportion 0.99(0.02) 0.99(0.02) F=0.43, p=0.510 0.005
False Alarm Rate, proportion 0.002(0.002) 0.003(0.005) F=2.33, p=0.130 0.025
Reaction time, ms 354(45) 382(101) F=3.47, p=0.066 0.037
CV of Reaction time, SD/M 0.21(0.07) 0.20(0.05) F=0.09, p=0.756 0.001
Open in a new tab
*

Statistically significant p-value after correction for overall false discovery rate.

The third stage of the analysis focused on the accuracy with which P300 ITV at one representative electrode site (Pz) could be used to predict the membership of individual cases in the PH-NEG and PH-POS groups. It was therefore a more rigorous test of the association of these variables than could be demonstrated via ANOVA. The analysis employed receiver operating characteristic (ROC) analysis which generates a classification accuracy, area-under-the-curve (AUC), statistic.

The fourth analysis stage was a formal mediation analysis in which P300 ITV at the Pz electrode was tested as a mediator of the parental obesity and offspring body mass index association. Admittedly, the validity of mediation analysis is controversial (Iacobucci, 2012) when one or more of the input variables are categorical. Accordingly, the present analysis should not be considered definitive but merely exploratory. The methods employed for mediation analysis were described by Baron and Kenny (Baron & Kenny, 1986) and subsequently revised by MacKinnon and colleagues (MacKinnon et al., 2007).

The final analysis stage was a test of the relative power of family and personal histories of excess body weight as a correlate of P300 ITV in the offspring. It computed and contrasted the simple bivariate correlations of the target stimulus P300 ITV recorded at the Pz electrode site with the student’s parental history of obesity versus her personal BMI.

3. Results

3.1. Background Characteristics of PH-POS and PH-NEG groups

As shown in Table 1, both groups were comprised of students who were white, predominantly non-Hispanic, and approximately 19 years of age. They did not differ significantly in race, ethnicity, or age. They also did not differ significantly in the prevalence of diabetes, use of illicit drugs, or in the number of alcohol or drug use problems respectively reported on the MAST and DAST. However, there was a trend (p=0.10) toward a higher prevalence of parental alcohol or drug abuse among PH-POS (45.8%) versus PH-NEG (27.9%) students. Also, there was an opposite trend (p=0.07) across groups in the use of prescribed drugs (41.7% versus 62.3%).

Table 1.

Background characteristics of groups [Mean(SD) or %(n)].

PH-NEG (n=69) PH-POS (n=24) T or X2
Age, y (SD) 19.51(1.25) 19.25(1.03) T=0.99, p=0.32
Black, %(n) 0(0) 0(0) --
Hispanic, %(n) 13.0(9) 16.7(4) X2=0.19, p=0.65
Diabetic, %(n) 0(0) 0(0) ---
BMI 23.38(3.28) 27.38(8.49) T=−2.24, p=0.03
BMI ≥ 30, % 2.9(2) 16.7(4) X2=5.59, p=0.01, Fisher Exact Test p=0.03
Episodes/wk with loss of control binge eating during past 6 mos 1.42(1.67) 2.54(2.16) T=−2.31, p=0.02
BIS-11 Attention 16.03(3.65) 18.33(4.86) T=−2.12, p=0.04
BIS-11 Motor 21.96(3.64) 24.33(5.33) T=−2.02, p=0.05
BIS-11 Non-planning 22.80(4.28) 24.54(5.16) T=−1.48, p=0.14
Routinely take prescription drugs, %(n) 62.3(43) 41.7(10) X 2=3.09, p=0.07
Ever taken recreational drugs, %(n) 36.2(25) 33.3(8) X 2=0.06, p=0.79
AUDIT total score 7.54(4.86) 8.13(6.18) T=−0.42, p=0.67
DAST 1.26(1.43) 1.50(1.88) T=−0.56, p=0.53
Parental history of alcohol or drug abuse, %(n) 27.9(19) 45.8(11) X 2=2.58, p=0.10
Open in a new tab

