RESPONSE Infant perseveration : Rethinking data , theory , and the role of modelling

commentaries on my model of the AK B task highlight disagreements in the field at three levels: 1) data, 2) theory, and 3) the role of modelling. Some of the data disagreements have actually been resolved by existing studies. I will not venture to claim that any of the debates about theory and the role of modelling have been resolved, but I will discuss additional data and arguments and attempt to clarify my stance on these issues. Data The commentaries highlight five disagreements about data on the AB B error: 1) Do looking measures show greater sensitivity than reaching measures? 2) Do infants perseverate to A when an object is visible at B just as they would if the object were hidden? 3) Does increasing the number of hiding locations help infant performance for non-artifactual reasons? 4) Does increasing the number of A trials increase the likelihood of the AB B error? and 5) Do infants show a U-shaped pattern of development in their AB B errors? I will argue that existing data answer Question 1 in the affirmative and Question 2 in the negative, whereas additional work may be needed to resolve the remaining questions. Diamond challenges some of the studies I cited as evidence for earlier sensitivity in violation-of-expectation variants of the AB B task that these were not equivalent to reaching versions of the task. Smith & Scheier go one step further to challenge the general claim that looking measures show greater sensitivity than reaching measures, arguing that looking and reaching are strongly coupled and the greater sensitivity claim is primarily based on 'anecdotal reports that babies sometimes look correctly but reach wrong and on experimental evidence from nonreaching tasks.' These are important cautions, but the collection of data on looking and reaching measures of AB B demonstrate that these behaviors, while strongly coupled, nonetheless show differential sensitivity in equivalent versions of the AB B task. This point is made perhaps most clearly in Ahmed and Ruffman's (1997) expectation-reach studies and Hofstadter and Reznick's (1996) gaze-reach study. In Ahmed and Ruffman's (1997) studies, infants were presented with equivalent expectation and reach variants of the AB B task; they were presented with an A trial and allowed to reach and then presented with a B trial, and either allowed to reach (generally reaching persevera-tively) or watch as the toy was revealed at A or B (generally looking longer …