Please recommend a book about simplification of detailed neuron models for the simulation of large networks

The basics of how to develop, test, and use models.
Post Reply
vogdb
Posts: 37
Joined: Sun Aug 13, 2017 9:51 am

Please recommend a book about simplification of detailed neuron models for the simulation of large networks

Post by vogdb »

Hi! I'm a beginner in this field. Can you please suggest a good book that explains and shows on examples how to simplify detailed neuron models for using in large networks? I have some questions, which I would be glad to be answered.
- What type of neuron is better to pick? Integrate-And-Fire? Rate-coded?
- How to translate axon/dendrite dimensions into delays, time constants etc?
- How different types of channels influence the choice? Would be there any difference if I use a simple HH model when I have various voltage and calcium dependent channels?
- How to replace an electrode? Form the experiment a have an exact data of voltage and current for my electrode. What the electrode should be translated into? Some generator?

I understand that the purpose of my simplified model defines my choice. So I am asking for materials where I can learn how to answer those questions when I'm simplifying.

I see some books like "Principles of computational modelling in Neuroscience" or "From single neurons to networks" and others. What would be the best choice for me?
ted
Site Admin
Posts: 6287
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: Please recommend a book about simplification of detailed neuron models for the simulation of large networks

Post by ted »

First realize that all understanding is based on simplification and abstraction. No hypothesis or conceptual model (synonyms, actually, for "an idea about how things work") takes all details of the physical world into account--it is always a simplification of reality. Any computational model that is useful for research is nothing more than a conceptual model expressed in computational form. The scientific challenge lies in coming up with a conceptual model. Expressing it in a form suitable for computation may take a lot of work, but it's actually only a secondary task.

With regard to methods for simplifying neurons, I am not aware of any book or article that presents an objective summary and comparison of approaches. There are plenty of books and articles by authors who have decided on one approach or another, for their own reasons that they generally state openly somewhere. A cynic might say it's like being surrounded by blind men who will tell you that an elephant is like a wall, or a rope, or a tree trunk, but an optimist would say that for some purposes a wall or rope or tree trunk might be a reasonable approximation to an elephant. Your best guide to deciding what approach to take is to first decide what kind of problems you are interested in. If you are really interested in the biological mechanisms that underlie neural function, and how function is disrupted by injury or disease, you'll need an approach that preserves an accurate representation of the details that you think are essential for that function. You have to decide which details are important and which are unimportant, and a rational decision can only be based on what is known about the phenomena of interest to you, and the likely biological substrate of those phenomena.

Knowledge of the scientific literature can help you develop experience and judgement that will help guide your decisions. Since you brought up the question of rate vs. spike timing, here's a paper that might be helpful:
Brette, R.
Philosophy of the Spike: Rate-Based vs. Spike-Based Theories of the Brain.
Front Syst Neurosci. 2015 Nov 10;9:151.
https://www.frontiersin.org/articles/10 ... 00151/full
PMID: 26617496
vogdb
Posts: 37
Joined: Sun Aug 13, 2017 9:51 am

Re: Please recommend a book about simplification of detailed neuron models for the simulation of large networks

Post by vogdb »

Thank you so much! I completely agree with you about conceptual model. The funny thing is that I chose an example of "rate vs spike timing" randomly and I am more interested in other things: 1. How to collapse a huge compartmental model into a IAF-like neuron. 2. What to do with multiples channels during this collapse? If you know any good material about, I would be happy to learn it.
ted
Site Admin
Posts: 6287
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: Please recommend a book about simplification of detailed neuron models for the simulation of large networks

Post by ted »

Suggest you examine the NEURON Bibliography for papers that contain the string
simplif
or
reduce

You might also want to read papers by authors who use software that does not easily accommodate multicompartment models--BRIAN and NEST come to mind--to see how they deal with this issue.
Post Reply