1/28
Looks like no tags are added yet.
What does a large cerebellum give an animal?
A large cerebellum affords an animal high-bandwidth, real-time forward internal models for sensorimotor prediction, and adaptive error correction.
The Sensory Problem of Self-Motion
Moving changes sensory evidence: As a little skate swims, its own body movements create massive water disturbances from self-motion (and self-generated electrical fields) that directly stimulate its sensory receptors.
The Challenge: How can the skate detect the subtle water disturbance from a nearby small fish (prey) without being blinded or deafened by the sensory noise of its own swimming?
What is the “Negative Image” Solution to the sensory problem of self-motion
Efference Copy & Internal Prediction: When the brain sends a motor command to move, it sends an internal copy to a cerebellum-like circuit
Generating a Negative Image: That circuit learns to predict the exact sensory consequences of the animal's own activity and creates an equal-and-opposite neural signal—a "negative image"
Canceling Self-Noise: By subtracting this predicted negative image from the incoming sensory data, the brain cancels out self-generated stimulation, leaving only unexpected external signals clearly visible
Weakly Electric Fish Example of “negative image”
Active Electrolocation (A): A weakly electric fish (like the elephantnose fish) uses an electric organ in its tail to generate an electric field around its body, which it senses using electroreceptors on its skin.
Without Prey (B): The fish's own electric discharges and tail movements create a predictable baseline electrical pattern on its skin, which its cerebellum-like circuit learns to cancel out completely.
With Prey (C): When a conducting prey item (such as a worm) enters the field, it distorts the electric field lines and creates a changed input at the skin. Because this change does not match the predicted negative image of self-activity, it stands out immediately to the fish's brain.
Predicting the Sounds We Make
When speaking,
the speech motor cortex sends a corollary discharge to the auditory cortex,
reducing its response to the expected sound of one's own voice.
The cerebellum also contributes to predicting and adapting to self-generated auditory feedback.
Predicting External Events the Animal Did Not Cause
A. Habituation to an External Cue: When a neutral external cue light (CS) is presented alone, a resting goldfish maintains a regular, baseline heart rate.
B. Conditioned Preparation: When that light reliably precedes an inevitable aversive event (US), the goldfish learns to use the cue to predict what is coming. Instead of panicking, it exhibits an anticipatory conditioned cardiac response—slowing its heart rate (bradycardia) to prepare its body before the event hits.
Feedback vs. Prediction
Using a sailing analogy, this slide highlights how motor control combines reactive and anticipatory strategies:
Feedback (Reactive — "Adjust to conditions now"): Watching the telltales (ribbons showing current airflow over the sail) lets the sailor adjust steering to wind conditions happening right now.
Prediction (Proactive — "Prepare for the approaching gust"): Spotting ripples on the water ahead signals an approaching wind gust before it hits the boat, allowing the crew to brace and adjust in advance
Cerebellar Anatomy & Brainstem Connections
Gross Anatomy & Brainstem Connections: The cerebellum connects to the brainstem via three major cerebellar peduncles:
Superior peduncles
Middle peduncles
Inferior peduncles
Deep Cerebellar Nuclei (DCN) & 3 nucleuses
The output nuclei embedded deep within the white matter (from medial to lateral):
Fastigial nucleus: Oldest phylogenetically (vestibulocerebellum/spinocerebellum connection).
Interposed nucleus (emboliform + globose): Controls limb and spinal motor execution.
Dentate nucleus: Largest in humans; highly interconnected with the cerebral cortex for planned, voluntary movements.
The Olivary Complex
Inferior Olive: A prominent, folded brainstem structure visible in cross-sections of the medulla.
Function: It is the sole source of climbing fibers to the cerebellar cortex, providing error signals essential for motor learning and timing.
The Cerebellum as a Forward Model
The cerebellum acts as a predictive feedforward controller to bypass sensory feedback delays:
Motor Command & Efference Copy: The motor cortex sends a descending command to the body via the corticospinal pathway while sending an efference copy into the cerebellum via mossy fibers.
Granule Cells: Combine command and body-state information.
Internal Prediction: Using learned cerebellar processing, it computes an estimated present state (forward model) before actual proprioceptive feedback arrives from the muscles.
Error Correction: Once delayed sensory feedback arrives from the muscle spindles, the cerebellum compares it against the prediction to update future motor commands.
Vestibulocerebellum
Anatomical Zone: Vermis / Flocculonodular
Deep Nuclei: Fastigial nuclei / Vestibular nuclei
Main Targets/function: Vestibular & reticular circuits; controls balance, posture, and eye movements.
Spinocerebellum
Anatomical Zone: Intermediate zone
Deep Nuclei: Interposed nuclei
Main Targets/function: Red nucleus & motor cortex; controls ongoing limb execution and muscle tone.
Cerebrocerebellum
Anatomical Zone: Lateral hemispheres
Deep Nuclei: Dentate nucleus
Main Targets/function: VA/VL thalamus, premotor/motor cortex; plans and times complex voluntary movements.
Cerebellar Circuitry & Forward-Model Loop
Descending command: M1, premotor, and SMA send motor commands down the corticospinal tract to drive movement.
Proprioceptive feedback: Muscle spindles/tendon organs detect muscle length and tension; signals travel to the cerebellum and inferior olive.
Efference copy: Motor cortex sends a copy of the command to the cerebellum via pontine nuclei (mossy fibers).
Integration: Granule cells/parallel fibers combine motor copy and sensory feedback to activate Purkinje cells.
Inhibitory sculpting: Purkinje cells inhibit cerebellar nuclei, shaping their output.
