Synapses not functioning

Support
1

Description of problem

I have simplified my network to 2 neurons, after realising that the conductances of post-synaptic neurons weren’t increasing following spikes of pre-synaptic neurons. The rpoblem persists.
When I plot g_exc for example it is a straight line.

Minimal code to reproduce problem

def define_network(parameters):
    # Define neuron model
    eqs = """
    dv/dt = (g_l*(e_l-v) + g_l*slope_f*exp((v-v_t)/slope_f) - g_exc*(v-E_exc) - g_inh*(v-E_inh) + i_stim - w)/c_m : volt
    dw/dt = (a*(v - e_l) - w)/tau_w : amp
    dg_exc/dt = -g_exc/tau_exc : siemens
    dg_inh/dt = -g_inh/tau_inh : siemens
    x : 1 (constant)
    y : 1 (constant)
    """
    ain_params = parameters['general'].copy()
    ain_params.update(parameters['ains'])
    mn_params = parameters['general'].copy()
    mn_params.update(parameters['mns'])

    N_A = parameters['ains']['N']

    N_M = parameters['mns']['N']

    ains_left = NeuronGroup(N_A, model=eqs, method='euler', threshold='v > 0*mV', reset='v = vr\nw = w + b', namespace=ain_params)
    mns_left = NeuronGroup(N_M, model=eqs, method='euler', threshold='v > 0*mV', reset='v = vr\nw = w + b', namespace=mn_params)

    ains_left.v = parameters['ains']['e_l']

    mns_left.v = parameters['mns']['e_l']
    
    ains_left.w = 0
    mns_left.w = 0
    
    ains_left.x = 1  # X position
    ains_left.y = 'i'  # Y position

    mns_left.x = 3  # X position
    mns_left.y = 'i'  # Y position


    synaptic_model = '''
                      w_syn : siemens
                      d : 1
                      '''

    S_mn_ain_l = Synapses(mns_left, ains_left, model=synaptic_model, on_pre='g_exc+=w_syn')   
    S_mn_ain_l.connect()
    S_mn_ain_l.w_syn = parameters['mns']['w_mn_ain']

    S_ain_mn_l = Synapses(ains_left, mns_left, model=synaptic_model, on_pre='g_inh+=w_syn')  
    S_ain_mn_l.connect()
    S_ain_mn_l.w_syn = parameters['ains']['w_ain_mn']

    return ains_left, mns_left

def run_simulation(parameters, simulation_time):

    ains_left, mns_left = define_network(parameters)

    N_A = parameters['ains']['N']
    N_M = parameters['mns']['N']

    spikes_ain_l = SpikeMonitor(ains_left)
    
    spikes_mn_l = SpikeMonitor(mns_left)
    
    mon_ain = StateMonitor(ains_left, 'g_exc', record=0)  # Record the membrane potential of the first aIN

    mon_mn = StateMonitor(mns_left, 'g_exc', record=0)    # Record the membrane potential of the first MN

    defaultclock.dt = 0.1 * ms
    time_secs = simulation_time
    run(time_secs)
    
    return mon_ain, mon_mn

(parameters is a dictionary of parameters, I set w_syn for the connections to large values (1000nsiemens) but nothing happened).

What you have already tried

Simplified the network and plotted the conductance. Discovered that the problem was that the conductance wasn’t changing.

2

Hi @awilson98. The problem is that the synapses objects are not returned from your define_network function. In run_simulation, the run function looks for “Brian objects” that should be included in the network (an alternative would be to create a Network object manually) ­– it sees ains_left, mns_left and the monitors, but no synapses, so they do not get included in the simulation. Additionally, given that nothing references thes synapses anymore, Python is free to delete the Synapses objects. This is probably a situation where we should raise a warning, we’ve recently discussed similar cases in this GitHub issue: Returned value needs to be retrieved for object to work · Issue #1499 · brian-team/brian2 · GitHub

Hope that helps!
Marcel

PS: For nicer display of code blocks, you can use the following markdown syntax (I’ve edited your post accordingly):

```
# Here goes the Python code
print("Hello")
```
3

Thank you, that solves it!

1 Like