• Open Access

Acute sleep deprivation increases food purchasing in men


  • Disclosure: The authors declared no conflict of interest.

  • Funding agencies: This research was supported by the Swedish Research Council, Swedish Brain Research Foundation, and Novo Nordisk Foundation.

    Author contributions: The authors' responsibilities were as follows—C.C., H.S., P.G., J.B., F.R., J.C., E.N., and C.B.: designed the study; C.B. and C.C.: analyzed data; E.N. and V.N.: enrolled patients; all authors: contributed to writing the manuscript; J.C., P.H., C.C., E.N., V.N., and C.B.: collected data and conducted experiments; S.D. and H.V.: ran laboratory analyses. All authors had full access to all data in the study and take responsibility for the integrity and accuracy of data analyses.



To investigate if acute sleep deprivation affects food purchasing choices in a mock supermarket.

Design and Methods

On the morning after one night of total sleep deprivation (TSD) or after one night of sleep, 14 normal-weight men were given a fixed budget (300 SEK—approximately 50 USD). They were instructed to purchase as much as they could out of a possible 40 items, including 20 high-caloric foods (>2 kcal/g) and 20 low-caloric foods (<2 kcal/g). The prices of the high-caloric foods were then varied (75%, 100% (reference price), and 125%) to determine if TSD affects the flexibility of food purchasing. Before the task, participants received a standardized breakfast, thereby minimizing the potential confound produced by hunger. In addition, morning plasma concentrations of the orexigenic hormone ghrelin were measured under fasting conditions.


Independent of both type of food offered and price condition, sleep-deprived men purchased significantly more calories (+9%) and grams (+18%) of food than they did after one night of sleep (both P < 0.05). Morning plasma ghrelin concentrations were also higher after TSD (P < 0.05). However, this increase did not correlate with the effects of TSD on food purchasing.


This experiment demonstrates that acute sleep loss alters food purchasing behavior in men.


Obesity is perhaps the most significant public health concern faced by modern society, both reducing life expectancy and increasing the likelihood of a plethora of diseases, including heart disease, type 2 diabetes, and some cancers [1]. Despite decades of research and public concern, the obesity epidemic sweeping the Western world continues to grow. Obesity rates have more than doubled since 1980, and in 2008 more than 1.4 billion adults, 20 and older, were overweight or obese [2]. A major contributor to the epidemic is consumption of excess calories [1]. The behavioral and lifestyle factors that contribute to this consumption are still poorly understood; however, behavioral experiments have found that participants tend to over-consume during the day following sleep deprivation [3, 4], particularly when high-caloric snacks are available [5]. There are a variety of mechanisms that have been proposed to account for these effects. Some studies have found neuroendocrine adaptations resulting from sleep deprivation, including an increase in circulating levels of the orexigenic hormone ghrelin [6, 7]. Additionally, sleep deprivation leads to an energy imbalance, as being awake is more energy demanding than sleeping, and thus requires added energy input [8].

In contrast to the existing literature on sleep deprivation's impact in energy intake and orexigenic hormones has been explored extensively, to date, no study has investigated whether sleep deprivation exerts acute effects on individuals' food purchasing behavior. Purchasing behavior is distinct from consummatory behavior as it affects future food intake long beyond the acute sleep deprivation period. Some studies have linked sleep deprivation to impaired decision making [9, 10], a condition that may also leave sleep-deprived individuals more likely to purchase calorically dense foods, and quantitatively more food in general. With this background in mind, the aim of this study was to determine if sleep deprivation alters economic decision making in individuals presented with a mock supermarket. Bearing in mind that sleep deprivation has been shown to increase food intake on the following morning [3, 4], we predicted that food purchases would be increased following total sleep deprivation (TSD).



Of the original 16, 14 healthy male subjects were considered eligible for analysis (age: 23.0 ± 0.8 years; body mass index: 23.4 ± 0.5 kg/m2; all nonsmokers). Two participants were excluded because their data was recorded improperly during the supermarket task procedure. An interview before the experiment ensured that all subjects enrolled in the study reported to have a normal sleep–wake rhythm, i.e., ∼8 h per night, bedtime between 22:30 and 23:30, and wake up time between 06:30 and 07:30 on working days, and not more than 2 h day-to-day variability with regards to their sleep duration. Subjects were excluded if they reported a history of physical or psychiatric disorders, if they were currently taking any medication for any such disorder, or if they displayed disrupted sleep patterns during the adaptation night. This adaption night also served the dual purpose of habituating the subjects to the experimental sleep conditions. This study was approved by the Regional Ethical Review Board in Uppsala, and the procedures followed were in accordance with the Helsinki Declaration. All participants gave written informed consent and were financially reimbursed for their participation in the study. The study is registered with ClinicalTrials.gov, number NCT01730742.

