Effect of Instructions on Categorization of Novel Stimuli

Previous research on free categorization has shown that people will group objects based on relational or topological alignability, meaning that different objects with corresponding parts in the same overall arrangement tend to be placed into the same category. Individual distinctive features may play a role in this process by drawing attention to relevant comparisons, but alignability should ultimately determine whether or not people see a given set of objects as belonging to the same "natural kind" category. In previous experiments, our participants grouped stimuli on the basis of both alignability and individual features. However, we suspect that their use of individual features is largely a task effect: People are actively looking for and expecting to find categories in the stimulus array, and will create ad hoc categories based on arbitrary relationships if obvious natural kinds cannot easily be found. To test this, we will be giving participants a free categorization task. Half will receive standard instructions about how to label categories in an array. The other half will receive additional instructions informing them that they may be in a control group where none of the stimuli belong to the same categories, and thus they should not feel pressured to create categories if they don’t see any in the array. We hypothesize that participants given instructions which explicitly discourage creating categories where none exist will be less likely to create single-dimension categories from otherwise non-alignable stimuli, while their tendency to create categories based on overall alignability will be unaffected.