Infant Search and Object Permanence: A Meta-Analysis of the A-Not-B Error

Research on Piaget's stage 4 object concept has failed to reveal a clear or consistent pattern of results. Piaget found that 8-12-month-old infants would make perserverative errors; his explanation for this phenomenon was that the infant's concept of the object was contextually dependent on his or her actions. Some studies designed to test Piaget's explanation have replicated Piaget's basic finding, yet many have found no preference for the A location or the B location or an actual preference for the B location. More recently, researchers have attempted to uncover the causes for these results concerning the A-not-B error. Again, however, different studies have yielded different results, and qualitative reviews have failed to yield a consistent explanation for the results of the individual studies. This state of affairs suggests that the phenomenon may simply be too complex to be captured by individual studies varying 1 factor at a time and by reviews based on similar qualitative considerations. Therefore, the current investigation undertook a meta-analysis, a synthesis capturing the quantitative information across the now sizable number of studies. We entered several important factors into the meta-analysis, including the effects of age, the number of A trials, the length of delay between hiding and search, the number of locations, the distances between locations, and the distinctive visual properties of the hiding arrays. Of these, the analysis consistently indicated that age, delay, and number of hiding locations strongly influence infants' search. The pattern of specific findings also yielded new information about infant search. A general characterization of the results is that, at every age, both above-chance and below-chance performance was observed. That is, at each age at least 1 combination of delay and number of locations yielded above-chance A-not-B errors or significant perseverative search. At the same time, at each age at least 1 alternative combination of delay and number of locations yielded below-chance errors and significant above-chance correct performance, that is, significantly accurate search. These 2 findings, appropriately elaborated, allow us to evaluate all extant theories of stage 4 infant search. When this is done, all these extant accounts prove to be incorrect. That is, they are incommensurate with one aspect or another of the pooled findings in the meta-analysis. Therefore, we end by proposing a new account that is consistent with the entire data set.