A few significant differences in group background features were detected. They were consistent with expectations. For example, PH-POS students had a statistically greater (M=27.38 versus M=23.38 kg/m2; t=−2.24, p=0.03) body mass index and a greater obesity prevalence (16.7% versus 2.9%, Fisher’s Exact Test p=0.03) than their peers. They also reported more episodes per week of binge eating with a loss of control (M=2.54 versus M=1.42; t=−2.31, p=0.02) on the EDDS and higher scores on the attention (M=18.33 versus M=16.03; t=−2.12, p=0.04) and motor (M=24.33 versus M=21.96; t=−2.02, p=0.05) impulsivity subscales of the BIS-11.

3.2. ERP and Task Performance Differences

The MANOVA evaluating the statistical effect of the grouping variable on a composite of the ERP and task performance summary measures was statistically significant [Pillai’s Trace=0.27, F(16,76)=1.73, p=0.05]. Accordingly, an analysis of the effect of the grouping variable on individual measures was justified.

As shown in Table 2, univariate ANOVAs of the average amplitude and ITV of the P300 response to the novel distractor did not reveal a statistical effect of Group. The Group effect was detected in simple univariate analyses of P300 average amplitude at Fz [F(1,91)=5.69, p=0.01] and in simple and FDR-corrected analyses of P300 amplitude at Cz [F(1,91)=6.82, p=0.01] where the PH-POS group demonstrated a lower amplitude. It was more robustly evident in simple and FDR-corrected analyses of P300 ITV at all three electrode sites--Fz [(F(1,91)=9.82, p=0.002)], Cz [(F(1,91)=14.66, p<0.001)], and Pz [(F(1,91)=11.58, p=0.001)]--where the PH-POS group consistently demonstrated greater inter-trial variability in amplitude. Figure 2 buttresses this statement by presenting the effect sizes, ήp2, of each comparison. The proportion of variance explained by P300 ITV (0.097–0.139) was approximately double the variance proportion explained by P300 average amplitude (0.030–0.070).

Figure 2.

Figure 2.

Open in a new tab

Effect sizes (ήp2) for comparisons of the PH-NEG and PH-POS groups on rare target P300 averaged amplitude and P300 inter-trial variability. Note the superiority of P300 ITV in explaining variance in the grouping variable.

Analyses of task performance data revealed only one trend. PH-POS students (M=382 ms) were marginally [F(1,91)=3.47, p=0.066)] slower than their PH-NEG peers (M=354 ms) in responding to the target stimulus. There were no significant group differences in hit or false alarm rates or the inter-trial variability in reaction time (CV of reaction time).

3.3. ROC Analysis

With parental history as the criterion and P300 ITV at the Pz electrode as the predictor, the ROC analysis finding was statistically significant. It yielded an AUC of 0.736 (SE=0.064, p=0.001).

3.4. Mediation Analysis

Comparative tests of the parental obesity and personal BMI association with (B=0.74, SE=0.22, t=3.29, p=0.001) and without (B=0.71, SE=0.24, t=2.91, p=0.005) Pz P300 ITV as a mediator did not reveal evidence of statistically significant mediation. The change in the unstandardized beta weight of only 0.03 BMI units, which is the indirect effect coefficient, was trivial.

3.5. Familial versus personal BMI as a correlate of P300 ITV

Simple Pearson product-moment correlations of target stimulus P300 ITV at the Pz electrode with parental weight status and personal BMI revealed a stronger association with the former variable. For the parental weight status analysis, the correlation with P300 ITV was r=0.336 (p<0.001). For the personal BMI analysis, the correlation was of a lower magnitude, r=0.146 (p=0.163). Yet, a formal comparison of the z-transformed correlation coefficients did not reveal a statistically significant decrement (z=0.81, p=0.21). Parental weight status and personal BMI were significantly intercorrelated (r=0.331, p<0.001).

