Not too simple, not too complex: The Goldilocks principle drives discrimination and search

Abstract

Like Goldilocks, human babies prefer to attend to events that are neither too simple nor too complex. What is so special — or unique — about moderately complex stimuli? And might such effects generalize to visual representations in adults? If so, core visual capacities such as search and memory might operate best over such Goldilocks stimuli. To explore whether such mental processes treat different levels of complexity differently, we generated a library of visual objects across 5 complexity levels based on their cumulative skeletal surprisal (from very simple ones like various triangles to very complex polygons). When we placed these objects in a search array, we found that subjects were best able to spot the presence of moderately complex objects, compared to very simple and very complex objects. Similarly, subjects demonstrated greater memory capacity (i.e., remembered more items) for moderately complex objects, relative to very simple and very complex objects. We suggest that these effects derive from the perception of moderately complex objects as “unique”. To test this hypothesis, a follow-up study asked subjects to judge whether two briefly presented objects were the same or not, and a “uniqueness score” was derived from performance accuracy. We found that both very simple and very complex objects were hard to distinguish from their counterparts of the same complexity level, and again that moderately complex objects were seen as most distinctive from one another. We propose that moderately complex objects are treated differently by the mind because they appear unique (compared to simple and complex objects). Thus, the Goldilocks principle — not too simple, not too complex — pervades the mind, spanning both infant and adult cognition and influencing core process of attention, memory, and representation more broadly.