200 likes | 366 Views
1. learning vs plasticity. 2. reinforcement learning vs supervised learning. 3. circuits: VOR & OKR. 4. open vs. closed loop controllers. 5. learning: adjusting the FF controller. 6. plasticity: cerebellar LTD. VOR Circuit. E. E. H. frequency. frequency. ipsi. contra. MR. LR.
E N D
1. learning vs plasticity 2. reinforcement learning vs supervised learning 3. circuits: VOR & OKR 4. open vs. closed loop controllers 5. learning: adjusting the FF controller 6. plasticity: cerebellar LTD
VORCircuit E E H frequency frequency ipsi contra MR LR oculomotor nucl. MLF PPRF abducens nucl. vestibular nucl. semicircular canals.
Ro + kE + rĖ 300 350 300 250 200 200 150 k 4 (sp/s) / deg 100 Ro = 100 sp/s 100 r 1 (sp/s) / (deg/s) 50 0 0 -150 -75 0 75 150 -45 -30 -15 0 +15 +30 +45 eye velocity (deg/s) From Robinson & Keller 1972 Rm = Motor Neurons r / k = = 0.25 s viscoelastic time constant firing frequency (spikes/s) eye position (deg) From Fuchs & Leschei 1970
Final Motor Pathway Saccade System (Burst Neurons in PPRF) Er Ein Pursuit System (Gaze Velocity Purkinje Cell) Eout Ep Neural Integrator (NPH for H INC for V) -H Vestibulo-Ocular Reflex (VOR) Extra- Ocular Muscle E ? Motor Neuron E + Ė
Integrator Lesion: Effects Cannon & Robinson 1987
Excitatory Motions for the SCC’s on the left. horizontal anterior posterior from Carpenter, fig. 2.7 Planes of the Semi-Circular Canals cochlea left right cochlear n. semicircular ducts: vestibular n. anterior horizontal vestibulo- chochlear n. posterior internal acoustic meatus foramen magnum from KS&J, fig. 33-9 Angle at which the plane of the anterior semicircular duct crosses the midsagittal line
Rabbit OK (Afferents) I direction selectivity E superior anterior posterior H ipsiversive U+post D+ant canal: horizontal anterior posterior 0.01 0.1 1 10 inferior stimulus speed (deg/sec) Accessory Optic System DTN/NOT MTN LTN data from Simpson 1984
Rabbit OK(adjusting VOR gain) I AOS/NOT Inferior Olive Cerebellum VN E SCC + climbing fibers error signal mossy fibers Final Motor Pathway Ė MN NPH INC E
Spectacles to change VOR gain Steve Lisberger
VOR before and after learning Miles & Fuller, Brain Res.80:512-6, 1974
Learning in the feedforward controller for saccades Time during adaptation experiment Straube et al. J. Neurophysiol.77: 874-895, 1997.
Double steps of target speed for studying learning in the feedforward controller for pursuit Kahlon & Lisberger, J. Neurosci.16: 7270-7283, 1996.
Sequence of a pursuit learning experiment Kahlon & Lisberger, J. Neurosci.16: 7270-7283, 1996.
LEARNING • H I I . . I G • General Scheme for Motor Learning e.g. AOS retinal inputs ~100 ms Feedback Controller Sensed Variable Feedforward Controller eye movement Goal Eyeball + • VN SC, PPRF MT, DLPN VOR Saccades Pursuit -E I I FMP •
Cerebellar Circuit . I . sites of learning H parallel fibers climbing fiber inhibitory interneurons HGVP GC GC (contra) mossy fibers FTN (ipsi) VRN inhibition excitation inferior olive to eye muscle motor neurons after Lisberger 1998
gain Version 2: synaptic plasticity CF + PF LTD . + contra I . ipsi H The learning rule:“Marr-Albus-Ito” or “Floccular” Hypothesis Version 1: operational vestibular information retinal slip
heterosynaptic LTD homosynaptic LTD PC CF PF cerebellar LTD associative LTD Long Term Depression after Linden & Connor 1995
Cerebellar LTD . Climbing Fiber: I (PC depolarization) Ca++ NO cGMP Na+ . AMPA Parallel Fiber: H (glutamate) Na+ LTD Induction Ca++ Ca++ Ca++ IP3 PKC* DAG PLC mGluR1 PIP2 HGVP after Linden & Connor 1995