Table 1 A list of model parameters used in the validation.

	Basic	Supplementary Figs. S2A–C, S5A, B	Supplementary Fig. S2E	Supplementary Fig. S5C	Supplementary Fig. S5D	Figure 6B	Figure 6D
m	0	0	0	0	0	0–1 in 0.1 increments	0
d	0	0	0	0	0	0	0–250 in 25 increments
C	2200	2200	2200	2200	100–5000 in 100 increments	2200	2200
α	100	100	100	100	100	100	100
β	3	3	3	3	3	3	3
γ	0.5	0.5	0.5	0.5	0.5	0.5	0.5
δ	0.5	0.5	0.5	0.5	0.5	0.5	0.5
I^RC1	26	26	26	26	26	26	26
I^RC2	30	30	30	30	30	30	30
O	3	3	3	3	3	3	3
N	1200	600–1200 in 200 increments	1200	1200	1200	1200	1200
S	300	100–600 in 100 increments	300	300	300	300	300
p	0.5	0.5	0.5	0–1 in 0.1 increments	0.5	0.5	0.5
τ	5	5–35 in 10 increments	5	5	5	5	5
g	1.0	1.0	0–2 in 0.1 increments	1.0	1.0	1.0	1.0
k	0.2	0.2	0 or 0.2	0.2	0.2	0.2	0.2

Parameters used in simulations are as follows: m, d, and C are parameters related to the input signal, where m is the non-coincidence rate of the input stimulus, d is the delay time, and C is the total number of stimuli, including random sequences. α, β, γ, and δ are parameters involved in the learning of the network. α plays a role in the learning rate in FORCE learning. β, γ, and δ are constants. For more information on α, see Sussillo⁴⁰. I^RC1, I^RC2, O, N, and S are the parameters that make up an RC. I^RC1 and I^RC2 are the numbers of input nodes, O is the number of read nodes, N is the number of reservoir nodes, and S is the number of reservoir nodes coupled to the read nodes of RC1 and RC2. p, τ, g, and k are the parameters that control the dynamics of the reservoir neurons. p is the connection probability of reservoir neurons, τ is the time constant that determines the dynamics of the reservoir neurons, g is the parameter that controls the dynamics, and k is the rate at which noise is introduced.

Quick links

Search