Lecture 18 6th April 2017
Neural Circuits
Modelling of Neural Circuits: Examples from the pyloric CPG of lobsters Part 2
Too quick conclusion:
“In the brain, the properties of cells vary widely between animals; every
brain has a different solution to the same problem”
Last time
We talked about the lobster pyloric CPG ...
We saw that parameters of its neurons and synapses can vary widely
from one animal to the next,
but the rhythm is the same.
And we saw that in a database of models, that in the models the same
activity patterns could also come from very different parameters
Lets put it to the test: Response to perturbations (agitations)
Experiment: Block Ia using 4AP (a specific blocker of that current) and
observe changes in network dynamics in different animals
Model: Remove Ia from all neurons and observe the changes in dynamics
in the “equivalent models”
The effect of remove IA from the equivalent models:
- Perturbing system and don't have same behaviour are different
Not only is the control behaviour of the pyloric circuit very reproducible but also
its response to perturbations.
The models are not equivalent, and they don’t seem to describe the response in
the real system
Experiment shows that if you block a particular current, it causes a
particular change in behaviour.
2nd attempt: Dynamics of single neurons with current injections
Measure the response of an identified neuron (LP) to current injections
Inject the same currents into “equivalent models” of the LP neuron
Discussion 1
The models chosen based on control behaviour are not as consistent as
real neurons are
Do we just need more stringent criteria? (Only select models which also
pass the two additional tests we just looked at)
One prerequisite for this approach to work would be that we are actually
examining “all relevant models”
Discussion 2
It seems impossible to sample all relevant models!
Is there a notion of covering relevant regions?
Neural Circuits
Modelling of Neural Circuits: Examples from the pyloric CPG of lobsters Part 2
Too quick conclusion:
“In the brain, the properties of cells vary widely between animals; every
brain has a different solution to the same problem”
Last time
We talked about the lobster pyloric CPG ...
We saw that parameters of its neurons and synapses can vary widely
from one animal to the next,
but the rhythm is the same.
And we saw that in a database of models, that in the models the same
activity patterns could also come from very different parameters
Lets put it to the test: Response to perturbations (agitations)
Experiment: Block Ia using 4AP (a specific blocker of that current) and
observe changes in network dynamics in different animals
Model: Remove Ia from all neurons and observe the changes in dynamics
in the “equivalent models”
The effect of remove IA from the equivalent models:
- Perturbing system and don't have same behaviour are different
Not only is the control behaviour of the pyloric circuit very reproducible but also
its response to perturbations.
The models are not equivalent, and they don’t seem to describe the response in
the real system
Experiment shows that if you block a particular current, it causes a
particular change in behaviour.
2nd attempt: Dynamics of single neurons with current injections
Measure the response of an identified neuron (LP) to current injections
Inject the same currents into “equivalent models” of the LP neuron
Discussion 1
The models chosen based on control behaviour are not as consistent as
real neurons are
Do we just need more stringent criteria? (Only select models which also
pass the two additional tests we just looked at)
One prerequisite for this approach to work would be that we are actually
examining “all relevant models”
Discussion 2
It seems impossible to sample all relevant models!
Is there a notion of covering relevant regions?
