**Study Design and Experimental Manipulation** In the context of an acute laboratory stressor (the Trier Social Stress Test), 150 female participants will be randomly assigned to one of five experimental conditions according to a 2 (type of food eaten: healthy vs. unhealthy) x 2 (timing of eating: before vs. after the stressor) + 1 (no-food control) design. Healthy foods will include whole fruits and vegetables, and unhealthy foods will include processed, high-calorie/fat/sugar foods. All foods served will be ranked by participants in pre-screening at a top choice food from a list of foods (see Food Opinions Questionnaire below). **Questionnaire Measures** *Chronic perceived stress.* The 14-item Perceived Stress Scale (PSS) will be administered as a measure of chronic perceived stress in the past month (Cohen, Kamarck, & Mermelstein, 1983). *Mood state.* Positive and negative mood state will be assessed at five time points during the lab visit using the Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988). Time 1 = baseline; Time 2 = ~10 minutes after the TSST instructions were delivered and post-"before" eating manipulation; Time 3 = immediately post-TSST; Time 4 = ~10 minutes post-TSST and post-"after" eating manipulation; Time 5 = 60 minutes post-TSST. *Post-stressor appraisals.* We created this 7-item Likert scale to assess participants’ appraisals of how stressful the TSST was. *Rumination.* Post-stressor rumination will be measured using The Negative Thoughts Subscale from the Modified Thoughts Questionnaire (Zoccola, Dickerson, & Zaldivar, 2008), which consists of 14 items assessing how much participants had certain negative thoughts in the time since the speech task had ended. *Trait comfort eating expectations (TCEE).* We created a 6-item Trait Comfort Eating Expectations Scale in collaboration with Jeff Hunger to assess the extent to which participants believe that food improves their mood. *Bartholomew and Horowitz’s Attachment Scale.* Participants will indicate which of 4 different paragraphs best describes their approach to social relationships (Bartholomew & Horowitz, 1991). *Brief COPE Scale.* In this 28-item scale (Carver, 1997), participants report how often they usually cope with life stress using certain techniques. *Primary Appraisal Secondary Appraisal Scale (PASA).* This 16-item scale will measure cognitive appraisal processes (i.e., threat vs. challenge perceptions) occurring when an individual is anticipating undergoing a stressful task (Gaab et al., 2003). *Risky Families Questionnaire (RFQ).* This 13-item scale will assess how organized and supportive the participant’s home environment was when she was at the age of 5-15 (Repetti, Taylor, & Seeman, 2002). *Biphasic Alcohol Effects Scale (BAES).* The BAES (Martin, Earleywine, Musty, Perrine, & Swift, 1993) consists of 14-items. Participants will rate the extent to which drinking alcohol has personally produced feelings like “energized” and “sedated. *Alcohol Expectancy Questionnaire (AEQ).* The Alcohol Expectancy Questionnaire (Brown, Christiansen, & Goldman, 1987) is comprised of 68-items. The measure contains six subscales which capture what an individual expects to happen after they drink alcohol: global positive, social physical pleasure, tension reduction, social assertiveness, aggression/arousal, and sexual enhancement. *Food Opinion Questionnaire (FOQ).* Participants will be given a list of 20 foods, from which they are asked to rate their top three comfort foods (i.e., “What foods would make you feel better if you were in a bad mood?”). They will then rate these top three comfort foods on, "How confident are you that this food would make you feel better if you were in a bad mood? (1 = not at all to 7 = very). Additional prompts are also provided (e.g., “What foods would you want if you were on-the-go?”), serving to distract participants from the main focus of the study (Wagner, Ahlstrom, Vickers, Redden, & Mann, 2014). **Physiological Measures** We will continuously measure the physiological autonomic nervous system (ANS) measures of electrodermal activity, heart rate variability, pre-ejection period, and respiration. For all ANS measures, we will analyze six time periods