Why do children from age four fail true belief tasks? A decision experiment testing competence versus performance limitation accounts

The developmental litmus test of Theory of Mind is the false belief (FB) task in which children have to represent how another agent misrepresents the world. Children typically start mastering this task around age four. The standard interpretation is that this marks the ontogenetic emergence of full-fledged meta-representational Theory of Mind. Recently, however, a puzzling finding has emerged that challenges this standard interpretation: Once children master the FB task, they begin to fail true belief (TB) control tasks (analogous to FB tasks with the exception that the agent has full informational access and thus veridical beliefs). This finding threatens the validity of the standard FB tasks and thus stands in urgent need of explanation. Here, we test two prominent explanatory approaches against each other. The perceptual access reasoning account is a competence limitation account. It assumes that children at age four do not yet engage in meta-representation, but only in simpler heuristics (“if an agent has perceptual access to relevant event, she knows and then acts successfully; otherwise she acts unsuccessfully”). These heuristics produce false positives: they allow children to perform correctly on FB tasks for the wrong reasons, but they make them fail TB tasks. In contrast, the pragmatics approach is a performance limitation account. It suggests that children at age four do have meta-representational Theory of Mind but are confused by pragmatic task factors of the TB task (involving utterly trivial, academic test questions about the beliefs of an agent in the absence of the possibility of any mis-representation etc.). The current study tested competing predictions of both accounts against each other. To this end, different versions of the TB task were implemented that systematically varied in task features deemed relevant by the two accounts. Results from 165 four- to seven-year-olds reveal that the basic TB effect (children from age four on pass FB and fail TB tasks) could be replicated but disappeared once TB tasks were modified: children mastered both FB and TB tasks when the latter were adapted in terms of heuristic and pragmatic factors. Importantly, this pattern held in conditions in which the pragmatics account predicts success, but the perceptual access account predicts failure. Overall, the present findings thus corroborate the standard picture (children use meta-representational ToM from age four, at the latest) and suggest that difficulties with TB tasks merely reflect pragmatic performance factors.