Where's my label?! Studying how a missing label and other missing features are perceived.

A. Capelier-Mourguy, Katherine Twomey and Gert Westermann presented this poster.


Previous studies have tried to determine whether labels are treated as features or markers, when learning categories. There is some evidence that the label might be treated as a feature among other features in an early stage, the behaviour shifting to be more marker-like in adults. However the role of the label is still debated, its interaction with other features remaining unclear. We plan to run an experiment challenging these questions on pre-lexical infants, pre-schoolers and adults. We will use the set of stimuli introduced in Kovic et al. (2010), namely a 5-4 categorization task with simple drawings of animal-like creatures, the categories being formed with either the more salient (head and tail) or less salient (wings and legs) features being diagnostical (highly predictive of the category). Unlike the many studies using this kind of task, we will treat the label as any other feature, making it vary amongst exemplars of a category instead of being totally diagnostical. We will then study how a missing feature influences subject’s behaviour; we are particularly interested in the effect of a missing label compared with other missing features. We will measure this effect in looking times during a test session. With respect to the Label-as-Feature theory, we expect the label to have the same role as other features, probably of high saliency. Thus, a missing label should be as surprising for the subject as a missing head or tail, depending on the exact level of saliency of the label. If however the label is treated as a marker, a missing label should have few or no effect at all on the subject’s looking times during the test session. Running this experiment at different ages will allow us to detect an eventual shift from a treatment of the label to the other during development.