Skip to main content
. 2020 Aug 19;40(34):6584–6599. doi: 10.1523/JNEUROSCI.0649-20.2020

Table 1.

Parameters used in model simulation

Parameters Letter Value
Retinal wave
    Retinal wave evolution time step Δt 100 ms
    ON RGC dendritic radius RON 400 μm
    AC dendritic radius RAC 40 μm
    ON RGC bursting threshold ΘON 14 unit
    AII AC activation threshold ΘAC 0.5 unit
    OFF RGC inhibition threshold ΘOFF −0.2 unit
    Wave filtering width σwave 0.85dOFF
RGC-V1 learning
    Retina-V1 size ratio γ 1.98 (monkey) (Jang et al.,2020)
0.75 (cat) (Jang et al., 2020)
0.20 (mouse) (Jang et al.,2020)
    Spatial decay parameter of RGC to V1connection dFF 18 μm (cat mosaics)
24 μm (monkey mosaics)
18 μm (mouse mosaics)
    Initial weight of RGC to V1 connection WinitFF 0.05
    Nonlinearity of V1 sigmoidal response curve ΘV1 0.5
    Nonlinearity of V1 sigmoidal response curve δV1 0.15
    Time constant of firing rate average τFF 15 learning steps
    Learning rate in Hebbian learning ϵFF 0.005
    Learning epochs --- 15 epochs
    Resource limit of single connection WlimitFF 0.14
    Image filter size σimg 36 μm (cat mosaics)
56 μm (monkey mosaics)
V1 horizontal connection network learning
    Initial weight of V1 horizontal connection WinitV1 0.01
    Time constant of firing rate average τV1 10 learning steps
    Learning rate in Hebbian learning ϵV1 2 × 10−7 (cat/monkeymodels)
2 × 10−5 (mouse model)
    Learning epochs --- 30 epochs (cat/monkeymodels)
10 epochs (mouse model)
    Resource limit of single connection WlimitV1 5 × 10−4
V1 spontaneous activity
    Normalization weight of V1 horizontalconnection WfinalV1 3
    Amplitude of local random stimulus Ilocal 10
Spatial scale of local random stimulus σlocal 20 μm
Amplitude of background noise stimulus Ibackground 0.01
Spatial scale of background noise stimulus σbackground 30 μm