Hot hand thinking in children
– by Andreas Wilke
People often see patterns in completely random data sets, expecting streaks and clumps where none exist. Maybe you’ve recently felt like you’re “on a roll,” or spotted an interesting pattern that later turned out to be non-existent. Here’s the fascinating twist: This tendency isn’t just found in adults. Our new research shows that children as young as three share this propensity—and often even more strongly than adults.
Our recent study published in Evolution and Human Behavior explores how these misperceptions of randomness develop in early childhood. Our findings confirmed something profound about human cognition: Our brains are essentially “wired” to expect streaks and clumps, because, for most of our evolutionary history, resources in nature that we foraged for weren’t random—they were patchy, where a resource encounter often predicted another encounter. If you cast your line and catch a fish, you’re actually pretty smart in thinking more fish may be nearby as well. Similarly, if the first person you encounter in a village speaks your language you will assume others will too. But while we evolved in a world filled full of positively auto-correlated patterns, there are many modern-day settings where events are completely independent leading us to make poor inferences.
Our research reminds us that what sometimes looks like a cognitive fallacy—such as the well-researched hot hand phenomenon—may actually be a design feature, a mental shortcut tuned for survival and resource search in patchy environments. Our brains evolved to expect clumps because, in nature, pure randomness was (and still is) quite rare. This evolved adaptation now still colors how we see the world, from interpreting sequential events to predicting spatial arrangements in various life domains.
So why do we do this? Evolutionary cognitive psychologists argue that this propensity is actually rational from an ecological and evolutionary perspective. For millions of years, humans foraged in environments where resources—plants, animals, water—were clumped together. If you found berries in one spot, chances were good you’d find more nearby. Assuming clumpiness was adaptive for survival, and using any other search strategy in truly random environments would not have lead to lower payoffs. Today, that same mental shortcut misfires in domains like sports, gambling and finance, where outcomes can be truly random.
Until now, research on the hot hand phenomenon focused on adults. Our team wanted to know: When does this bias emerge developmentally? Is it learned through experience, or is it an evolved default present from birth?
To find out, we tested more than 300 children aged 3 to 10 in the U.S. and Germany, using three playful, tablet-based decision-making tasks:
- The animal foraging task: Children helped a cartoon rabbit guess whether a carrot was hidden under each spot along a path. The resource sequence path was random, but we measured whether they expected to find clumps of carrots.
- The raindrop task: Young participants tapped where they thought raindrops would fall on a basketball court, revealing their mental model of randomness also in two-dimensional space.
- The tree task: Children placed apples (which “like to grow close together”) and bird nests (which “like to be spaced out”) on a tree, testing their grasp of positive and negative spatial dependency.
Across all three tasks, children showed a strong bias toward streaks and clumps.
In the animal foraging task, younger children (ages 3–7) consistently predicted streaks, with subjective alternation probabilities well below the objective random benchmark. Older children (ages 8–10) were somewhat better in the task, but often reverted to seeing more clumps in the sequence the longer they played the game.
In the raindrop task, nearly every child produced a pattern that was highly aggregated, as if raindrops fall in clusters. Very few children—often only of older age— created actual random distributions.
In the tree task, kids were better at grouping apples together than spreading bird nests apart, suggesting an earlier readiness for reasoning with positive spatial dependency (clumping) and a delayed grasp of negative dependency (dispersion).
Overall, these assumptions of clumpiness weakened with increased age, but even 10-year-olds still lacked a sound understanding of randomness. Adults, tested in comparison benchmark samples, were more accurate—but also far from perfect.
These findings suggest that hot hand thinking isn’t just a cultural artifact or learned disposition. It appears early in life, likely reflecting an evolved cognitive default shaped by our species’ foraging past. For most of human history, assuming clumps was a smart bet. Today, that same bias can lead us astray.
There are intriguing possibilities for future research. On the one hand, we believe it would be informative to look for ways to identify hot hand thinking even earlier in life—including during infancy, to see what expectations of clumpiness look like, with the most minimal experience and exposure to socio-cultural context. On the other hand, further investigations could look at the effects in later life, say during adolescence, and develop educational trainings to help young people learn what truly random patterns look like.
Understanding this developmental trajectory has practical implications. Teaching statistical literacy early on could help children distinguish real patterns from random noise. This is crucial for science education and decision-making under uncertainty. Since stronger misperceptions of randomness are linked to increased gambling risk, interventions targeting these biases in youth could help reduce problem or pathological gambling later on in life.
So the next time you catch yourself seeing a streak or a pattern in the world out there, remember that it might not be real. It’s a deeply human tendency—one that starts in childhood and tells a story about our evolutionary past.
