Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

Supporting Information Edin et al. 10.1073/pnas.0901894106

Summary: Supporting Information
Edin et al. 10.1073/pnas.0901894106
Fig. S1. Graphical representation of H(IX) in the equation for pcap (Eq. 2) with f(I) (1 e I) 1
Edin et al. www.pnas.org/cgi/content/short/0901894106 1 of 8
Fig. S2. Effect of changes in G , G , or IX on memory capacity and persistent activity. Changes in parameters that lead to improved capacity generally increase
network activity. (Top) Firing rate of memories at different loads. (Middle) Graphical solutions of the capacity equation for two different loads, p 1 and 3. As
p increases, the line representing the effective network synaptic strength shifts leftward. (Bottom) Potential energy landscape of the network at loads 1 and 3.
Energy minima represent stable states. (A) Reference parameter set used in Fig. 2: A 0.1, G 22, G 2, IX 6.25. When load equals capacity (thin line;
here, pcap 3.57) the synaptic line is tangential to the neuronal-input output curve in the Middle. (Bottom) The energy minimum corresponding to the memory
state disappears at capacity (thin line). (B) Increasing capacity from 3 to 4 by increasing G leads to increased persistent activity at the same load (arrows indicate
memory activity at load p 3). For comparison, the reference network is shown in gray. (C) Same for G . (D) Same for IX. Generally, networks that have higher
capacity caused by a change in a G , G , or IX also have higher activity at the same load.
Edin et al. www.pnas.org/cgi/content/short/0901894106 2 of 8
Fig. S3. Working memory capacity in a balanced network. (A) The network is stable when the line and the curve intersect in three places. For the value of IX
in the figure, the golden area represents the region of stable memory activity. The size of this area is determined by the ratio between G and H(IX), which
determines the right and left border of the area. (B) G /H(IX) is in turn determined by the spontaneous activity. For spontaneous rates above 1.5, G /H(IX)
1.5. (C) Having determined G /H(IX), Eqs. 3 and S6 (see SI Appendix) can now be used to calculate pcap
for a given level of spontaneous activity. Note that although


Source: Andrzejak, Ralph Gregor - Departament de Tecnologia, Universitat Pompeu Fabra


Collections: Computer Technologies and Information Sciences