On the representation of semantic and motor knowledge. Evidence from brain damaged patients

When we think of an apple, do we actually feel the same as when we eat it? The central theme of this work is to understand whether the permanent representation of an object corresponds to a reactivation of sensations we perceived when we actually had it in our hands. A recent debate in cognitive neuroscience, in fact, is concerned with the possibility that the neural systems that mediate overt action and sensory experience are causally involved in the neural representation of actions and real objects. On the other hand, more classical models postulate a relative separation between the how system and the what system, the former being more related to action, the latter more related to visual and semantic object representation. Such a classical view does not deny that the two streams normally have a close interaction but, based on neuropsychological and behavioral evidence, it holds that they can work separately in the case of selective brain damage or in particular experimental conditions. In this thesis I will explore the possible role of the motor processes in understanding objects and actions by studying brain damaged patients performing a series of action- and object-related tasks. In Chapter I, I will briefly introduce the literature on the relationship between actions and concepts of both healthy and brain damaged subjects. Chapter II reports a study on a group of 37 stroke patients who have been tested for their ability to recognize and use objects, as well as to recognize and imitate actions. In this group I found double dissociations suggesting that these tasks depend on separable cognitive processes. In Chapter III, I will describe a double dissociation study in which we compared the performance of two patients with apraxia with that of two patients with semantic impairment, and I will show how the object knowledge of the latter patients decline in time although they maintained relatively good ability to use objects. Finally, in Chapter IV I will analyze the performance of a new series of apraxic patients on a set of tasks aimed at testing a computational model which accounts for the errors that apraxic patients make when using objects. The results will not completely fit with the embodied theories of knowledge. Rather, they are compatible with “disembodied” models that postulate a separation between the object conceptual knowledge and the sensory-motor input and output systems.