4. Discussion

A problem faced by many researchers undertaking studies of either genetic or family characteristics promoting obesity is the confounding of these contributors. Research studies that examine parenting behavior, food insecurity, or related aspects of the family environment are not pure and independent of a competing genetic explanation for their findings. Studies that focus on candidate genes or genome wide associations are likewise not immune to criticisms. Thus far, the predominant findings in the literature using either single gene or polygenic risk score as predictors of BMI or obesity class are minor and inconsistent (Manfredi et al., 2021).

Because family history often explains more variance than any genetic polymorphism or combination of polymorphisms (Zhang et al., 2022), the present investigation focused on a family history of obesity as the marker of risk. We attempted to show that healthy college students with a parental history of obesity possess a nervous system that is less successful in maintaining a stable orienting response to rare, task-salient events occurring in the environment. We were successful in this attempt. In comparison to the PH-NEG group, students in the PH-POS group demonstrated greater ITV in P300 amplitude at Fz, Cz, and Pz sites. They also demonstrated smaller P300 average amplitude at the Cz electrode site.

A separate and interesting discussion can be formed around this co-occurrence of enhanced P300 ITV and reduced P300 average amplitude in the PH-POS group. Both findings suggest a dysfunction. But, there are unanswered questions about common versus independent sources for the dysfunction. For example, does greater ITV in single trial P300 amplitudes fully explain the reduction in mean P300 amplitude or vice-versa? If the answer to either question is affirmative, then one would expect P300 ITV and P300 average amplitude to be equally powerful in differentiating the groups. Yet, they were not equally powerful: P300 ITV explained a markedly greater proportion of the variance in the group effect (Figure 2). One might accordingly conclude that P300 ITV indexes a different source of dysfunction.

A factor that limits our ability to ascribe the elevation in P300 ITV to parental and/or genetic history with a high level of confidence is the co-occurrence of a greater average BMI among the members of the PH-POS group. Parental weight status and personal BMI were significantly intercorrelated (r=0.331, p<0.001). Therefore, entering personal BMI as a covariate in the analysis of the parental history is not an optimal analytic solution. In its place, we computed and compared the correlations of P300 ITV with parental obesity and with offspring BMI. The former correlation was larger than the latter. However, the correlations were not significantly different. It appears that parental obesity may be a more powerful to P300 ITV than personal BMI but the contributions of the latter variable cannot be definitively dismissed.

Another limitation is our failure to show that P300 ITV mediates the association between parental obesity and offspring BMI. To maximize power for detecting mediation, we tried to select a participant sample that was not as truncated in risk factors as typical samples recruited from college student populations. We recruited from both state and private colleges. Also, we focused on college freshman and sophomores because inherited personality and psychological disorders (Auerbach et al., 2018; Sedgwick, 2018) known to contribute to childhood- or adolescent-onset obesity (Anderson et al., 2006; Girela-Serrano et al., 2022) are more prevalent during the early college years and less prevalent during later years (Hjorth et al., 2016). These adjustments in the recruiting strategy and eligibility criteria were sufficient to reveal group differences in the expected direction (PH-POS>PH-NEG) in BMI, binge eating frequency, and levels of impulsivity on the attention and motor subscales of the BIS-11 (Table 1). Of course, a simple ANOVA comparison of two groups and a multi-factor mediation analysis have different power requirements (i.e., sample size, within- and between-group variance, error estimation) for statistical significance. Questions about P300 ITV as a mediator of the parental obesity by offspring BMI association can be better answered in a future and larger investigation.

4.1. Conclusions

We suspect that the temporally unstable single trial P300 amplitudes found among PH-POS students represent an inherited and dysregulated brain mechanism that translates their family risk into personal risk. By this mechanism, their processing of satiety cues or other cues that influence eating behavior may be episodically distorted to elicit binge eating or other unhealthy eating behaviors. It is noteworthy that this theory of episodic inattention and episodic neural dysregulation is different from many neurocognitive theories of obesity. It recognizes that many people who are obese or members of at-risk-for-obesity groups are capable of appropriately monitoring and regulating their eating behavior at most times.

