Object permanence, a result revised twice

Jean Piaget described in the 1930s and 1940s that an infant under eight months stops searching for an object as soon as it is hidden, and drew from this the idea that the object does not yet exist for them outside perception.

He also described a persistent error after that age: the child searches for the object where they found it the previous time, even having seen it moved. That error has remained one of the best replicated phenomena in the field. It is known as the A-not-B error, after the two locations used in the task.

From the 1980s, Renée Baillargeon and others proposed protocols measuring fixation time in front of possible and impossible events, without asking the child to act. The best-known experiment, published in 1985 with Elizabeth Spelke and Stanley Wasserman, showed a rotating screen that seemed to pass through an object placed behind it.

That work reported signs of object permanence far earlier than Piaget predicted, in infants a few months old. The interpretation was that the failure Piaget observed stemmed from the demands of action, not from representation.

Fixation time is an indirect measure: prolonged looking may signal surprise, but also a perceptual preference for an unusual scene, independently of any expectation. Critics showed that some results could be explained by the perceptual familiarity or novelty of the scenes presented, without assuming any representation of the hidden object.

Several multi-laboratory replications published from the 2010s found weaker effects than the initial publications, and the field now discusses methods as much as results.

The error Piaget described is robust and replicates. The explanations offered have changed: rather than a representational deficit, difficulty inhibiting an already practised motor response is now invoked. Adele Diamond in particular linked performance on this task to the maturation of the prefrontal cortex, in work conducted from the 1980s onwards.

That revision is an example of ordinary scientific continuity: the observation holds, the explanation moves, and the phenomenon changes name without changing content.

The same failure may have several causes, and the choice of measure decides what is attributed to the child. A task requiring action measures something other than a task requiring only looking. Each task also draws on distinct capacities, memory, motor control, attention, which do not develop at the same pace.

That is one reason published ages vary so much across protocols, and why claims about what an infant understands always call for the question of how they were asked.