Abstract
We study the memory performance of a class of modular attractor neural networks, where modules are potentially fully-connected networks connected to each other via diluted long-range connections. On this anatomical architecture we store memory patterns of activity using a Willshaw-type learning rule. P patterns are split in categories, such that patterns of the same category activate the same set of modules. We first compute the maximal storage capacity of these networks. We then investigate their error-correction properties through an exhaustive exploration of parameter space, and identify regions where the networks behave as an associative memory device. The crucial parameters that control the retrieval abilities of the network are (1) the ratio between the number of synaptic contacts of long- and short-range origins (2) the number of categories in which a module is activated and (3) the amount of local inhibition. We discuss the relationship between our model and networks of cortical patches that have been observed in different cortical areas.
Similar content being viewed by others
References
Amit, D.J. (1989). Modeling brain function: Cambridge University Press.
Amit, D.J., & Fusi, S. (1994). Dynamic learning in neural networks with material sysnapses. Neural Computation, 6, 957–982.
Bosking, W., Zhang, Y., Schofield, B., & Fitzpatrick, D. (1997). Orientation selectivity and the arrangement of horizontal connections in tree shrew striate cortex. J Neurosci., 17, 2112–2127.
Braitenberg, V., & Schutz̈, A. (1991). Anatomy of the cortex: Springer-Verlag.
Brunel, N. (2003). Network models of memory. In Methods and Models in Neurophysics, Volume Session LXXX: Lecture Notes of the Les Houches Summer School, pages 407–476.
Brunel, N. (2005). Network models of memory. In Chow, C., Gutkin, B., Hansel, D., Meunier, C., & Dalibard, J. (Eds.) Methods and Models in Neurophysics, Volume Session LXXX: Lecture Notes of the Les Houches Summer School. Elsevier.
Buzas, P., Eysel, U.T., Adorjan, P., & Kisvarday, Z.F. (2001). Axonal topography of cortical basket cells in relation to orientation, direction, and ocular dominance maps. J Comp Neurol, 437, 259–285.
DeFelipe, J., Conley, M., & Jones, E.G. (1986). Long-range focal collateralization of axons arising from corticocortical cells in monkey sensory-motor cortex. J Neurosci., 6, 3749–3766.
Dubreuil, A.M. (2014). Memory and cortical connectivity. Université Paris Descartes: PhD thesis.
Dubreuil, A.M., Amit, Y., & Brunel, N. (2014). Memory capacity of networks with stochastic binary synapses. PLoS computational biology, e1003727(8).
Enoki, R., Hu, Y.L., Hamilton, D., & Fine, A. (2009). Expression of long-term plasticity at individual synapses in hippocampus is graded, bidirectional, and mainly presynaptic: optical quantal analysis. Neuron, 62(2), 242–253.
Funahashi, S., Bruce, C.J., & Goldman-Rakic, P.S. (1989). Mnemonic coding of visual space in the monkey’s dorsolateral prefrontal cortex. J. Neurophysiol., 61, 331–349.
Fuster, J.M. (1995). Memory in the cerebral cortex: MIT Press.
Fuster, J.M., & Alexander, G. (1971). Neuron activity related to short-term memory. Science, 173, 652–654.
Gilbert, C.D., & Wiesel, T. (1989). Columnar specificity of intrinsic horizontal and corticocortical connections in cat visual cortex. J Neurosci., 9, 2432–2442.
Hellwig, B. (2000). A quantitative analysis of the local connectivity between pyramidal neurons in layers 2/3 of the rat visual cortex. Biological cybernetics, 82(2), 111–121.
Holmgren, C., Harkany, T., Svennenfors, B., & Zilberter, Y. (2003). Pyramidal cell communication within local networks in layer 2/3 of rat neocortex. J. Physiol., 551, 139–153.
Hopfield, J.J. (1982). Neural networks and physical systems with emergent collective computational abilities. Proc. Natl. Acad. Sci. U.S.A., 79, 2554–2558.
Huth, A.G., Nishimoto, S., Vu, A.T., & Gallant, J.L. (2012). A continuous semantic space describes the representation of thousands of object and action categories across the human brain. Neuron, 76(6), 1210–1224.
Johansson, C., & Lansner, A. (2007). Imposing biological constraints onto an abstract neocortical attractor network model. Neural Comput., 19(7), 1871–1896.
Kalisman, N., Silberberg, G., & Markram, H. (2005). The neocortical microcircuit as a tabula rasa. Proc Natl Acad Sci U S A, 102(3), 880–885.
Knoblauch, A., Palm, G., & Sommer, F.T. (2010). Memory capacities for synaptic and structural plasticity. Neural Computation, 22(2), 289–341.
Kropff, E., & Treves, A. (2005). The storage capacity of Potts models for semantic memory retrieval. J. Stat. Mech., 8, P08010.