Predicted state: Cerebellar nuclei send excitatory signals through VL thalamus back to the motor cortex.
Error comparison: Inferior olive helps detect mismatches between predicted and actual movement.
Error learning: Climbing fibers activate Purkinje cells, driving plasticity to refine future movements.
sensory prediction error:
The discrepancy between the sensory consequence expected from an action and the consequence observed.
That discrepancy can update the state estimate and help recalibrate future control
Cerebellar agenesis
The cerebellum fails to develop during embryonic growth or is destroyed early in gestation
presentation varies dramatically from person to person as other regions of the brain attempt to compensate
Paralysis does not follow cerebellar lesions because the cerebellum is a coordinating and predictive coprocessor rather than the direct source of motor output.
The core deficit without a cerebellum
Anticipatory Braking: Normally, an accurate reach requires proactive deceleration using antagonist muscles before the hand hits the target.
Dysmetria: Without the cerebellum's predictive timing, the nervous system fails to brake properly, causing the hand to overshoot or undershoot the target
Key Deficits with Cerebellar Damage
Ataxia: Uncoordinated, jerky, and poorly sequenced voluntary movements (including cerebellar gait, which presents with a wide-based, unsteady stance).
Delayed Movement Onset: A noticeable lag between the intention to move and the actual initiation of the motor command.
Cerebellar Intention Tremor: A rhythmic, coarse tremor that worsens as the limb approaches the target (unlike resting tremors seen in Parkinson's).
Nystagmus: Involuntary, rhythmic eye oscillations (especially during eccentric gaze) due to impaired vestibulocerebellar control of gaze holding.
Patterned Movement Errors: Loss of rhythm and smooth multi-joint coordination, resulting in irregular trajectories and impaired rapid alternating movements.
What the VOR Does
Goal (Retinal Image Stability): When your head turns left, your eyes must simultaneously turn right at the exact same speed (ideal gain = 1.0) so the image stays stationary on your retina.
Feedforward / Open-Loop Speed (~10 ms): Visual feedback is too slow to keep up with rapid head movements.
Instead, head-motion signals from the inner ear drive compensatory eye movement immediately—even in complete darkness—without waiting for visual error.
VOR "Three-Neuron Arc" Pathway
First neuron: Vestibular afferent (CN VIII) from the left horizontal semicircular canal sensing head rotation.
Second neuron: Left vestibular nucleus neuron relaying the signal across the midline.
Third neuron: Right abducens motor neuron (CN VI) contracting the right lateral rectus (and via the MLF to the left oculomotor nucleus [CN III], contracting the left medial rectus) to turn both eyes right.
lesion of the left cerebellar hemisphere produces
ataxia of the left limbs
This occurs because the cerebellar pathways cross twice (or "double cross") before influencing lower motor neurons, meaning the left cerebellum controls the left side of the body.
What is Retinal Slip? How does the cerebellum play a role in this
If the VOR is poorly calibrated (for instance, due to aging, injury, or new glasses), the eyes move too slowly or too fast.
Visual Error Pathway to the Cerebellum:
Retinal ganglion cells detect the direction and speed of image slip.
This visual motion signal travels via cortical/pretectal pathways (LGN V1 MT/MST and NOT / accessory optic nuclei) to the inferior olive (dorsal cap) in the medulla.
The inferior olive sends climbing fibers directly into the cerebellar flocculus, carrying a visual teaching/error signal.
How does the cerebellum fixes the retinal slip?
Input: Mossy/parallel fibers signal head movement to Purkinje cells.
Brake: Purkinje cells inhibit vestibular nuclei, regulating eye speed.
Learning (LTD): Climbing fiber error signal (image slip) weakens parallel fiber–Purkinje synapses.
Result: Less Purkinje inhibition → stronger eye movement → compensates for image slip and restores stable vision.
How does the cerebellum learn?
it uses supervised learning, meaning it relies on an explicit "teacher" signal that yells "Mistake!" whenever a movement goes wrong
Parallel fibers constantly feed the cerebellum’s main output cells, Purkinje cells, on what the body is doing
Based on that context, Purkinje cells fire rapid, steady pulses that act as an internal prediction or braking signal for movement
If movement misses the mark or feels unexpected —> inferior olive sends an urgent error signal up the climbing fiber
forces the Purkinje cell to fire a massive, unmistakable complex spike which floods the cell with calcium and flags whichever context wires were active right before the mistake.
The Regression Metaphor of Cerebellum learning
Mossy fibers bring in the raw clues (shoulder angle, muscle stretch, motor commands).
Granule cells remix those clues into every possible combination (e.g., "shoulder angle + muscle stretch happening together") and pass them along parallel fibers.
Purkinje cells hold the volume dials (synaptic weights) for each combination. Learning is simply turning up the dials for helpful clues and turning down the dials for clues that led to an error.
How the Dials Turn: Removing vs. Adding "Ears" (LTD & LTP)
Purkinje cells control movement by acting as an inhibitory brake
LTD (Turning the Dial Down = Stronger Movement): if context signal arrives and the climbing fiber yells error —> reduces AMPA receptors —> Purkinje cell fires less —> less inhibitory brake —> stronger/faster movement
LTP (Turning the Dial Up = Softer Movement):): if context signal arrives without error —> more AMPA receptors —> Purkinje cell fires more —> more inhibitory brake —> reduced movement
Dysdiadochokinesia
The impairment of rapid alternating movements (like quickly patting a hand back and forth).
Speech deficits
Speech can become poorly timed and variably stressed.