Abstract
We previously evaluated the impact of visual similarity and action similarity on visual object identification using a learning paradigm where novel associations were formed between objects and actions. We taught participants to associate novel objects with novel actions, and to identify these novel objects with non-word labels. Specific objects were paired with specific actions, and findings revealed that visually similar objects paired with similar actions were confused more often in memory than when these same objects were paired with dissimilar actions. The actions associated with objects served to increase or decrease their separation in memory space, and influenced the ease with which these objects could be identified. In earlier experiments, this pairing process occurred in two steps: participants first visually identified actions performed on a cylinder, and when participants could correctly identify all actions, they visually identified stationary objects. However, this two-step process may not represent the way we learn about objects in the real world. In the present study, we varied task characteristics to more adequately represent how we learn about objects, and asked participants to concurrently focus on action and object information. We contrasted the performance of participants who completed the original experiment to those who were asked, during test trials, to visually identify objects and either (1) produce their associated action, or (2) visually identify their associated action. All three tasks produced similar patterns of results. An analysis of the effect sizes revealed that the impact of action similarity on visual object identification was strongest in the two-step process and weakest when participants were asked to visually identify objects and their associated actions. Because learning trials and the object-naming part of test trials were identical for all tasks, the findings suggest that characteristics of the action-testing task modulated the impact of action information on visual object identification.
This work was supported by an Ontario Graduate Scholarship to the first author, as well as Natural Science and Engineering Research Council grants to the second and third author. Special thanks to Ben Cole, William Oud, Leigh Bloomfield, Maria Cristina Pensa, Zarsheesh Divecha, and Michelle Manios.