Curse of Simplicity in High-Dimensional Environments

Human beings are constantly bombarded with information. In any given moment, our senses take in a flood of details: the color of a car, the sound of a siren, the texture of a table, the smell of rain. Our brains cannot process every single piece of data at once, so they rely on a strategy called selective attention. This is the mental ability to focus on what seems important while ignoring the rest. It is a survival mechanism that allows us to function in a complex world by filtering out the noise and zeroing in on the signal. Usually, this process works beautifully, helping us learn quickly and efficiently. But what happens when the world becomes too complex, and the very strategy that usually helps us becomes a trap?

Researchers at The Ohio State University recently investigated this question by studying how people learn to sort objects into categories when faced with a high number of confusing details. They discovered a phenomenon they call the "curse of simplicity." The core idea is that our brains have a strong preference for simple explanations. When we learn something new, we naturally try to find the easiest rule that works, even if that rule isn't the most accurate one. In a simple environment, this preference for simplicity is a strength. But in a complex environment filled with many distracting options, this same preference can lead us to ignore the most important information, causing us to get stuck in a state of learning that feels right but is actually wrong.

To test this, the researchers designed a computer game where participants had to learn to identify two different types of fictional amoebas. Each amoeba was made up of several distinct "arms," and each arm had a specific feature, like a color or a shape. The participants' goal was to figure out which arms determined whether an amoeba belonged to Group A or Group B. The game was rigged with three types of arms. One arm was the "truth-teller": it always correctly predicted the group. Another arm was a "distractor": it had no connection to the group at all. The third type was a "tricky predictor": it was usually correct, but it was wrong about a quarter of the time.

The experiment varied the difficulty by changing how many arms the amoebas had. In the easiest version, the amoebas had three arms. In the medium version, they had five. In the hardest version, they had seven. To see exactly what the participants were looking at, the researchers used a special cursor-contingent display. The screen was black except for a small circle around the mouse cursor. Participants could only see the features of the amoeba when they moved their mouse close to a specific arm. This allowed the researchers to track exactly which arms the participants chose to look at and for how long, giving a precise map of their attention.

As the participants played through dozens of rounds, receiving feedback on whether their guesses were right or wrong, a clear pattern emerged. In the easiest version with three arms, most people quickly figured out which arm was the truth-teller. They stopped looking at the other arms and focused almost entirely on the one that mattered. Their learning was fast and accurate. However, as the number of arms increased, the participants' behavior changed. In the versions with five or seven arms, people were much less likely to find the single truth-telling arm. Instead, they often latched onto one of the "tricky predictors." Because the tricky predictor was correct most of the time, it felt like a good rule. The participants would focus on it, ignore the truth-teller, and convince themselves they had solved the puzzle.

This preference for the simpler, slightly flawed rule came at a cost. When the researchers tested the participants with tricky examples where the truth-teller and the tricky predictor gave opposite answers, the people who had fallen for the simple rule made many mistakes. They had built a mental model that was simple but incomplete. The more arms the amoebas had, the more likely people were to fall into this trap. In the seven-arm condition, a significant number of participants never found the true rule, even after extensive practice. They had successfully simplified the problem, but in doing so, they had sacrificed accuracy.

To understand why this happened, the researchers built a computer model that simulated how people learn. They tested different theories of how the brain might balance the desire for accuracy with the desire for simplicity. Some models assumed the brain only cared about being right. Others assumed the brain tried to minimize the total amount of mental effort. The model that best matched the human data was one that treated learning as a trade-off. It suggested that the brain is constantly trying to reduce the effort of paying attention while also trying to get the answer right. When the environment is simple, these two goals work together. But when the environment is complex, the brain's urge to save effort can overpower the need for perfect accuracy, leading it to settle for a "good enough" rule that is actually a dead end.

The study suggests that our natural instinct to simplify the world is a double-edged sword. In a world with few variables, it helps us learn quickly. But in a world overflowing with data, that same instinct can blind us to the most critical facts. The researchers found that as the number of options grew, the likelihood of getting stuck in a local optimum—a solution that feels right but isn't the best—increased dramatically. This "curse of simplicity" reveals a fundamental limit in human cognition: we are so eager to find a simple pattern that we sometimes stop looking for the truth before we have found it. The findings do not suggest that simplicity is bad; rather, they show that the benefits of simplicity depend entirely on the complexity of the environment. In a crowded, noisy world, the urge to make things simple can be the very thing that keeps us from understanding them.