Loewenstein, Y., Kuras, A., & Rumpel, S. (2011). Multiplicative dynamics underlie the emergence of the log-normal distribution of spine sizes in the neocortex in vivo. J. Neurosci., 31(26), 9481–9488.
Mari, C.F. (2004). Extremely dilute modular neuronal networks: Neocortical memory retrieval dynamics. Journal of Computational Neuroscience, 17, 57–79.
Mari, C.F., & Treves, A. (1998). Modeling neocortical areas with a modular neural network. Biosystems, 48 (1), 47–55.
Markram, H., Lubke, J., Frotscher, M., & Sakmann, B. (1997). Regulation of synaptic efficacy by coincidence of postsynaptic APs and EPSPs. Science, 275, 213–215.
Meli, C., & Lansner, A. (2013). A modular attractor associative memory with patchy connectivity and weight pruning. Network, 24, 129–150.
Miller, E.K., Erickson, C.A., & Desimone, R. (1996). Neural mechanisms of visual working memory in prefrontal cortex of the macaque. J. Neurosci., 16, 5154–5167.
Miyashita, Y. (1988). Neuronal correlate of visual associative long-term memory in the primate temporal cortex. Nature, 335, 817–820.
Montgomery, J.M., & Madison, D.V. (2004). Discrete synaptic states define a major mechanism of synapse plasticity. Trends Neurosci., 27(12), 744–750.
Nadal, J.-P. (1991). Associative memory: on the (puzzling) sparse coding limit. J. Phys. A: Math. Gen., 24, 1093–1101.
O’Connor, D.H., Wittenberg, G.M., & Wang, S. S.-H. (2005). Graded bidirectional synaptic plasticity is composed of switch-like unitary events. Proc Natl Acad Sci U S A, 102, 9679– 9684.
O’Kane, D. & Treves, A. (1992). Short-and long-range connections in autoassociative memory. Journal of Physics A: Mathematical and General, 25, 5055.
Perin, R., Berger, T.K., & Markram, H. (2011). A synaptic organizing principle for cortical neuronal groups. Proc. Natl. Acad. Sci. U.S.A., 108, 5419–5424.
Petersen, C.C., Malenka, R.C., Nicoll, R.A., & Hopfield, J.J. (1998). All-or-none potentiation at CA3-CA1 synapses. Proc.Natl.Acad.Sci.USA, 95, 4732–4737.
Pucak, M.L., Levitt, J.B., Lund, J.S., & Lewis, D.A. (1996). Patterns of intrinsic and associational circuitry in monkey prefrontal cortex. J. Comp. Neurol., 338, 360–376.
Romo, R., Brody, C.D., Hernández, A., & Lemus, L. (1999). Neuronal correlates of parametric working memory in the prefrontal cortex. Nature, 399, 470–474.
Roudi, Y., & Treves, A. (2004). An associative network with spatially organized connectivity. Journal of Statistical Mechanics: Theory and Experiment, 2004(07), P07010.
Roudi, Y., & Treves, A. (2006). Localized activity profiles and storage capacity of rate-based autoassociative networks. Physical Review E, 73(6), 061904.
Sjöström, P.J., Turrigiano, G.G., & Nelson, S. (2001). Rate, timing, and cooperativity jointly determine cortical synaptic plasticity. Neuron, 32, 1149–1164.
Stepanyants, A., Martinez, L.M., Ferecsko, A.S., & Kisvarday, Z.F. (2009). The fractions of short-and long-range connections in the visual cortex. Proceedings of the National Academy of Sciences, 106(9), 3555–3560.
Tsao, D.Y., Freiwald, W.A., Knutsen, T.A., Mandeville, J.B., & Tootell, R.B.H. (2003). Faces and objects in macaque cerebral cortex. Nature Neuroscience, 6(9), 989–995.
van Vreeswijk, C., & Sompolinsky, H. (2003). Irregular activity in large networks of neurons. In Methods and Models in Neurophysics, Volume Session LXXX: Lecture Notes of the Les Houches Summer School, pages 341–402.
Wang, X.-J. (2001). Synaptic reverberation underlying mnemonic persistent activity. Trends Neurosci., 24, 455–463.
Willshaw, D., Buneman, O.P., & Longuet-Higgins, H. (1969). Non-holographic associative memory. Nature, 222, 960–962.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Additional information
Action Editor: Alessandro Treves
Appendix
Appendix
Estimating the storage capacity requires to compute the distributions of the inputs on foreground and background neurons. Because neural activity and synapses are binary, these inputs are sums of binary random variables. We first present general results about such sums, that will be used later to compute the distributions of inputs to neurons.
