Larter-Breakspear model strange extreme values for V and Z

50 views
Skip to first unread message

NeuroLife

unread,
May 29, 2021, 2:19:41 PM5/29/21
to TVB Users
Hi,

I am interested in working with the Larter-Breakspear model to investigate resting-state brain activity. I've been using the default connectome only. When I try to run the model with all default values (also default values for the hyperbolic tangent coupling function & the conduction speed) and record with a temporal average monitor (2000Hz), I get strange behavior from a subset of the nodes.

Their values of V keeps decreasing radically (reaching -3000 at the 30000ms mark or so) and their values of Z keeps increasing radically in a similar fashion. I tried multiple integration schemes, with very low (0.001ms) to high (0.5ms) integration step size, low (1e-6, even 0) to high (1) D value for additive noise (stochastic Heun) and I can't seem to get rid of this effect.

I tried to see what the nodes with strange behavior had in common and found that they all have relatively high (north of 1) initial value of Z. I tried observing the behavior of the model on the phase plane to see if the initial values were involved. It appears that the model does produce the same type of behavior as those nodes with those initial conditions, but since they are the default values, I assumed that it is considered reasonable behavior and that the dynamics between the nodes should compensate for this drop in V through excitatory input.

So, I also tried to play with the coupling parameters (including the C parameter of the model) along with the conduction speed, to see if I can get the nodes to have V values within a reasonable range, but the best I was able to get was reducing the rate at which the value of V in those nodes drop. Since I was doing a lot more simulations at the same time I was mostly doing shorter simulations ranging from 3000ms to 6000ms, but when I did longer ones with the parameters that seemed to reduce this effect, it did not really improve the results. The best I was able to get was V reaching -200 at the 30000ms mark if I recall correctly. An issue that I encountered while doing that is that if I change things too much, some other nodes start blowing up with values of V going well above reasonable range (north of 1000)

I am very new to TVB and to any kind of work in modelling/simulations, so I am wondering if there is something that I am missing that can account for this or explain it. However, considering that V should represent the mean membrane potential of pyramidal cells then it doesn't really make sense that it drops this way, no? 

To synthesize my questions:
Is it normal to see that kind of behavior in the model at default parameter values?
If not, how could I adjust the model so that the values of the state variables remain within a physiologically realistic range?

Thank you for your time,
And sorry for the long post.
Dominic Boutet

WOODMAN Michael

unread,
May 31, 2021, 6:59:30 AM5/31/21
to tvb-...@googlegroups.com

hi


I'm not very familiar with this model, but I suspect the initial conditions and/or parameter values aren't quite right.  I would start by trying to reproduce results from the papers, before trying a network model.



cheers,

Marmaduke

From: tvb-...@googlegroups.com <tvb-...@googlegroups.com> on behalf of NeuroLife <domin...@hotmail.com>
Sent: Saturday, May 29, 2021 8:19:41 PM
To: TVB Users
Subject: [TVB] Larter-Breakspear model strange extreme values for V and Z
 
--
You received this message because you are subscribed to the Google Groups "TVB Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tvb-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tvb-users/c02c5bd7-d113-48eb-bbd9-664ef941179fn%40googlegroups.com.

NeuroLife

unread,
May 31, 2021, 9:24:14 AM5/31/21
to TVB Users
Thank you for the response.

I will look more into the papers to see if I can get some insight about this. 

On a side note, I am working on this for a summer project and my time is limited, therefore if I find that it takes too much time to get a hold of using this model I will probably be looking for another one to use. So, I was wondering if you or anyone in this group would have any insights on which models could be of interest for me. 
I am looking for a model with physiologically meaningful parameters that I could use to model resting state brain activity. Ideally, I would want it to be stable even at a relatively high integration step size because I only have the summer and would like to do a lot of parameter space exploration.

Thank you for your time,
Have a nice day.
Dominic Boutet

WOODMAN Michael

unread,
Jun 1, 2021, 3:15:09 AM6/1/21
to TVB Users

hi,


The reduced Wong-Wang, Wilson-Cowan and Jansen-Rit models, in order of increasing complexity, have state variables and parameters with physiological interpretation.  I'd suggest the Wilson-Cowan as a good compromise.


The integration step size is usually inversely proportional to the fastest time scale in the system, so it is worth taking time to understand which activities you are trying to model: if you choose to model alpha activity, for instance, you can use a larger integration step than e.g. gamma band. 



cheers,

Marmaduke

Sent: Monday, May 31, 2021 3:24:14 PM
To: TVB Users
Subject: Re: [TVB] Larter-Breakspear model strange extreme values for V and Z
 
Reply all
Reply to author
Forward
0 new messages