Coding body and world for reaching movements

Loading...
Thumbnail Image
Date
Authors
Abedi Khoozani, Parisa
Keyword
Reaching Movements , Coordinate Transformations , Neural Network , Motor Control , Signal Dependent Noise
Abstract
To reach to visual targets, we need to transform the visual representation into the coordinate system of the moving arm, i.e. coordinate transformations. Varying body postures such as lying or rolling the head leads to higher movement variability. However, it is unknown how uncertainty in body posture estimations propagates to reaching movements. It has been speculated that higher movement variability is a by-product of noisy coordinate transformations stemming from the uncertainty in estimating the body posture. In the first study, we investigated how uncertainty of body posture estimations effects action by our novel head roll and neck load paradigm. We showed that both rolling the head and loading the neck resulted in biased and more variable movements. Using our 3D computational model, we showed that higher movement variabilities resulted from uncertainties in coordinate transformations caused by head roll and neck load. In the second study, we asked whether the brain accounts for the increased movement variability caused by varying body postures. We showed that, rolling the head results in adapted movements, such as increased safety margins while passing an obstacle. In the third study we asked how humans incorporate information about surrounding visual objects, i.e. allocentric information, when reaching to a memorized object location. We showed that the brain uses clustering mechanisms to efficiently memorize scene configurations, i.e. code allocentric information. By including the clustered allocentric information in reaching, our model predicts shifted reaching end points in the direction of shifted objects, comparable to human data. Lastly, we investigated the role of coordinate transformations in performing multisensory integrations at the neuronal level. We showed that a neural network trained to perform multisensory integrations replicates the neuronal activation associated with multisensory integrations. Our model behave comparable to a divisive normalization model but in a neurophysiologically feasible manner. Together these studies suggest that coordinate transformations must be considered as noisy processes which have tangible effect on reaching movements. Since coordinate transformations are ubiquitous, this work represents a significant advance in our understanding of how the brain performs one of its most fundamental functions.
External DOI