ICTEAM Institute, Dept. Mathematical Engineering
Institute of Neuroscience, COSY
Contact and registration: Philippe LEFEVRE
Previous
studies have shown that learning to manipulate efficiently a non-rigid
object can be challenging, and that the provision of haptic feedback is
critical. In addition previous work has shown that the ability to track
with the eye a moving target is substantially improved when the target
is self-moved in comparison to when being externally-moved. To account
for this latter observation, it has been proposed that the oculomotor
system has access to an estimate of the current hand position by means
of a forward model that receives the arm efferent copy.
During the
first part of my talk I will present an adaptation study in which the
hand-target mapping was perturbed by simulating an elastic relationship
between hand motion and target motion. During the second part of my
talk, I will present some preliminary data exploring 1) the
contribution of haptic feedback in eye-hand coordination, and 2) the
contribution of the primary motor cortex (M1) in eye hand-coordination.
When we move, sensory sensitivity is strongly reduced around the time of movements. In the case of saccades, previous work has suggested that saccadic suppression contributes to maintaining spatial consistency during movement. However, that idea cannot explain the facts that saccadic suppression considerably precedes saccade onset, and lasts about twice as long as the saccade duration (>150ms). Here we show how these effects emerge naturally by simply considering that signal dependent noise makes delayed sensory signals less reliable around the time of movement. Thus, when we move, sensory signals should have less weight relative to other sources of information, such as internal priors. One easy way for the brain to implement this is to lower the gain of neural activities related to such sensory input. Thus, single modality sensation should be biased downwards during movements, as is experimentally observed. According to our view, it is the sharing of neural resources between single modality estimates and the combined estimates needed for movement that gives rise to the observed suppression. We capture these effects by constructing a model that makes estimates based on the optimal weighting of delayed feedback and its dependence on the noise associated with motor commands. More precisely, we derive an analytical expression of the conditional distribution of the present state given delayed sensory feedback, and show that the variance associated with this estimate scales with the intensity of motor commands. As a result, the weighting of sensory feedback is reduced to maintain statistically efficient control. Simulations from this model generate an important reduction of visual weight that quantitatively matches the behavioral and neural dynamics of saccadic suppression. These results suggest that reducing sensory sensitivity may be a simple mechanism whereby online motor commands are statistically well tuned to the actual state of the body during movement.
The
human hand is an amazing tool. It is one of the most important
interface between us and the world by allowing us to touch and
manipulate it. Its complexity is certainly similar to its usefulness.
To control the manipulatory and the exploratory functions of the hand,
the brain needs to collect, to store, and to process adequately a
considerable quantity of data about the objects and the environment.
When gripping an object with the index finger and the thumb only, all
this information is used to control the kinematics and the dynamics of
the movement. In some situations, we have to manipulate unbalanced
objects that produce load torques at the finger/object. The objective
of the present work was to further investigate the control strategies
during object manipulation in presence of load torques.
The results
show that the grip force adaptation is slowed down when a load torque
is present. The reproduction of a slower adaptation when manipulating
torques under microgravity suggests that the presence of load torques
experienced during movement may alter our internal estimates of
required grip force. The last hypothesis tested was that the difference
in control strategies when a load torque is present comes from a poor
ability to discriminate load torques. Indeed, a poor ability to
discriminate load torque was found during the active task. However,
this bad discrimination was not due to an incapacity of the sensory
system as the discrimination during the passive stimulation was much
better.