Analysis of the category structure of Experiment 1.

<p>Rows correspond to values of “nucleus roundness”; columns correspond to values of “membrane thickness.” Any cell resides at one entry in the matrix (circles in the matrix). A single exemplar consists of three cells (i.e., three circles from the matrix). Members of X consist of three cells connected by lines of finite positive slope (such as the dashed line). Members of Y consist of three cells connected by lines of finite negative slope (such as the solid line). Gray bars indicate values withheld from training exemplars for use during transfer. Circles marked “G” correspond to cells that can appear only in category X; circles marked “g” appear only in X during training, but in either X or Y during transfer. Cells marked “H” only appear in category Y; those marked “h” appear only in Y during training, but in either X or Y during transfer. Values “g” and “h” are misleading in that a participant who learns to categorize based on them (i.e., based on feature conjunctions) will make systematic errors during transfer. Unmarked cells appear with equal likelihood in X or Y.