of interest: baseline (5 minutes), the TSST instructions period (3 minutes), the speech preparation period (3 minutes), the speech task (5 minutes), the math task (5 minutes), and the post-TSST recovery period (12 minutes). For each time period, each ANS outcome will be calculated in MindWare in 1-minute segments. Salivary cortisol samples will be collected at baseline, and at 15, 25, and 60 minutes post-stressor to index HPA axis reactivity and recovery. Cortisol reactivity to the TSST will be defined as change in cortisol secretion from baseline through 15 and 25 minutes post-TSST, as cortisol reactivity is known to peak at 15-25 minutes after a stressor has ended (Kudielka et al., 2007). Cortisol recovery to the TSST will be defined as change in cortisol secretion from 25-60 minutes post-TSST. We will measure participants’ weight and height without shoes. BMI will be incorporated as a covariate in later analyses that include ANS outcomes. **Other Measures** *Volume of Food Consumed.* We will measure the weight of food that participants consume by calculating the difference in the dish weight before versus after eating. *Food Content.* We will record the amount of calories, fat, and sugar for the food served using packaging information (when available) or online databases. *Independent Ratings of TSST Performance.* Three independent raters blind to the study purpose, study hypotheses, and participants' experimental condition will review videotaped recordings of participants' performance on the TSST. After viewing, they will rate participants' performance on the tasks using items from a published study (Brooks, 2014). *TSST Evaluator Ratings of TSST Performance.* The staff members who administer the TSST (blind to experimental condition) will rate their respective participants on how nervous/stress they seemed to be during the TSST tasks. **Primary Hypotheses** 1. Both healthy and unhealthy foods consumed pre-TSST will reduce psychophysiological stress reactivity and enhance psychophysiological stress recovery, compared to consuming no food. That is, compared to those who consume no food pre-TSST, those in groups 1 and 2 will: a. Appraise the TSST as more of a challenge and less of a threat (PASA scale) b. Show a smaller increase in negative affect from PANAS 1 through PANAS 3 c. Show a smaller decrease in positive affect from PANAS 1 through PANAS 3 d. Show a faster and larger decrease in negative affect from PANAS 3 through PANAS 5 e. Show a faster and larger increase in positive affect from PANAS 3 through PANAS 5 f. Show fewer negative ruminatory thoughts post-TSST g. Show more positive ruminatory thoughts post-TSST h. Show smaller decreases in parasympathetic nervous system (PNS) activation (indexed by heart rate variability) during the stress reactivity phase i. Show dampened physiological stress reactivity of the HPA axis (indexed by cortisol levels) and SNS (indexed by electrodermal activity and pre-ejection period) j. Show enhanced physiological stress recovery of the HPA axis (indexed by cortisol), SNS (indexed by electrodermal activity and pre-ejection period), and PNS (indexed by heart rate variability) 2. Both healthy and unhealthy foods consumed post-TSST will enhance psychophysiological stress recovery, compared to a non-food control. That is, compared to those in group 5, those in groups 3 and 4 will: a. Show a faster and larger decrease in negative affect from PANAS 3 through PANAS 5 b. Show a faster and larger increase in positive affect from PANAS 3 through PANAS 5 c. Show fewer negative ruminatory thoughts post-TSST d. Show more positive ruminatory thoughts post-TSST e. Appraise the TSST as less stressful at post-TSST (Post-Stressor Appraisals scale) f. Show enhanced physiological stress recovery of the HPA axis (indexed by cortisol), SNS (indexed by electrodermal activity and pre-ejection period), and PNS (indexed by heart rate variability) 3. Comfort eating before versus after the TSST (collapsed across healthy and unhealthy foods) will better enhance psychophysiological stress recovery. That is, comparing groups 1 and 2 (combined) versus groups 3 and 4 (combined), the former will: a. Show fewer negative ruminatory thoughts post-TSST