The adaptivity of children’s search processes

Andreas Wilke, PhD

Clarkson University, Department of Psychology

Humans and other organisms must search effectively for the resources they need, whether these
are physical (e.g., food or shelter) or informational (e.g., patterns in the world, or concepts stored in
memory). Most human search studies have focused on brief (static) laboratory tasks, but being
effective in realistic search settings requires adapting to changing environments over both short
and long terms, and to changing individual abilities developmentally. In this talk, I will report on
two ongoing projects that investigate how young children grow to understand searchable patterns
as clusters and sequences.

A tendency to perceive illusory streaks or clumps in random sequences of data—the hot hand
phenomenon—has been identified as a human universal tied to our evolutionary history of
foraging for clumpy resources. In the first project, we investigated how this misperception of
randomness and ecologically relevant statistical thinking broadly develops ontogenetically. Based
on our work with adults, we developed three iPad-based decision-making tasks that assess how 3-
to 10-year-old children decide that sequential events will continue in a streak or not, their
understanding of randomness, and their ability to reason in spatially dependent terms. In a second
project, we assessed more specifically to what extent children can differentiate among objective
alternation probabilities of various kinds, specifically when they are non-random and do indeed
contain statistical regularities. The distributions that were used included clumpy ones (where a hit
is more likely to predict another hit nearby or vice versa) and dispersed ones (where a hit is more
likely to predict a miss nearby or vice versa). Our project collected data at research sites in the
United States and in Germany. Our analyses suggest that children, indeed, hold strong
expectations of clumpy resources when they search through and reason with various statistical
distributions.