Hello, Thank you very much for your answer. The model I am trying to implement in NEST is explained in details in the article in attachment of this email. In case it is more convenient for you, here is also the reference : Vincent Fontanier, Matthieu Sarazin, Frederic Stoll, Bruno Delord, Emmanuel Procyk. Local in- hibitory control of frontal network metastability underlies the temporal signature of cognitive states. 2021. hal-03094565 The article is quite long, there is no need at all to read it entirely. The only part related to my model is starting at the middle of page 40 and ending page 42. It is the "Model of local recurrent neural networks in frontal areas" (bold tittle), not the "Cellular model of pyramidal neurons in frontal areas" that starts page 38. Thus, to answer your questions: Le 2021-11-28 21:56, Charl Linssen a écrit : Hi, Thanks for writing in. Just some questions to make sure I understand it right. Does the channel opening probability in a given synapse only depend on the history of presynaptic spikes? In that case, it would be computationally most efficient to compute these probabilities in the (presynaptic) neuron objects, rather than in the synapses, because each synapse downstream from that neuron would compute the same probabilities anyway. It is possible to use the postsynaptic membrane potential to modulate synaptic plasticity, but this ignores the synaptic delay associated with the connection. For an example, please see: https://nestml.readthedocs.io/en/latest/tutorials/active_dendrite/nestml_ac… It might help if you have a link to the paper, or a full description of the model you would like to implement. With kind regards, Charl On Thu, Nov 25, 2021, at 15:24, barthelemy wrote: Dear all, I am a new NEST user. I have a question concerning the range of neuron/synapses model possibilities of NEST. I would like to implement my own neuron/synapse model with NESTML, but I am unsure that it would be possible. Indeed, in my model, synaptic currents are not only relying on pre-synaptic spikes. To compute synaptic currents, the opening probability of pre-synaptic channel receptors are required. Those pre-synaptic channel receptors opening probabilities are evolving according to differential equations involving second order dynamics, with specific decays and taking into account the pre-synaptic spikes arrivals times at this specific synapse. Those differential equations for the opening probabilities are relying on different parameters, according to the neurotransmitter type (GABA A,GABA B, NMDA, AMPA ). Furthermore, additionally to the input spikes and the pre-synaptic channel receptors opening probabilities, the current membrane potential of the post-synaptic neuron is also required to compute the synaptic currents. Do you know if one of the NEST models implement similar dynamics? Is it possible to compute such synaptic dynamics with NESTML by creating a synapse or (and) a neuron model? Or is it not, due to specific limitations? Thank you, Best regards, JB _______________________________________________ NEST Users mailing list -- users(a)nest-simulator.org To unsubscribe send an email to users-leave(a)nest-simulator.org _______________________________________________ NEST Users mailing list -- users(a)nest-simulator.org To unsubscribe send an email to users-leave(a)nest-simulator.org

Dear Jeanne, Sorry for the late reply, we had to discuss this a bit internally. In general, it looks like this model should be possible, but the heterogeneous transmission delays could lead to issues when the length of the delay exceeds the length of the postsynaptic (dendritic) delay. This is a consequence of the magic under the hood of NEST. We actually just had a Master's student join our team to work exactly on making your use case possible! But it might still be a few months until we have some code to share. We'll be sure to keep you posted on that, but in any case, if the delays you are planning to use are shorter than any of the dendritic delays in the network (what we call the "min_delay" period), it should already work without any major changes. With kind regards, Charl On Mon, Nov 29, 2021, at 11:47, barthelemy wrote: Sorry, I made a mistake and sent the message before I could finish writing it. Below is the end of it: *Thus, to answer your question: The opening probabilities are indeed following first or second order dynamics, but the only external input in the differential equations are the pre-synaptic spikes. Though, parameters used to compute the opening probabilities are not the same for every neurotransmitter. Furthermore, I am actually going to change the model in the paper a little bit. I am planning