The Organization and Planning of Movement Ch The Organization and Planning of Movement Ch. 33, “Principles of Neural Science”, 5th Ed. 陳韋達 MD PhD 臺北榮總神經醫學中心 主治醫師 國立陽明大學醫學系/腦科所 副教授 哈佛大學醫學院麻省總醫院生醫影像中心 進修
Outline Motor action = the ultimate function of the sensory system Step 1: Sensory input Step 2: Sensorimotor transformation Step 3: Motor output Conscious processes are not necessary for movement control Motor learning
Step 1: Sensory Inputs Extrinsic information Intrinsic Eg. Spatial location of a target Egocentric space Mainly from visual or auditory input Intrinsic Kinematic information Position, velocity and acceleration of the hand, joint angles, muscle lengths From muscle spindles [Golgi tendon organs] Kinetic information The forces generated or experienced by our body Reference
Egocentric space and coordinates The choices of coordinate system in the brain is dependent on motor tasks and different stages of sensorimotor transformation
Difference sensory processing for action and perception Optic ataxia Visual agnosia Reference
Step 2: sensorimotor transformation Reference
kinematic vs. dynamic tasks Dynamic and kinematic tasks involves different sensory modality Dynamic: proprioception > vision Kinematic: vision > proprioception Http://brainconnection.brainhq.com/2011/04/26/the-day-his-world-stood-still/ “Motion study” is a catch-all term for simulating and analyzing the movement of mechanical assemblies and mechanisms. Traditionally, motion studies have been divided into two categories: kinematics and dynamics. Kinematics is the study of motion without regard to forces that cause it; dynamics is the study of motions that result from forces. Other closely related terms for the same types of studies are multibody dynamics, mechanical system simulation, and even virtual prototyping. Kinematic analysis is a simpler task than dynamic analysis and is adequate for many applications involving moving parts. Kinematic simulations show the physical positions of all the parts in an assembly with respect to the time as it goes through a cycle. This technology is useful for simulating steady-state motion (with no acceleration), as well as for evaluating motion for interference purposes, such as assembly sequences of complex mechanical system. Many basic kinematic packages, however, go a step further by providing “reaction forces,” forces that result from the motion. Dynamic simulation is more complex because the problem needs to be further defined and more data is needed to account for the forces. But dynamics are often required to accurately simulate the actual motion of a mechanical system. Generally, kinematic simulations help evaluate form, while dynamic simulations assists in analyzing function.
Step 3: motor output (internal models) Evidence of presence: motor equivalence Reference
Step 3: motor output (internal models) The forward model forms predicted behaviors and works as a guide for INITIAL motor output The actual behavior starts from initial motor output (predicted behavior derived from feedforward model) then corrected by feedback controls Reference
Movement errors and variability The difference between desired and actual movement Causes for movement errors Neural noise of sensory neuron (input) Inaccurate internal models Neural noise of motor neuron (output) Example 1: variability of constant force Increased with force level Example 2: speed-accuracy trade-off
Speed-Accuracy Trade-off
Fitt’s law: speed vs. accuracy =the index of difficulty
Invariant and variant features of complex movements Reference
Invariant feature Variant feature Variant feature Hand path and speed are invariant (stereotypical) feature Variant feature
Movement Schemas = The simple spatiotemporal elements of a complex movement Reference
Bibliotics
Movement control Feedforward control (open-loop) No sensory feedback is needed Feedback control (closed-loop) Sensory-dependent Some movements only involve forward model Saccade, deliberate reading, vestibular-ocular reflex (VOR) Pros: save the time for sensorimotor delay Cons: inaccuracy Reference
Sensorimotor delay Sensory feedback is noisy and slow The delayed information dose not reflect the the present state of the body and world Two compensation strategies Intermittency of movement Prediction of changes in body states (better) Reference
Prediction corrects sensorimotor delays Reference
Observer model of motor prediction Gain = the movement error
Motor learning Most forms involve procedural or implicit learning (without consciously thinking) Require different forms of sensory information Improves motor control in novel situations Motor control circuits are not static but modified throughout life Reference
Machines can beat a chess master but not the dexterity of a boy !