The present theory of dysregulation is more similar to neurocognitive theories that have emerged from studies of reaction time variability (RTV) among children with attention-deficit-hyperactivity disorder (ADHD) and older adults with signs of cognitive impairment. In the ADHD literature (Kofler et al., 2013), there are numerous reports of enhanced RTV among children with the diagnosis. Enhanced RTV is evident across tasks of different types, including tasks that challenge behavioral inhibition, working, memory, or motor control among other skills. In the literature examining neurocognitive problems in older adults, enhanced RTV is likewise a common finding. It predicts many unfortunate outcomes, including increased risk of all-cause mortality (Batterham et al., 2014) and impairments in prospective memory (Haynes et al., 2016) among other unfortunate outcomes.

We did not detect a difference between our groups of healthy young adults in RTV that correspond to the difference in P300 ITV. One explanation may be found in the evidence suggesting that there are cognitive processes (response selection and programming) that may contribute to reaction time and its variability but not to P300 amplitude and its variability. However, we did detect a trend (p=0.06) suggesting that the average response times of PH-POS college students are slower than their PH-NEG peers. This trend suggests that the temporal dysregulation of the P300 orienting response indicated by an elevation in P300 ITV found among PH-POS students is not behaviorally silent. It remains to be determined if P300 ITV is malleable and can become an objectively-measured target for biofeedback or other interventions that may reduce their obesity risk.

Variability in brain activation has not been tested as a correlate of obesity risk.

P300 amplitude variability was tested as a correlate of parental obesity.

The offspring of obese parents exhibited greater P300 variability and body mass.

Excessive P300 variability may objectively indicate brief lapses in sustained attention.

Episodic inattention is rarely addressed in obesity prevention programs.

Acknowledgements

This research was supported by PHS grant P50AA027055.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

CRediT authorship contribution statement

The author is fully responsible for the study’s concept, experimental design, and data analysis, as well as the preparation of this manuscript.

Declaration of competing interest

The author has no conflicts of interest to disclose.

Declaration of interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.