to improve the model by adding transmission delays in the differential equation of the opening probabilities. Thus, unlike the research paper, in the formulas of the differential equations, I would not have anymore just the kronecker delta of (t-t_spike), but of (t-t_spike-t_trans), with t the current biological time, t_spike the spiking biological time of the pre synaptic neuron, and t_trans the transmission delay from the pre-synaptic neuron to the synapse, which will not be the same for every synapse, even for spikes from the same pre-synaptic neuron. Thus, because of those transmission delays that I am planning to add to the model in the research paper, it might be difficult to compute the opening probabilities in the pre-synaptic neuron...And I am not sure if this specific configuration is possible in NEST. As for the postsynaptic membrane potential, thank you very much for sharing the link! Though, as I will be taking into account transmission delays, it might not be adapted. I will check this. Finally, this is the link of the research paper with the model I would like to implement & improve with transmission delays : https://hal.archives-ouvertes.fr/hal-03094565/document) Do not hesitate to ask more questions if anything is unclear! Thank you, Best regards, Jeanne Barthélemy Le 2021-11-29 10:59, barthelemy a écrit : Hello, Thank you very much for your answer. The model I am trying to implement in NEST is explained in details in the article in attachment of this email. In case it is more convenient for you, here is also the reference : Vincent Fontanier, Matthieu Sarazin, Frederic Stoll, Bruno Delord, Emmanuel Procyk. Local in- hibitory control of frontal network metastability underlies the temporal signature of cognitive states. 2021. hal-03094565 The article is quite long, there is no need at all to read it entirely. The only part related to my model is starting at the middle of page 40 and ending page 42. It is the "Model of local recurrent neural networks in frontal areas" (bold tittle), not the "Cellular model of pyramidal neurons in frontal areas" that starts page 38. Thus, to answer your questions: Le 2021-11-28 21:56, Charl Linssen a écrit : Hi, Thanks for writing in. Just some questions to make sure I understand it right. Does the channel opening probability in a given synapse only depend on the history of presynaptic spikes? In that case, it would be computationally most efficient to compute these probabilities in the (presynaptic) neuron objects, rather than in the synapses, because each synapse downstream from that neuron would compute the same probabilities anyway. It is possible to use the postsynaptic membrane potential to modulate synaptic plasticity, but this ignores the synaptic delay associated with the connection. For an example, please see: https://nestml.readthedocs.io/en/latest/tutorials/active_dendrite/nestml_ac… It might help if you have a link to the paper, or a full description of the model you would like to implement. With kind regards, Charl On Thu, Nov 25, 2021, at 15:24, barthelemy wrote: Dear all, I am a new NEST user. I have a question concerning the range of neuron/synapses model possibilities of NEST. I would like to implement my own neuron/synapse model with NESTML, but I am unsure that it would be possible. Indeed, in my model, synaptic currents are not only relying on pre-synaptic spikes. To compute synaptic currents, the opening probability of pre-synaptic channel receptors are required. Those pre-synaptic channel receptors opening probabilities are evolving according to differential equations involving second order dynamics, with specific decays and taking into account the pre-synaptic spikes arrivals times at this specific synapse. Those differential equations for the opening probabilities are relying on different parameters, according to the neurotransmitter type (GABA A,GABA B, NMDA, AMPA ). Furthermore, additionally to the input spikes and the pre-synaptic channel receptors opening probabilities, the current membrane potential of the post-synaptic neuron is also required to compute the synaptic currents. Do you know if one of the NEST models implement similar dynamics? Is it possible to compute such synaptic dynamics with NESTML by creating a synapse or (and) a neuron model? Or is it not, due to specific limitations? Thank you, Best regards, JB _______________________________________________ NEST Users mailing list -- users(a)nest-simulator.org To unsubscribe send an email to users-leave(a)nest-simulator.org _______________________________________________ NEST Users mailing list -- users(a)nest-simulator.org To unsubscribe send an email to users-leave(a)nest-simulator.org