Study design and procedure

This experiment was a component of a larger study investigating a variety of other measures related to sleep deprivation's impact on hunger levels and immune system functioning, and thus parts of the study have been reported previously [11, 12]. Data regarding ghrelin levels was also reported in a previous paper [11]. However, as this orexigenic peptide has a demonstrated role in the decision to consume or seek food through its action on central nervous system reward pathways [13], it is included here for the purpose of analyzing its relationship to our target measurement, i.e., performance on the supermarket task.

Figure 1A shows a scheme of our experimental setup. All participants engaged in two conditions (sleep vs. sleep deprivation) using a within-subject, randomized crossover design, where each condition was separated by 4 weeks. Before the experimental day, participants stayed in the experimental setting for a baseline day, including the aforementioned night of adaption sleep. During the baseline day, participants were provided with a standardized breakfast at 8:00 (pancakes with jam), a warm lunch at 13:00 (ready-to-eat meal with chicken, rice, and vegetable), a snack at 15:30 (cinnamon bun), and a warm dinner at 20:00 (ready-to-eat meal with minced meat, baked potato, and vegetable). Between meals, participants were allowed to engage in leisure activities under sedentary lab conditions and were provided with two 30-min walks, monitored by experimenters. The evening of this day comprised the experimental night (involving sleep or TSD) from 22:30 to 6:30. To prevent possible anticipatory effects, participants were blinded to the condition they were in until the evening of the experimental day (∼21:00). In the sleep condition, sleep was monitored using Embla A10 recorders (Flaga hf, Reykjavik, Iceland) comprising electroencephalography (EEG), electrooculography, and electromyography. Sleep stages were determined according to standard criteria [14]. In the TSD condition, participants were given a selection of movies, games, and books and were monitored by experimenters to ensure wakefulness. They were provided with an unlimited supply of water; however, no food items were provided during this period. In the sleep condition, lights were turned off at 22:30, and switched on the next morning at 6:30.

Figure 1.

Total grams purchased, total calories purchased, and circulating ghrelin levels after one night of either sleep or total sleep deprivation. A: Scheme of our experimental setup. B: Total calories and grams purchased averaged across all price conditions (i.e., 75%, 100% (reference price condition), and 125% for high-calorie food items). In each price condition, subjects were instructed to use their budget of ∼50 USD to the maximum, i.e., they were instructed to purchase as much as they could out of a possible 40 food items, including 20 high-calorie (>2 kcal/g) and 20 low-calorie foods (<2 kcal/g). While subjects purchased significantly more calories and grams after sleep loss than they did after one night of 8-h sleep opportunity, this effect was mediated neither by the type of food offered nor by the price condition. C: Plasma ghrelin levels between total sleep deprivation (TSD) and sleep conditions. Data are shown as mean ± SEM. *P < 0.05.

At 7:30 in the morning following the sleep intervention, blood was sampled for the assessment of plasma concentrations of total ghrelin. Blood samples were immediately centrifuged and frozen at −80°C until analysis. A commercially available ELISA was used for assaying total ghrelin (EZGRT-89K; Millipore, Billerica, MA). At 8:00 in the morning after sleep or wakefulness, subjects consumed a breakfast comprising a 500 mL yogurt drink (Gainomax Recovery Vanilj; Norrmejerier, Umeå;100 kcal/100 g) and 2 bars of crisp bread (Wasa Sandwich Cream Cheese Naturell; Barilla Sverige AB, Stockholm) (75 kcal/bar), providing 650 kcal in total. Afterward, a portion size task followed by the supermarket task presented herein was administered to the participants. Note that sleep deprivation resulted in the selection of significantly larger portions of food on the portion size task, as previously reported [11].

