A Neurocomputational Model of Impaired Imitation

This abstract addresses the question of human imitation through convergent evidence from neuroscience, using tools from machine learning. In particular, we consider a deficit in imitation of meaningless gestures (i.e., hand postures relative to the head) following callosal brain lesion (i.e., disconnected hemispheres). We base our work on the rational that looking at how imitation in apraxic patients is impaired can unveil its underlying neural principles. We ground the functional architecture and information flow of our model in brain imaging studies. Finally findings from monkey brain neurophysiological studies drive the choice of implementation of our processing modules. Our neurocomputational model of visuo-motor imitation is based on selforganizing maps receiving sensory input (i.e., visual, tactile or proprioceptive) with associated activities [1]. We train the connections between the maps with anti-hebbian learning to account for the transformations required to translate the observation of the visual stimulus to imitate to the corresponding tactile and proprioceptive information that will guide the imitative gesture. Patterns of impairment of the model, realized by adding uncertainty in the transfer of information between the networks, reproduce the deficits found in a clinical examination of visuo-motor imitation of meaningless gestures [2]. The model makes hypotheses on the type of representation used and the neural mechanisms underlying human visuo-motor imitation. The model also helps to gain more understanding in the occurrence and nature of imitation errors in patients with brain lesions.

Neurocomputational modeling of imitation deficits

Aude Billard, Biljana Petreska von Ritter-Zahony

We are interested in the question of human imitation and we address this through convergent evidence from neuroscience. The rationale of our work is that the nature of imitation deficits following brain lesion can unveil some of the neural and computational principles underlying normal imitation. In particular, we consider how imitation of meaningless gestures (i.e., hand postures relative to the head) is impaired in apraxia [1], i.e., an inability to perform voluntary movements that cannot be explained by elementary motor, sensory or cognitive deficits. We have first considered a clinical study of visuo-motor imitation of meaningless gestures in an apraxic patient with damage to the corpus callosum [2]. Interestingly the patient made different errors depending on the visual field of presentation of the stimulus and the hand used, with preserved imitation only in the right visual field – right hand condition. These observations brought us to consider modeling of the information pathway across the two hemispheres and the type of connectivity impairment which would account for them. We have thus developed a leaky integrator neurocomputational model of the neural representations of different sensory information (e.g., visual, tactile or proprioceptive) and reciprocal nonlinear transformations necessary to perform the imitation task [3]. The sensory representations in our model are self-organizing maps whose interconnections are trained with anti-hebbian learning. The information flow and implementation details of the model are consistent with evidence from brain imaging and neurophysiological studies [4,5]. To simulate callosal apraxia, we added uncertainty in the transfer of information between sensory representations localized in different hemispheres, successfully reproducing the results found in [3]. To summarize, our model makes hypotheses on the type of neural representations used and the computational mechanisms underlying human visuo-motor imitation and could possibly help to gain more understanding in the occurrence and nature of imitation errors in patients with brain lesions. In addition, to further test and validate the model against human motion experimental data, we conduct, in collaboration with the Geneva University Hospital (HUG) and Vaud University Hospital Center (CHUV), kinematic studies with brain damaged adults specifically disabled in gesture imitation.

2008

A Neurocomputational Model of an Imitation Deficit following Brain Lesion

Aude Billard, Biljana Petreska von Ritter-Zahony

This paper investigates the neural mechanisms of visuo-motor imitation in humans through convergent evidence from neuroscience. In particular, we consider a deficit in imitation following callosal brain lesion, based on the rational that looking at how imitation is impaired can unveil its underlying neural principles. We ground the functional architecture and information flow of our model in brain imaging studies and use findings from monkey brain neurophysiological studies to drive the choice of implementation of our processing modules. Our neural model of visuo-motor imitation is based on self-organizing maps with associated activities. Patterns of impairment of the model, realized by adding uncertainty in the transfer of information between the networks, account for the scores found in a clinical examination of imitation (Goldenberg et al, 01). The model also allows several interesting predictions.

2006

Modeling the neural correlates of imitation from a neuropsychological perspective

Biljana Petreska von Ritter-Zahony

Imitation is a fundamental mechanism by which humans learn and understand the actions of others. This thesis addresses the low-level neural mechanisms underlying the imitation of meaningless gestures, using tools from computational neuroscience. We investigate how the human brain perceives these gestures and translates them into appropriate motor commands. In addition, we take a relatively unexplored neuropsychological perspective, which looks at imitation following a brain lesion. The analysis of how imitation breaks down in apraxia, a complex disorder of voluntary movement, enables us to reverse engineer brain function through the identification of those building blocks that are preserved. To better understand the phenomenon of apraxia, we develop a neurocomputational model of imitation that proposes potential neuroanatomical correlates, such as the flow of information across the two brain hemispheres. The model accounts for the pattern of errors observed in apraxic patients with disconnected brain hemispheres. To validate the predictions of our model, we further analyze the experimental errors and uncover a goal-dissociation, where a goal is defined as the spatial relation between two body parts. The experimental observations suggest that the imitation deficit in apraxia arises from an incorrect coordination between the reproductions of multiple goals. A prediction of this hypothesis was validated on three apraxic patients. The collected body of kinematic and neuropsychological data allowed us to refine our neurocomputational model of imitation, and to propose a biologically plausible mathematical model for the execution stage of the imitation. The model controls movement by following nonlinear dynamics, and precisely reproduces both the spatial and temporal aspects of unconstrained and natural three-dimensional reaching movements. Importantly, the model is stable and robust against external perturbations. Overall, our computational models and neuropsychological experiments contribute to a better understanding of how the brain performs the imitation of meaningless gestures; that is, by first decomposing the gesture into imitation goals, and then reproducing these goals through the association of different sensory modalities.

EPFL2009

Neurocomputational modeling of imitation deficits

Aude Billard, Biljana Petreska von Ritter-Zahony

2008

Modeling the neural correlates of imitation from a neuropsychological perspective

Biljana Petreska von Ritter-Zahony

EPFL2009