b. Show more positive ruminatory thoughts post-TSST c. Show enhanced physiological stress recovery of the HPA axis (indexed by cortisol), SNS (indexed by electrodermal activity and pre-ejection period), and PNS (indexed by heart rate variability) d. Show a faster and larger decrease in negative affect from PANAS 3 through PANAS 5 e. Show a faster and larger increase in positive affect from PANAS 3 through PANAS 5 f. Appraise the TSST as less stressful at post-TSST (Post-Stressor Appraisals scale) 4. Unhealthy foods consumed pre-TSST will reduce psychophysiological stress reactivity and enhance psychophysiological stress recovery, compared to healthy foods consumed pre-TSST. That is, compared to those in the group 2, those in group 1 will: a. Appraise the TSST as more of a challenge and less of a threat (PASA scale) b. Show a smaller increase in negative affect from PANAS 1 through PANAS 3 c. Show a smaller decrease in positive affect from PANAS 1 through PANAS 3 d. Show a faster and larger decrease in negative affect from PANAS 3 through PANAS 5 e. Show a faster and larger increase in positive affect from PANAS 3 through PANAS 5 f. Show fewer negative ruminatory thoughts post-TSST g. Show more positive ruminatory thoughts post-TSST h. Show smaller decreases in parasympathetic nervous system (PNS) activation (indexed by heart rate variability) during the stress reactivity phase i. Show dampened physiological stress reactivity of the HPA axis (indexed by cortisol levels) and SNS (indexed by electrodermal activity and pre-ejection period) j. Show enhanced physiological stress recovery of the HPA axis (indexed by cortisol), SNS (indexed by electrodermal activity and pre-ejection period), and PNS (indexed by heart rate variability) 5. Unhealthy foods consumed post-TSST will enhance psychophysiological stress recovery, compared to healthy foods. That is, compared to those in group 4, those in group 3 will: a. Show a faster and larger decrease in negative affect from PANAS 3 through PANAS 5 b. Show a faster and larger increase in positive affect from PANAS 3 through PANAS 5 c. Show fewer negative ruminatory thoughts post-TSST d. Show more positive ruminatory thoughts post-TSST e. Appraise the TSST as less stressful at post-TSST(Post-Stressor Appraisals scale) f. Show enhanced physiological stress recovery of the HPA axis (indexed by cortisol), SNS (indexed by electrodermal activity and pre-ejection period), and PNS (indexed by heart rate variability) 6. Unhealthy comfort eating versus healthy comfort eating (collapsed across pre- and post-TSST) will better enhance psychophysiological stress recovery. That is, comparing groups 1 and 3 (combined) versus groups 2 and 4 (combined), the former will: a. Show fewer negative ruminatory thoughts post-TSST b. Show more positive ruminatory thoughts post-TSST c. Show enhanced physiological stress recovery of the HPA axis (indexed by cortisol), SNS (indexed by electrodermal activity and pre-ejection period), and PNS (indexed by heart rate variability) d. Show a faster and larger decrease in negative affect from PANAS 3 through PANAS 5 e. Show a faster and larger increase in positive affect from PANAS 3 through PANAS 5 f. Appraise the TSST as less stressful at post-TSST (Post-Stressor Appraisals scale) 7. Compared to those who consume no food pre-TSST, those in groups 1 and 2 will: a. Give speeches that are of a greater time duration b. Make fewer mistakes on the TSST math task c. Get a greater number of correct subtractions on the TSST math task d. Be rated by independent raters (blind to the study purpose and experimental condition) as performing better on the TSST speech and math tasks e. Be rated by the TSST evaluators (blind to experimental condition) as seeming less nervous/stressed during the TSST tasks **Moderation Hypotheses** 1. Any observed stress-buffering effects outlined above for food versus no food (1a-1j and 2a-2f) will be greater in those individuals who: a. Express greater expectations that eating will improve mood (TCEE) b. Express greater confidence in pre-screening that the specific food selected for their lab day comfort food, “would make you feel better if you were in a bad mood” (FOQ) c. Exhibit