Supermarket task

Experimenters explained to participants that they would be participating in a mock supermarket scenario. It was explained that they should imagine that they are purchasing food items for the next several days, that their stocks at home are empty, and that there is no opportunity to purchase food during the next several days. They were given 300 Swedish kronor (∼50 USD) with which to purchase as much as they could out of a possible 40 items, including 20 high-calorie (>2 kcal/g) and 20 low-calorie foods (<2 kcal/g, see Supporting Information Table S1). Importantly, subjects were instructed to use their budget to the maximum, i.e., they were not allowed to make money savings. For each category, an equal number of easily consumable items (i.e., candy bars) and items requiring preparation (i.e., frozen goods) were selected. Each item was presented as an equally sized picture of the item, with equally sized and formatted labels describing price, calorie density (i.e., kcal/g), and weight. Calorie density, weights, baseline prices, and pictures were all drawn from items at a representative Swedish supermarket. Participants made their purchase choices by placing actual Swedish currency (in coins; provided by the experimenters; collected after the experiment) on the pictures of the items they wished to buy, and could, within the constraints of their budget, buy an unlimited number of any given item. After completing the task, participants were then presented with the same items twice more, but in these trials the prices of the high-calorie foods were varied in random order, i.e., 25% lower than baseline, and then 25% higher, or vice versa, with prices rounded up to the nearest Swedish kronor.

Data analysis

Data are presented as means ± SEM. A repeated measures three-factorial ANOVA was used to explore the interaction between the sleep intervention (Sleep vs. TSD), price levels (i.e., low, medium, and high price), and food category (i.e., high-caloric vs. low-caloric food items), for both the amount of calories and grams purchased. A recent meta-analysis from our group has shown that sleep loss increases food intake in humans [3]. Further, following TSD, increased morning fasting ghrelin concentrations have been observed in healthy young men [6, 7]. Thus, planned contrasts were used for comparisons of food purchases and plasma ghrelin between the sleep and TSD conditions. Overall, a P-value < 0.05 was considered significant.


Effects of total sleep deprivation on food purchases

Sleep was typical for the laboratory condition (Table 1). The descriptive results of the supermarket task are included in Table 2. Repeated measures ANOVA revealed significant main effects for purchased calories and grams (P = 0.04 and P = 0.01 for the ANOVA Sleep/TSD main effects), i.e., when kept awake the entire night, participants purchased significantly more total calories (+9%) and grams (+18%) the next morning (Figure 1B), as compared with purchases after one night of laboratory sleep. The three-factorial repeated measures ANOVA (sleep/TSD × food category × price), however, did not reach significance (P = 0.33 for calories and P = 0.46 for grams, respectively).

Table 1. Sleep characteristics of the sleep condition
Sleep parameter 
  1. Data are mean ± SEM.

Time in bed, min480
Total sleep time, min441 ± 6
Wake, min32 ± 4
Sleep stage 1, min6 ± 1
Sleep stage 2, min219 ± 12
Slow wave sleep, min114 ± 7
Rapid eye movement (REM) sleep, min103 ± 9
Table 2. Descriptive food purchasing task results
 Price of high-calorie food items  
 (100% = reference price)TSDMean ± SEMSleepMean ± SEM
  1. Table demonstrating the mean total grams and kcals of items purchased in each price condition under both sleep and total sleep deprivation (TSD).

Purchased grams
All food items (n = 40)75%11,913 ± 89410,207 ± 651
 100%10,992 ± 7059,037 ± 531
 125%10,935 ± 1,0319,449 ± 810
High-calorie food items (n = 20)75%3,961 ± 2803,747 ± 260
 100%3,205 ± 2613,129 ± 254
 125%2,433 ± 3032,196 ± 158
Low-calorie food items (n = 20)75%7,952 ± 1,0686,460 ± 703
 100%7,787 ± 8645,908 ± 643
 125%8,502 ± 1,2447,253 ± 878
Purchased kcal
All food items (n = 40)75%19,061 ± 80717,657 ± 842
 100%16,479 ± 86515,160 ± 1,044
 125%14,161 ± 99112,990 ± 671
High-calorie food items (n = 20)75%13,458 ± 1,00512,552 ± 951
 100%10,912 ± 1,06910,604 ± 1,183
 125%8,252 ± 1,3167,192 ± 758
Low-calorie food items (n = 20)75%5,603 ± 7635,105 ± 316
 100%5,567 ± 6144,556 ± 332
 125%5,910 ± 8465,798 ± 439

Effects of total sleep deprivation on plasma ghrelin

As reported in a different subset of the participants of the original study [11], plasma levels of total ghrelin were also higher after TSD than after a night of sleep in this sample of 14 males (Figure 1C). However, a Pearson's correlation analysis revealed that the effects of acute sleep loss on fasting plasma ghrelin the next morning were statistically unrelated to subsequent increases in food purchasing (for grams: r = 0.17, P = 0.56; for kcal: r = 0.14, P = 0.64).


