Physiology configuration

The physiological configuration consists of

The physiology configuration consists of constants (variable-values) and dictionaries (variable-key-values).

There are five hard-coded neuron types in CxSystem2 (two excitatory, SS and PC; three inhibitory, BC, MC and L1i), but the user can easily define subtypes, e.g. L4_MC. Subtypes can have arbitrary names (e.g. MyFavouriteBasketCellType, L5_LBC). Each ‘model’ refers to Brian equations, which are explained in the Neurodynlib section. Advanced users are able to add new neuron types, including new membrane equation models. See the Developer’s Guide.

General model parameters

General model parameters affect all neurons and synapses. They include the point neuron model, compartmental PC neuron model, receptor models, receptor weights and background noise rates. Note that the model names must be surrounded by single-quotes.

These parameters are included in the example configuration files, including some short explanations in comments.

The parameters are:


model_variation{0,1}: This should be 1. Value 0 is only for backwards compatibility, i.e. old models that use equations hard-coded in CxSystem1.

neuron_model{'string'}: Specifies which neuron model to use for point neurons (other cells than PCs). Available neuron models.

excitation_model{'string'}: Specifies the model for excitatory receptors in point neurons. Available receptor models.

inhibition_model{'string'}: Specifies the model for inhibitory receptors in point neurons. Available receptor models.

pc_excitation_model{'string'}: Specifies the model for excitatory receptors in pyramidal cells. Available receptor models.

pc_inhibition_model{'string'}: Specifies the model for inhibitory receptors in pyramidal cells. Available receptor models.

background_rate{float*Hz}: Sets the rate for excitatory background synapses (Poisson-distributed). Receptors will be modeled as excitation_model.

background_rate_inhibition{float*Hz}: Sets the rate for inhibitory background synapses (Poisson-distributed). Receptors will be modeled as inhibition_model.

background_E_E_weight{float*unit}: Weight of excitatory-to-excitatory synapses.

background_E_I_weight{float*unit}: Weight of excitatory-to-excitatory synapses.

background_I_E_weight{float*unit}: Weight of inhibitory-to-excitatory synapses.

background_I_I_weight{float*unit}: Weight of inhibitory-to-inhibitory synapses.

Neuron type-specific parameters

Neuron type-specific parameters are given as dictionaries. Subtype-specific parameters (e.g. L23_MC, L4_MC) will override parameters for the hard-coded types (MC). Parameter names must match those defined in neurodynlib. For example, to define parameters for BC neurons that are modelled as exponential integrate-and-fire (EIF) neurons, you would write:

BC C 100 * pF
  gL 10 * nS
  EL -70 * mV
  VT -40 * mV
  DeltaT 2 * mV
  Ee 0 * mV
  Ei -75 * mV
  tau_e 3 * ms
  tau_i 8 * ms
  V_res VT - 4 * mV
  Vcut VT + 5*DeltaT
  V_init_min EL
  V_init_max VT

This would define a general basket cell type. You could then continue by defining parameters for L23_NBC (L2/3 nest basket cell) and setting the neuron_subtype as L23_NBC for the corresponding group in the Anatomy configuration.

As shown in the example, you can refer to parameters (and do computations using them!) defined earlier in the Physiology configuration.

Connection- and synapse-type specific parameters

Connection weights and connections delays are also given as dictionaries (cw and delay, respectively). These are given with reference to the hard-coded cell types. Pathway-specific connection weights can be set in the anatomy configuration using the custom_weight parameter. If there is no custom_weight defined, then values in the cw dictionary will be used. Currently there is no way to define pathway- or neuron subtype-specific delays.

These dictionaries are included in the example configuration files. Even though you might not have all the hard-coded cell types in your model, you don’t need to delete the redundant lines.

Other simulation parameters

There are some additional parameters that are sometimes required. The most important are parameters related to short-term plasticity (STP) and to connection weight scaling by extracellular calcium concentration. To see how these are implemented, please see Methods in Hokkanen et al. 2019 Neural Computation.

For depressing synapses, you need the following parameters:


U_E{float}: Utilization factor for depressing excitatory synapses.

U_I{float}: Utilization factor for depressing inhibitory synapses.

tau_d{float*unit}: Recovery time constant (from depression).

For facilitating synapses:


U_f{float}: Utilization increment for facilitating synapses.

tau_f{float*unit}: Facilitation decay time constant.

tau_fd{float*unit}: Recovery time constant for facilitating synapses.

If you want to scale synapse weights with respect to extracellular calcium level, you should define the following parameters:


calcium_concentration{float}: Calcium concentration in mM. If set to 2.0, there is no scaling.

flag_background_calcium_scaling{0,1}: Sets whether background inputs are also scaled with respect to calcium level.