cerebellar model lecture 4 Flashcards
What is the main framework and what is it trying to do
The Marr- Albus framework.
Explain how the cerebellum is important for motor control.
What is task analysis
Complex information processing problems need to be modelled at more than one level.
The computational level. - what problem is the cerebellum trying to solve:
need to know how big a command to send to the muscles wherever the target is located.
Algorithmic level
The cerebellum motor commands are learnt, using supervised learning.
- babies spend a lot of time learning about movements.
what is the learning rule
The supervised learning rule (delta rule) gradient descent.
IF movement is too small increase command if too big reduce the command. If accurate do nothing.
what do parallel fibres signal
the location of the target
when a PF fires so does…
its target Purkinje cell with the motor command
How much a purkinje cell fires depends on?
the weight of the synapse between the PF and PC
what is the weight of the synapse adjusted by
climbing fibre input/
increase weight if movement too small/. decrease it movement is too big.
Why is it plausible… granule
explains why there might be so many granule cells. A lot of factors can affect the size needed for an accurate movement
why is it plausible… climbing
explains why the climbing fibre input is so unusual. Low frequency should not drive output just correct motor command.
very reliable- whenever CF fires PC also does and whole dendritic tree affected
How does this apply to eyeblink;
- the movement is too small - no blink at all.
- the error signal (US) therefore needs to increase movement size.
- because of circuitry the US should teach purkinje cells to fire less.
The pairing of PF (CS) with climbing fibre (US) produces
long term depression in the excitatory synapses between parallel fibres and PC.
Cerebellar chip
Same basic circuit used for eyeblink conditioning as for movements in general
Thus, a cerebellar model used to explain role in motor control should also apply to classical conditioning.
Two way traffic:
Use general models (not devised with conditioning in mind) to explain classical conditioning (this lecture)
Use classical conditioning data to test these models
