Table 1.

Symbols with descriptions.

Symbol	Description
y	Binary row vector of length Nin. Represents the firing state of input neurons. Codifies a stimulus
z	Binary row vector of length Nrec. Represents the firing of the recurrent network
u	Real-valued row vector of length Nrec. Represents the activation states of neurons in the recurrent network
Win	Matrix of synaptic weights from sensory neurons to the recurrent network. Columns are incoming connections
Wrec	Matrix of synaptic weights among neurons of the recurrent network. Columns are incoming connections
Nin	Number of input neurons ($N_{\mathit{in}}=2$ throughout this work)
Nrec	Number of neurons in the recurrent network
N	Total number of neurons (input and integration)
M	Total number of network firing states
$\mathbf{\theta }$	Row vector of neuron’s thresholds
c	Row vector, obtained after concatenating one y vector with one z vector
C	Matrix whose rows are c vectors. Coefficient matrix in a system of linear equations
U	Matrix composed of row vectors u. Contains activation states reached by the network from the firing states in matric C
W	Matrix resulting from concatenating matrices Win and Wrec
Ubase	Matrix of row vectors u picked at random
Ulc	Matrix composed of the rows in Ubase and linear combinations thereof. There is one row for each network state
Z	Binary matrix, obtained by applying threshold $\mathbf{\theta }$ to matrix Ulc
$\mathbf{\Delta }$	Real-valued vector of length Nrec. Each component is the difference between activations after s1 and s2 presentation, when starting from the same network firing state
fr	Redundancy factor: quotient between the number of neurons and the number of sequences codified in an s-task
fcc	Multiplying factor to induce signal correlation
fbc	Number of network states reachable after s1 or s2 presentation, divided by the total number of network states