[Eeglablist] Meaning of ITC

[Arnaud Delorme]

Dear Pal and Baris,

from my perspective, ITC is a measure (like ERP). It tells you that  
when you present a stimulus, some neural assemblies get synchronized  
in a given frequency band to this stimulus presentation and thus  
generate an EEG signal that can be recorded at the scalp. If the  
neural response has a fixed latency with respect to stimulus  
presentation, then the ITC will be high (close to 1) and we will have  
a true ERP (a true ERP occurs when the ERP is present in every single  
trial). If this neural response is desynchronized with respect to  
stimulus onset then the ITC value will be low (closer to 0).

ITC has no meaning by itself: it is only defined with respect to the  
presence of repetitive events in the EEG data and it is a measure of  
the average synchronization between these events and the EEG signal in  
the frequency domain.

Arno

On 17 avr. 09, at 06:34, Pål Gunnar Larsson wrote:

>>
>> Dear Baris,
>>
>> Thank you for your response. I agree with you. As you mentioned, ITC
>> must be related to the 'optimal point which maximizes the neural
>> operation' in a certain way. In my poor understanding of neural
>> physiology, punctuality of neuronal firing should derive from the  
>> fact
>> that each neuron has a specifically optimized 'time limit' to accept
>> pulses from other neurons, and only pulses arrived within the limit  
>> are
>> summed to decide whether to fire. In principle, temporal regulation  
>> of
>> a macroscopic electrophysiological phenomenon such as EEG still  
>> must be
>> explained by the prerequisite... that's my guess. Any comment?
>
> First very brief to your question. An action potential the jumps to  
> a postsynaptic neuron , will change its membrane potential. The cell  
> will reset the potentila relative fast. If the synaps is far from  
> the cell body ithas little influence and will soon be "forgotten" as  
> a synaps close to the cell body that will have much more "powewr"  
> over a longer time. To that there is a significant dynamic in  
> location and matabolism of reseptors which would be expected to  
> change the system. To that input to the system will change the state  
> of the system. Input may be sensory, chemical, activity from other  
> parts of the brain and so on, so yes, there are lots of static and  
> dynamic prerequisits in the system. Also, signals you put on to the  
> system will probably make changes to it.
>
> Here is a small program you can look at: If you add 100000 action  
> potentials to a time segment, you get something that look like white  
> noise, if there is no driving in the system. You may try this:
>
> for i=1:100000
> punkt=floor(rand(1,1)*987)+1;
> eeg_data(punkt:punkt+11)=eeg_data(punkt:punkt+11)+action';
> end
>
> Where action is: action=[0 -1 -2 0 2 4 2 0 -1 -3 -1 0]
>
> If you continue with in the same variable
>
> for i=1:10000
> punkt=floor(rand(1,1)*20)+500;
> eeg_data(punkt:punkt+11)=eeg_data(punkt:punkt+11)+action';
> end
>
> That is 10000 now and it may only vary over a small segment of the  
> time line (20 points).
>
> Now you see that the 10000 constrained "action potentials" totally  
> dominates the 100000 unconstrained.
>
> So to conclude, anything that influences the probability of the  
> firing rate will influence the signal in a larger degree than one  
> would initially believe and there are lots of factors changing the  
> probability. The firing rate per se is not that important.
>
> Regards
>
> Pål