higher scores on trait emotional eating (DEBQ) d. Exhibit lower scores on trait restraint eating (DEBQ) e. Report higher chronic stress levels (PSS) f. Report lower depressive symptoms (CESD) g. Show higher self-distraction coping (Brief COPE) 2. Any observed stress-buffering effects outlined above for unhealthy food versus healthy food (4a-j, 5a-f, and 6a-f) will be greater in those individuals who: a. Show higher substance use coping (Brief COPE) b. Report greater coping with palatable foods (PEMS) c. Report higher tension reduction alcohol expectancies (AEQ) d. Report sensitivity to the sedating effects of alcohol (BAES) **Mediation Hypotheses** 1. Any observed stress-buffering effects outlined above (1a-1j, 2a-2f, 4a-j, 5a-f, and 6a-f) will mediated by: a. Amount of sugar consumed from food in the lab b. Amount of fat consumed from food in the lab c. Amount of calories consumed from food in the lab d. Feelings of guilt at PANAS time 2 **Additional Hypotheses** 1. Trait emotional eating (DEBQ) will be positively associated with: a. Depressive symptoms (CESD) b. Trait comfort eating expectations (TCEE) c. Perceived stress (PSS) d. Weight of food consumed in the lab **Data Analytic Plan** After cleaning the data, we will explore descriptive characteristics for all study variables, including means, standard deviations, and normality of distributions. We will also compute bivariate associations between study variables, which will aid in determining whether given covariates are appropriate for each model. We will use multilevel modeling to test the hypotheses with repeated measures outcomes (i.e., PANAS, cortisol, HRV, PEP, EDA), as this statistical method is appropriate for modeling data with repeated measures nested within individuals. Here, multilevel modeling allows the simultaneous estimation of both within-subjects variance (repeated measures of psychological and physiological stress) and between-subjects variance (group differences related to experimental condition). Separate analyses will be conducted for each outcome, and each multilevel model will test for the effects of experimental group, time, and the group by time interaction. For repeated measures *reactivity* analyses, we will test the effects of eating healthy versus unhealthy versus no food pre-TSST on given outcomes. For repeated measures *recovery* analyses, we will conduct a 5-condition omnibus test of experimental condition on the given outcomes. For analyses with ANS outcomes, we will examine respiratory rate, age, and BMI as potential covariates, as these factors can influence physiological outcomes. We will follow up on any significant group by time interaction effects by conducting ANOVAs at each appropriate time point to test for group differences. Any significant omnibus tests will be followed with pairwise comparisons. In addition, we will use ANOVA (or ANCOVA, if appropriate) models to test hypotheses with outcomes that were measured at only one time point (e.g., PASA scale, rumination, post-stressor appraisals, independent ratings of participant performance). For pre-TSST coutcomes, we will conduct 3-condition (unhealthy vs. healthy vs. no food pre-TSST) omnibus ANOVAs to test any effects of experimental condition, following up on any significant results with planned comparisons between: (1) those who ate healthy food versus no food pre-TSST, (2) those who ate unhealthy food versus no food pre-TSST, (3) those who ate healthy food versus unhealthy food pre-TSST. For post-TSST outcomes, we will conduct 5-condition omnibus ANOVAs to test any effects of experimental condition, following up on any significant results with planned comparisons between: (1) those who ate healthy food versus no food pre-TSST, (2) those who ate unhealthy food versus no food pre-TSST, (3) those who ate healthy food versus no food post-TSST, (4) those who ate unhealthy food versus no food post-TSST, (5) those who ate healthy food versus unhealthy food pre-TSST, and (6) those who ate healthy food versus unhealthy food post-TSST. To follow up on any significant stress-reducing effects of comfort eating, we will use the PROCESS statistical macro (Hayes & Matthes, 2009) to test mediation and moderation hypotheses.