AbstractThis peer commentary emphasizes the importance of implementing more sophisticated analytical techniques in infant research. It provides examples of information extracted from classical paradigms, the habituation–dishabituation paradigm, the visual pair preference task and the visual expectation paradigm, by means of models such as a latent variables mixture model. Although modelling techniques have been implemented in infant research for many years, they have always been a marginal phenomenon because infant data were mainly analysed using mean level comparisons. This commentary outlines the principal advantage of modelling infant behavioural data: there is no other way to obtain information on behaviour and its development at the individual infant level than through a model, and models can be adapted to the specific requirements of an experimental setting's data. However, there are expenses to such techniques: it takes time both to build and to understand a model for a specific experimental setting, and both data and code need to be made available (and thus readable, i.e. following certain style requirements). If as a field, we do not refrain from investing such time, modelling approaches can be a very useful tool for a more reliable infant research.
Read full abstract