(Over)Estimating Shared Cultural Knowledge: Evidence for early use of a knowledge prediction heuristic in children

Both adults and children accurately and efficiently predict what other people know, despite interacting with a diverse range of individuals who each have different knowledge sets. To reduce the cognitive cost of predicting each individual’s knowledge, there is evidence that we use heuristics to make generalisable predictions about the way specific kinds of knowledge are shared with others. Yet, little research examines the function of a knowledge prediction heuristic, the input needed to produce accurate knowledge predictions, or changes across development. I propose that children use a heuristic to predict others’ knowledge, and that this heuristic functions by considering the type of knowledge being predicted, and characteristics of the individual whose knowledge is being predicted. Chapter 2 demonstrates that 3- to 6-year-old children accurately and selectively predict who shares different pieces of their knowledge. Children also predict knowledge accurately in a third-party task, providing evidence for the use of a generalisable heuristic rather than simple associations or personal experience. Chapter 3 and Chapter 4 demonstrate knowledge overestimation errors, predicted by the heuristic I propose. 4-year-olds, but not 6-year-olds, overattribute knowledge to others if the knowledge item being predicted is an example of a cultural knowledge item (typically shared with strangers from the same social groups). Yet, even 4-year-olds do not make this over-attribution error when predicting an example of a typically episodic knowledge item (not typically shared with any strangers). Chapter 4 provides initial evidence that feelings of closeness or shared episodic knowledge with a partner (but not simply shared group membership) decrease 4- and 6-year-olds consideration of this partner’s perspective. Taken together, these findings provide evidence for an early-emerging knowledge prediction heuristic which considers the type of knowledge being predicted and characteristics of the individual whose knowledge is being predicted to facilitate accurate yet efficient knowledge predictions.