Implicit updating of object representation via temporal associations

The cognitive system can flexibly update the representations of objects upon changes in the physical properties of the objects. Can the changes ripple to the representations of other associated objects that are not directly observable? We propose that statistical learning allows changes in one object to be automatically transferred to related objects. Observers viewed a temporal sequence with pairs of colored circles where the first circle always preceded the second. When the first circle increased or decreased in size, the second circle was judged to be larger (or smaller), suggesting that changes were automatically transferred to the second object (Experiment 1). When the second circle changed in size, the first circle was unaffected (Experiment 2). The strength of transfer seemed to depend on the conditional probability between objects (Experiment 3). The findings were replicated using pairs of faces that changed in expressions (Experiments 4&5). Importantly, no observer was explicitly aware of the pairs. Thus, statistical learning enables automatic and implicit updating of object representations upon changes to temporally associated objects.