Abstract
Where stimuli are possibly discriminable but do not elicit spontaneous differential looking it is customary to use one or other variation of the habituation–dishabituation method in order to investigate their discriminability, and the literature abounds with such studies. However, there are many variations on habituation–dishabituation procedures, varying both within and between the sensory modalities investigated. Additionally, infant attention is a dynamic process which is likely to change over time, and given the great variations in procedures it can often be difficult to know whether infants are giving a familiarity or a novelty preference at test, which can make interpretation of results a difficult, and often controversial problem. Houston‐Price and Nakai give some clear examples of the changing nature of infant preferences, and relate them particularly to the amount of familiarisation prior to testing. The model they favour (and which receives a good deal of support) is that put forward by (among others) Hunter and Ames (1988) in which preferences change over the course of familiarisation time, from preference for neither familiarity or novelty, to familiarity, to no preference, to a novelty preference. They suggest that ‘If an attentional shift from what is familiar to what is novel is always found as the encoding of a familiar stimulus… is completed… then the pattern of change in preference over time should provide crucial information regarding the identity of the obtained effect’. I will argue that this attentional shift is not always found, and that in many instances we need converging evidence in order to clarify experimental findings. Copyright © 2004 John Wiley & Sons, Ltd.