Why do infants make A not B errors in a search task, yet show memory for the location of hidden objects in a nonsearch task?

In 4 experiments, infants aged 8 to 12 months were tested on A not B search tasks, and nonsearch A not B tasks following the violation-of-expectation paradigm. A 1-location task and 2 control tasks were also conducted. In the nonsearch tasks, a toy was hidden in A, moved to B, and retrieved after a delay from either A (impossible) or B (possible). Results showed significantly longer looking times at impossible events, indicating some memory for where the object was hidden and an expectation of where it should be found. This effect occurred at delays at which infants made the A not B error when searching, and at a longer delay of 15 s. The results showed clearly that infants have some memory for the object's location, even at delays at which they search at the incorrect location. Discussion centers on how these results are accounted for within explanations of the A not B error.