Mind over Material: An fNIRS Study Attempting to Manipulate 3-year-olds away from a Shape Bias and into a Material One

During word learning, children’s attention becomes tuned to the regularities in their vocabulary and language manifesting as a bias to attend to shape over other object features (e.g., material) when categorizing objects. The novel-noun generalization (NNG) task is the most common assessment of this, requiring children to categorize objects based on a common feature dimension. The shape bias appears around age 2, corresponding with the onset of language production, and is most rigid at 3 (Landau et al. 1992). While participants above or below this age can be manipulated out of this bias, 3-year-olds remain consistent (Samuelson et al. 2008). Research on cognitive control development has demonstrated limitations in children’s attentional control in 3-year-olds, though the relationship between attentional control and performance on the NNG task has not been examined. The NNG uses task known, shape-based orientation trials prior to the novel trials. These warmup trials could create a task-based bias to attend to shape that children with poor attentional control skills cannot overcome in the context of manipulations to shift attention away from shape.

The current project takes a cognitive neuroscience approach examining behavioral and neural indices of attention in the NNG task using functional near-infrared spectroscopy measured bilaterally in fronto-temporo-parietal regions. The NNG task followed the protocol created by Samuelson et al. (2008) with the addition of a typical known, shape-based training condition and a material-based training condition. We also examined whether general attentional control skills are related to performance on the NNG task using the triad classification (TC) task and an embedded shape identification (ESI) task. Children in the shape-based training (M=.77) exhibited a stronger shape bias than children in the material training (M=.59, p=.039), providing clear evidence that the standard training trials bias attention towards the shape dimension.

Neural analyses of the NNG task revealed differences activation between familiar vs unfamiliar object identification and training condition children were placed in. The TC task was correlated with shape bias such that tendency to pick the identity match in TC was associated with a stronger shape bias, but no correlation was found between shape bias score and ESI.