Abstract
There are two seemingly contradictory hypotheses about how humans process complex patterns across domains: one proposes that people are predisposed towards simple patterns that focus on local, proximal relationships (Frank et al., 2012; Futrell et al., 2015); the other proposes that they are “dendrophiliacs', preferring more complex hierarchical patterns (Fitch, 2014; Santolin & Saffran, 2018). Adults and children can both learn to preferentially generate hierarchical sequences (Dedhe et al., 2022; Ferrigno et al., 2020). One unanswered question is whether children and adults generate complex patterns spontaneously, on the first try. We investigated the predispositions or inductive biases of human children and adults in a visual patterning task. We used a novel sequencing task where participants spontaneously generated patterns with visual stimuli. In this task, if people are biased toward simple patterns they will minimize dependency lengths between items in the sequence. If instead they are biased toward hierarchical patterns, they will exhibit moderate or high dependency lengths to generate 1- or 2-tier tail-embedded or center-embedded hierarchies. We found that human children and adults display strong biases to generate simple patterns. Both groups have a predisposition to reduce dependency length, a metric of local proximality. Both groups successfully overcame this inductive bias by generating more complex patterns of higher dependency length in subsequent learning and generalization tasks. However, children found it more difficult than adults to override their biases. Our results dispute a strong version of the “dendrophilia hypothesis” which suggests that humans seek complex hierarchical patterns whenever possible. Instead, our findings support a weaker version of the dendrophilia hypothesis - people possess an inductive bias towards simple non-hierarchical patterns as well patterns that are hierarchical in the simplest sense. Yet, with a little experience, humans can override their inductive biases and learn to generate more complex hierarchical visual patterns.