Participants purchased significantly more calories and grams of food, within the same budget, following sleep deprivation. These effects were found after an energy-typical breakfast, thereby minimizing the potential confound that the energy deficit induced by sleep deprivation [8] was responsible for the observed effects of sleep deprivation on food purchasing. Going beyond recent findings that have shown that acute sleep loss increases acute food intake in humans [3, 4], the present analysis suggests that increased food purchasing in the morning after nocturnal wakefulness may represent another mechanisms through which a repeated or chronic lack of sleep, e.g., through shift work, promotes weight gain.

Participants were driven to purchase more calories and grams of food following TSD across price conditions, suggesting an increased hedonic drive, an increased level of homoeostatic hunger, or both. As we used a heavy caloric preload before administering the task to help reduce the impact of homeostatic drive, this explanation is less likely; however, there was still a residual difference in hunger between the two conditions, and thus we cannot rule out this possibility. On the other hand, there is a breadth of evidence suggesting the reward system may have been involved. As we and another lab recently reported, sleep deprivation increases the brain's response to pictures of food, as measured by functional magnetic resonance imaging [15-17]. Specifically, these studies found increases in activity levels in brain regions involved in reward, such as the putamen, nucleus accumbens, thalamus, insula, anterior cingulate cortex, and the prefrontal cortex, suggesting that sleep deprivation leaves the brain vulnerable to an increased hedonic drive. Additionally, as previously reported, following the breakfast pre-load sleep deprived participants only selected larger portions of snack foods, and not meal foods, on a portion size task [11]. Thus, sleep deprivation's impact on hedonic drive, in addition to its impairment of general decision making, provides a more plausible explanation for its associated tendency to induce participants to buy more calories and grams of food.

In support of this theory, sleep deprivation is known to increase ghrelin levels [6, 7], a peptide integrally involved in the hedonic value of food (18-21). Ghrelin also has a demonstrated role in the decision to consume or seek food through its action on central nervous system reward pathways [13]. As, under our experimental conditions, we found an increase in ghrelin levels in the TSD condition, this could be a neuroendocrine mechanism that facilitated the impact of hedonic drive on the observed increase in purchasing. However, we also report data contradicting this line of reasoning, as in our sample we found no correlation between sleep deprivation's impact on ghrelin, and its impact on food purchasing. It is important to note that this null finding could be related to our sample size or the fact that plasma ghrelin was measured ∼1 h before task administration to avoid stress associated with blood sampling. However, an alternate explanation would be that it is largely sleep deprivation's impact on higher level cognition and decision making [9, 10], and not hedonic circuitry or homeostatic hunger per se, which drove the observed effects.


The design of the study was limited to a mock scenario, which may not have induced behavior typical to actually shopping in a supermarket. The experiment was also run in a male population; while in general the impact of sleep deprivation on at least hunger and food consumption (which might serve as a loose proxy for our task) appears to be gender neutral, at least one study has found differences [22], suggesting that additional studies should confirm the effect in females. This study used a TSD paradigm—as opposed to partial sleep deprivation—which is a strength in that it increased the study's power, but a weakness in that it hampered external validity for everyday people who more frequently miss some, and not all of a given night's sleep. Additionally, this study was run with a relatively small sample size. Thus, we cannot exclude that this may have masked possible interactions between food purchasing, food type (i.e., high-caloric vs. low-caloric foods), and price conditions. Finally, there was a dearth of high priced, nonredundant high-calorie items common to Swedish supermarkets to include for analysis. That said, the results of the study may have been dampened by the relative lack of low-priced high-calorie items to choose from.


Our findings demonstrate that participants purchased significantly more calories and grams of food, within the same budget, following sleep deprivation. This is significant as the stocks that one purchases last beyond the acute sleep deprivation, and will influence food consumption choices long after the purchases are made. Bearing in mind that we chose TSD to investigate the influence of sleep loss on food purchasing behavior in humans, our findings are broadly significant for people working in a variety of professions, including shift workers, cab drivers, nurses, doctors, and other jobs requiring work at night. Follow-up studies should address whether these sleep deprivation-induced alterations in food purchasing behavior are exaggerated in obese populations, and in those with chronic sleep problems—both populations where decision making following disrupted sleep may be markedly different. Additionally, studies should investigate whether or not this impact on purchasing behavior extends to other items beyond food, including high-price items, where purchasers could fall victim to disrupted decision making.


We are grateful to Lina Lundberg and Sanaz Zarei (Department of Neuroscience, Uppsala University, Sweden) for their expert and invaluable laboratory work. This study was supported by the following funders: Swedish Research Council, Swedish Brain Research Foundation, and Novo Nordisk Foundation. The funding sources had no input in the design and conduct of this study, in the collection, analysis, and interpretation of the data, or in the preparation, review, or approval of the manuscript.