Since the human metabolic system is far too messy to be formalized into neat numbers, calories counts are virtually meaningless, a new study has found. We all process food differently, and our caloric intake is largely determined by the digestive process itself rather than the hard value ascribed to the food product. For this reason, weight loss diets should rely on foods that are hard to digest rather than foods with low calorie content.

The researchers add that much of our current nutritional data is based on outdated values measured in the 19th century, and can no longer be considered accurate.  

In an accompanying editorial published in Scientific American, biologist Rob Dunn of North Carolina State University explains that humans “engage in a kind of tug-of-war with the food we eat, a battle in which we are measuring the spoils—calories—all wrong.” All foods interact with our metabolic system in different ways, and some are easier to absorb than other. For example, 170 calories of raw almonds will only yield a caloric intake of 129 calories. Conversely, the same amount of sugary cereal will yield caloric intake that often exceeds the value given on the label.

To illustrate this, the researchers conducted an experiment with mice, in which test subjects were fed either raw or cooked food. Whereas test subjects fed raw sweet potatoes lost four grams of weight, subjects fed identical amounts of cooked food gained weight instead.

According to The Daily Mail, calorie intake is also determined by each individual’s gut bacteria. For this reason, digestion of identical foods will always vary from person to person. The discrepancy helps explain why some people find it harder to lose weight than others. Obese people, for instance, may have a natural excess of highly efficient calorie-absorbing bacteria.

Rather than counting calories, people intent on losing weight should seek out raw and organic produce that is harder for the body to digest.

“Digestion is so intricate that even if we try to improve calorie counts, we will likely never make them perfectly accurate,” said Dunn.