Graphic by Maia Brown


A new Northwestern study found out why people prefer junk food when sleep deprived.

Sleep helps problem-solving, because we consolidate memories during sleep.

Data scientists predict terror groups’ lifetime lethality based on their first 10 to 20 attacks.

Why high-fat food after a restless night?

After a sleepless night, whether you are reviewing for an exam or having a party with friends, do you find yourself reaching out for fries, burgers and pizza? Scientists have been wondering why a lack of sleep changes people’s food preferences. They are also looking for ways to intervene in this unhealthy behavior because overeating high-fat food can result in health problems, such as being overweight. A new study, conducted by a Northwestern team, found out that your nose, or your sense of smell, is the one to blame.

Researchers designed an experiment to examine 25 healthy human volunteers. After a week of getting a normal night’s sleep, participants were randomly assigned to one night of deprived sleep for four hours or non-deprived sleep for eight hours. All volunteers participated in both deprived sleep and non-deprived sleep sessions, which were separated by 4 weeks to allow for sufficient recovery time. The day after each night, participants received standardized breakfast, lunch and dinner to ensure identical food intake in both sleep sessions. But researchers also prepared a buffet of snacks for participants to measure the effects of sleep deprivation on food choices and intake. They found out that participants in the deprived sleep session consumed food items with a significantly higher energy density, like chocolate chip cookies and potato chips.

Scientists found that after a sleepless night, the human brain can better distinguish between food and non-food odors. An odor-processing region in the brain called the piriform cortex is encoding smells more strongly when people do not get enough sleep. Lack of sleep also reduces the connection between the piriform cortex and the insula, which integrates information to control food intake, like the smell and taste and how much food you need. Blood analyses showed that individuals also had increased amounts of 2-oleoylglycerol after a deprived sleep. 2-oleoylglycerol is a molecule that is part of the endocannabinoid system, which is a complex network that controls appetite. An increase in 2-oleoylglycerol makes people prefer food with high energy. These changes in the brain induce people to take in more high-calorie food.

In the United States, one in three people sleeps less than six hours a night.  However, besides getting enough sleep, another way to avoid reaching out for high-fat food is simply to avoid smelling it. Food smells attract us more after a sleepless night, because our brains are more sensitive to food odors. By avoiding the enticing smells, the cravings for junk food can decrease.  

Sleep helps problem-solving

If you get stuck on a problem, try getting some sleep. Scientists have shown that you may have a better chance of solving the problem the next day after you wake up.

Scientists have already known that people process information and reactivate memories during sleep. Now, a new Northwestern study further showed that we also reactivate previously unsolved problems during sleep, and because of this, we sometimes solve a problem better after a sound sleep. Researchers hypothesized that by consolidating unsolved problems during sleep time, we facilitate our memories of problems and can find new ways to solve them.

In the experiment, researchers gave 57 participants several puzzles, each arbitrarily associated with a different sound cue. While they slept, a program presented half of the sounds associated with the unsolved puzzles. The next morning, participants tried to solve the unsolved puzzles again. They solved 31.7% of cued puzzles and 20.5% of uncued puzzles. This result supported researchers’ hypothesis that providing information about the sound related to the puzzle during sleep did improve problem-solving the next day and support sleep’s role in problem incubation.

Because difficult problems often require thinking in another way, this result showed that the brain reorganizes problem memories during sleep, and therefore possible correct ways of solving could emerge the next morning.

However, researchers also pointed out that this consolidation of problem memory during sleep only helps when you have enough background knowledge of the problem. In other words, you need to have the ability and knowledge to solve the problem, and you just haven’t found the correct way of dealing with it yet. So if you know nothing about a question and just try to sleep on it, it will not work out.

Preventing future terror attacks with data science and the business model

As the United States and many countries around the world spend more and more to battle terrorism, it has become increasingly important to predict how lethal a terror group can be and stop the most destructive group in its early stage before it grows out of control. Researchers in the Kellogg School of Management developed a model that predicts the future lethality of a terror group based on its first 10 to 20 attacks.

Researchers gathered data of terror groups between 1970 and 2014 from the Global Terror Database (GTD) and the RAND Database of Worldwide Terrorism Incidents (RDWTI). Then they used the business model to understand the organization of terror groups. They treated terror groups as businesses producing lethality, and they asked: How could you predict the success of terror groups producing this product?

Since it is almost impossible to know a terror group’s inner network and operation, researchers developed proxies from the information they could get. For example, by viewing the timing of attacks as a proxy for resources, they could know the resources and organizational strength of the terror organization. They collected data on the diversity of weapons used, the sophistication of those weapons and their attack capabilities, which measures their abilities to carry out an attack successfully. By combining those data into their model, researchers could successfully sort out the most dangerous terror organizations in the early stage. They tried their model on Islamic State (ISIS). Based on the data of its first 10 attacks, the model placed ISIS as an extremely dangerous terror group with the most potential for committing exceptionally deadly attacks.

An 800 percent increase in global terror attacks from 2000 to 2015 raised international concern over public safety. This early-warning model developed by Kellogg could help the government to use its budget cleverly and target the most destructive terror organizations before they grow too powerful.