Testing the Embodied Account of Object Representations

Abstract

Theories of embodied cognition propose that sensorimotor experiences constrain visual cognitive process. Specifically, these theories posit that object concepts are represented by simulations within modality-specific cortices. These theories make two critical predictions about the role of sensorimotor simulations in representing manipulable versus non-manipulable objects. First, sensorimotor simulations are functionally necessary during visual cognitive tasks. That is, in visual processing of objects, simulations are involved in producing successful performance (e.g. object naming). Second, sensorimotor simulations are coactivated in response to the visual presentation of objects. That is, we should observe evidence for incidental sensorimotor simulations during the performance of visual cognitive tasks (e.g. orientation judgments). I conducted two studies to test these hypotheses. In the first, I employ a concurrent motor task during an object-naming experiment in an attempt to disrupt sensorimotor simulations. Using this paradigm, I failed to show evidence that concurrent motor activity selectively impairs manipulable object naming; rather, I show that a concurrent motor task affects naming across categories of objects. In the second, I investigate behavioural evidence for coactivated sensorimotor simulations in response to the visual presentation of objects. I show a ‘motor potentiation effect’ for both animals and artifacts, a finding that cannot be explained by coactivation of sensorimotor simulations. In the other studies reported here, I adopt eye-tracking and electrophysiological techniques to investigate ways in which visual attention is biased by artifact and animal stimuli and show covert and overt attentional biases for the handle of manipulable objects. Overall, the results in the present set of studies are more parsimoniously accounted for by a general bias in visual attention that is determined by factors such as experimental task, rather than embodied object representations. I conclude that we must abandon a strong form of the embodiment hypothesis.