References

  1. Anderson SE, Cohen P, Naumova EN, & Must A (2006). Relationship of childhood behavior disorders to weight gain from childhood into adulthood. Ambul Pediatr, 6(5), 297–301. http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=17000421 [DOI] [PubMed] [Google Scholar]
  2. Auerbach RP, Mortier P, Bruffaerts R, Alonso J, Benjet C, Cuijpers P, Demyttenaere K, Ebert DD, Green JG, Hasking P, Murray E, Nock MK, Pinder-Amaker S, Sampson NA, Stein DJ, Vilagut G, Zaslavsky AM, & Kessler RC (2018). WHO World Mental Health Surveys International College Student Project: Prevalence and distribution of mental disorders. Journal of Abnormal Psychology, 127(7), 623–638. 10.1037/ABN0000362 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Baron RM, & Kenny DA (1986). The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol, 51(6), 1173–1182. http://www.ncbi.nlm.nih.gov/pubmed/3806354 [DOI] [PubMed] [Google Scholar]
  4. Batterham PJ, Bunce D, Mackinnon AJ, & Christensen H (2014). Intra-individual reaction time variability and all-cause mortality over 17 years: A community-based cohort study. Age and Ageing, 43(1), 84–90. 10.1093/ageing/aft116 [DOI] [PubMed] [Google Scholar]
  5. Bauer LO (1997). Frontal P300 decrements, childhood conduct disorder, family history, and the prediction of relapse among abstinent cocaine abusers. Drug and Alcohol Dependence, 44(1), 1–10. 10.1016/S0376-8716(96)01311-7 [DOI] [PubMed] [Google Scholar]
  6. Bauer LO (2020). Neural responses to signals for behavior change: Greater within-person variability is associated with risk factors for substance dependence. Alcohol Clin Exp Res. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bauer LO (2022). Altered brain activation in uncomplicated pediatric obesity: The second moment is more sensitive than the first. Brain and Cognition, 164. 10.1016/J.BANDC.2022.105923 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Burke LE, Wang J, & Sevick MA (2011). Self-monitoring in weight loss: a systematic review of the literature. Journal of the American Dietetic Association, 111(1), 92–102. 10.1016/J.JADA.2010.10.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Clark VP, Fannon S, Lai S, Benson R, & Bauer L (2000). Responses to rare visual target and distractor stimuli using event-related fMRI. J Neurophysiol, 83(5), 3133–3139. http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=10805707 [DOI] [PubMed] [Google Scholar]
  10. Elliott ML, Knodt AR, Ireland D, Morris ML, Poulton R, Ramrakha S, Sison ML, Moffitt TE, Caspi A, & Hariri AR (2020). What Is the Test-Retest Reliability of Common Task-Functional MRI Measures? New Empirical Evidence and a Meta-Analysis. Psychological Science, 31, 792–806. 10.1177/0956797620916786 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Ellison-Barnes A, Johnson S, & Gudzune K (2021). Trends in Obesity Prevalence Among Adults Aged 18 Through 25 Years, 1976–2018. JAMA, 326(20), 2073–2074. 10.1001/JAMA.2021.16685 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Girela-Serrano BM, Guerrero-Jiménez M, Spiers ADV, & Gutiérrez-Rojas L (2022). Obesity and overweight among children and adolescents with bipolar disorder from the general population: A review of the scientific literature and a meta-analysis. Early Intervention in Psychiatry, 16(2), 113–125. 10.1111/EIP.13137 [DOI] [PubMed] [Google Scholar]
  13. Hales CM, Carroll MD, Fryar CD, & Ogden CL (2020). Prevalence of Obesity and Severe Obesity Among Adults: United States, 2017–2018. NCHS Data Brief, 360, 1–8. [PubMed] [Google Scholar]
  14. Hall MH, Schulze K, Rijsdijk F, Picchioni M, Ettinger U, Bramon E, Freedman R, Murray RM, & Sham P (2006). Heritability and reliability of P300, P50 and duration mismatch negativity. Behav Genet, 36(6), 845–857. 10.1007/s10519-006-9091-6 [DOI] [PubMed] [Google Scholar]
  15. Haynes BI, Kliegel M, Zimprich D, & Bunce D (2016). Intraindividual reaction time variability predicts prospective memory failures in older adults. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn, 25, 1–14. 10.1080/13825585.2016.1268674 [DOI] [PubMed] [Google Scholar]
