McCulloch-Pitts or Boolean neurons in NEURON?

Anything that doesn't fit elsewhere.
Post Reply
pjthomas
Posts: 6
Joined: Fri Jun 05, 2009 9:00 am

McCulloch-Pitts or Boolean neurons in NEURON?

Post by pjthomas »

I am collaborating with a biologist who would like to incorporate Boolean switching elements (like McCulloch-Pitts units) into a network model.
Is there an artificial cell that has a 2-state (or n-state) discrete "membrane potential", e.g. V is either 0 or 1; the artificial cell can receive inputs (e.g. via a NetCon or similar), it can change state based on those inputs; and when the cell is in state 1 it continuously emits an "on" signal that can be read by other cells (it could open a synaptic conductance in another cell, for example)? What I am hoping to do is different from any of the integrate-and-fire models, which only send a "spike" signal once when the spike is triggered.

If I need to introduce a fundamentally new kind of artificial cell, can I do that with NMODL or do I need to change something deeper in NEURON's core logic?
ted
Site Admin
Posts: 5889
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: McCulloch-Pitts or Boolean neurons in NEURON?

Post by ted »

pjthomas wrote: Wed Feb 17, 2021 2:49 pm I am collaborating with a biologist who would like to incorporate Boolean switching elements (like McCulloch-Pitts units) into a network model.
Some biologist. McCulloch-Pitts! Maybe thinking about gene expression, or perhaps "high-level phenomena" like learning or social interaction.
Is there an artificial cell that . . .
All things are doable via programming. In this case, NMODL with a NET_RECEIVE block is the way to go (alternatively could be implemented with a class definition written in hoc or Python that makes use of FInitializeHandler and cvode.event, but NMODL seems "cleaner" and would execute much faster). I'll cook up a toy example and email it to you.
pjthomas
Posts: 6
Joined: Fri Jun 05, 2009 9:00 am

Re: McCulloch-Pitts or Boolean neurons in NEURON?

Post by pjthomas »

Great, thank you Ted, I look forward to your toy example! As for the biological significance, here is the paper for you to enjoy.

Webster-Wood, Victoria A., et al. "Control for multifunctionality: bioinspired control based on feeding in Aplysia californica." Biological Cybernetics 114.6 (2020): 557-588.

It's a motor control model in which the bursting neurons are represented as boolean switching elements
(like a firing rate model only more numerically efficient) and the slower muscle components are
represented in continuous time. My goal is to reproduce the whole thing in NEURON instead of
matlab. Part of the motivation is speed: to be able to run a large ensemble with different parameters
in faster than real time while plugged in to an actual slug to predict (and maybe even steer)
their real-time, real-life movement. So yes, a real biologist got me into this...

You can stop by Vickie's presentation at COSYNE if you want to skip the paper and just see the model in action.
Post Reply