  16. Hjorth CF, Bilgrav L, Frandsen LS, Overgaard C, Torp-Pedersen C, Nielsen B, & Bøggild H (2016). Mental health and school dropout across educational levels and genders: a 4.8-year follow-up study. BMC Public Health, 16(1). 10.1186/S12889-016-3622-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Iacobucci D (2012). Mediation analysis and categorical variables: The final frontier. Journal of Consumer Psychology, 22(4), 582–594. 10.1016/J.JCPS.2012.03.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Kofler MJ, Rapport MD, Sarver DE, Raiker JS, Orban SA, Friedman LM, & Kolomeyer EG (2013). Reaction time variability in ADHD: a meta-analytic review of 319 studies. Clin Psychol Rev, 33(6), 795–811. [DOI] [PubMed] [Google Scholar]
  19. MacKinnon DP, Fairchild AJ, & Fritz MS (2007). Mediation Analysis. Annual Review of Psychology, 58, 593. 10.1146/ANNUREV.PSYCH.58.110405.085542 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Manchester R (2020). How has the health of college students changed in the last 50 years? Journal of American College Health : J of ACH, 68(8), 795–797. 10.1080/07448481.2020.1859853 [DOI] [PubMed] [Google Scholar]
  21. Manfredi L, Accoto A, Couyoumdjian A, & Conversi D (2021). A Systematic Review of Genetic Polymorphisms Associated with Binge Eating Disorder. Nutrients, 13(3), 1–19. 10.3390/NU13030848 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. McCabe SE, Boyd CJ, Cranford JA, Morales M, & Slayden J (2006). A modified version of the Drug Abuse Screening Test among undergraduate students. J Subst Abuse Treat, 31(3), 297–303. http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=16996392 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Opel N, Redlich R, Kaehler C, Grotegerd D, Dohm K, Heindel W, Kugel H, Thalamuthu A, Koutsouleris N, Arolt V, Teuber A, Wersching H, Baune BT, Berger K, & Dannlowski U (2017). Prefrontal gray matter volume mediates genetic risks for obesity. Molecular Psychiatry, 22(5), 703–710. 10.1038/MP.2017.51 [DOI] [PubMed] [Google Scholar]
  24. Patton JH, Stanford MS, & Barratt ES (1995). Factor structure of the Barratt impulsiveness scale. J Clin Psychol, 51(6), 768–774. http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=8778124 [DOI] [PubMed] [Google Scholar]
  25. Polich J (2007). Updating P300: an integrative theory of P3a and P3b. Clin Neurophysiol, 118(10), 2128–2148. 10.1016/j.clinph.2007.04.019 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Rapuano KM, Zieselman AL, Kelley WM, Sargent JD, Heatherton TF, & Gilbert-Diamond D (2017). Genetic risk for obesity predicts nucleus accumbens size and responsivity to real-world food cues. Proceedings of the National Academy of Sciences of the United States of America, 114(1), 160–165. 10.1073/PNAS.1605548113 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Rath SR, Marsh JA, Newnham JP, Zhu K, Atkinson HC, Mountain J, Oddy WH, Hughes IP, Harris M, Leong GM, Cotterill AM, Sly PD, Pennell CE, & Choong CS (2016). Parental pre-pregnancy BMI is a dominant early-life risk factor influencing BMI of offspring in adulthood. Obesity Science & Practice, 2(1), 48–57. 10.1002/OSP4.28 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Saunders JB, & Lee NK (2000). Hazardous alcohol use: its delineation as a subthreshold disorder, and approaches to its diagnosis and management. Compr Psychiatry, 41(2 Suppl 1), 95–103. http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=10746911 [DOI] [PubMed] [Google Scholar]
  29. Sedgwick JA (2018). University students with attention deficit hyperactivity disorder (ADHD): a literature review. Irish Journal of Psychological Medicine, 35(3), 221–235. 10.1017/IPM.2017.20 [DOI] [PubMed] [Google Scholar]
  30. Semlitsch H. v., Anderer P, Schuster P, & Presslich O (1986). A Solution for Reliable and Valid Reduction of Ocular Artifacts, Applied to the P300 ERP. Psychophysiology, 23(6), 695–703. 10.1111/j.1469-8986.1986.tb00696.x [DOI] [PubMed] [Google Scholar]
  31. Shearrer GE, Stice E, & Burger KS (2018). Adolescents at high risk of obesity show greater striatal response to increased sugar content in milkshakes. The American Journal of Clinical Nutrition, 107(6), 859–866. 10.1093/AJCN/NQY050 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Stice E, Fisher M, & Martinez E (2004). Eating disorder diagnostic scale: additional evidence of reliability and validity. Psychological Assessment, 16(1), 60–71. 10.1037/1040-3590.16.1.60 [DOI] [PubMed] [Google Scholar]
  33. Zhang Y, Meigs JB, Liu CT, Dupuis J, & Sarnowski C (2022). Leveraging family history in genetic association analyses of binary traits. BMC Genomics, 23(1). 10.1186/S12864-022-08897-8 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Data will be made available on request.

RESOURCES