We consider a random variable h
where the X k are independent binary random variables described by a parameter q:
The sum h is then distributed according to a binomial distribution
Note that to get this binomial distribution, we have to assume the X k ’s are independent. In our case, this means that two synapses on the same neuron \(W_{ij_{1}}^{m,n}\) and \(W_{ij_{2}}^{m,n}\) are treated as independent variables. This is a reasonable assumption to make as we have
and
and similarly for long-range connections \(\mathbb {P} (W_{ij_{1}}^{m,n} = 1\vert W_{ij_{2}}^{m,n} = 1) \underset {f\rightarrow 0,pc\rightarrow \infty }{\simeq } \mathbb {P} (W_{ij_{1}}^{m,n} = 1)\).
We will consider cases in which K and S are large. We can then use Stirling formula to express the binomial coefficients and write
with
We use the superscript fc as this expression will be mainly used to describe fully connected sub-networks. For diluted enough networks, we will have \(q, \frac {S}{K} \ll 1\), it is then useful to introduce
In our networks, when testing the stability of a given pattern \(\boldsymbol {\Xi }^{\mu _{0}}\) it is useful to separate the total input into a local part and an external part. The local part is described by a couple (K, q), where K = f N is the number of neurons active in a given local network, and q = 1 or g depending on whether we are considering the input onto a foreground or a background neuron. The external part can also be described by a couple (K, q) with K = F(M−1)f N or K = F(M−1)(1−f)N and \(q=\frac {D}{N}\) or \(\frac {D}{N}G\) for foreground or background neurons.
The distribution of the total input on a neuron can be written
To compute it, we first need to express the distribution of the inputs generated by the local module and the distribution of the inputs generated by the other modules. In the asymptotic limits we consider, this sum will be dominated by the most probable term of the sum, we will thus need to find the couple (S l , S e ) that maximizes \(\mathbb {P}_{l}(S_{l})\mathbb {P}_{e}(S_{e})\).
1.1 Distribution of inputs and probability of no-error in multi-modular network
1.1.1 Case F M → + ∞
We apply the method sketched above, first to compute the distribution of inputs on foreground neurons, and then on background neurons.
Foreground neurons The distribution of local inputs is a delta function \(\mathbb {P}_{l}(S_{l}) = \delta (S_{l} - fN)\) as exactly f N neurons are active in each module in each pattern, and because of the fact that each module is a fully-connected network. The external component is the sum of the activity in each of the other F M−1≃F M active modules when their states coincides with the pattern \(\boldsymbol {\Xi }^{\mu _{0}}\) we are trying to retrieve. Given the above results on sum of binary variables, it writes
with \(s_{e} = \frac {S_{e}}{fN}\).
The total input on foreground neurons is then
Background neurons in active modules The local input now fluctuates because inputs are mediated by synapses that have been potentiated during the presentation of randomly drawn patterns \(\boldsymbol {\Xi }^{\mu \neq \mu _{0}}\). It is distributed according to
where g is defined in Eq. (25). Similarly, the external part of the input is distributed according to
The distribution of the total input is now written
where \(s_{l}^{*} = \frac {S_{l}^{*}}{fN}\) and \(s_{e}^{*} = \frac {S_{e}^{*}}{fN} = s - s_{l}^{*}\) (where \(s=\frac {S}{fN}\)) are given by the condition
Solving this equation yields
Background neurons in silent modules There is only long-range inputs in this case and the fraction of activated long-range synapses G ′≃α c F is given by (28),
Probability of no errors We have derived the expressions for the distribution of inputs to both foreground and background neurons. In order to compute the probability that there is no error ¶ n e in the retrieval of pattern \(\boldsymbol {\Xi }^{\mu _{0}}\), we have to estimate the probability that the inputs are above or below threshold, as written in the main text in (14). To do so we first note that
where the ‘\(\sum \)’ term will not contribute to the final expression of \(\mathbb {P}_{ne}\) in the large N limit, as has been shown in (Dubreuil et al. 2014). In practice we thus replace the probability to be above threshold by the probability to be at threshold. We can apply the same reasoning for the probability to be above the activation threshold for background neurons. We now have all the elements to express Φf, Φb and \({\Phi }^{b^{\prime }}\) in formulas (14):
and
1.1.2 Case F M = O(1)
In this case the microscopic dilution term \(\frac {D}{N}\) is finite and we have to use Φfc instead of Φdc to describe the external inputs. Following the same reasoning as above, the rate functions are
and
with \(s_{l}^{*}\) and \(s_{e}^{*}\) given by
with
The expression for \({\Phi }^{b^{\prime }}\) remains unchanged as connectivity between two randomly taken modules is effectively highly diluted.
1.2 Capacity calculation for a single module with diluted connectivity
Here we focus on a model made of a single module with N neurons whose dynamics obey Eq. (1), where we store P patterns ξ μ with coding level f (\({\sum }_{i} \xi _{i}^{\mu } = fN\)). As for modular networks we focus on the limits N → + ∞, \(f = \beta \frac {\ln N}{N}\) and \(P=\frac {\alpha }{f^{2}}\) with α, β = O(1). Patterns are stored on a diluted connectivity matrix, such that at the end of the learning phase the synaptic matrix is given by W i j = w i j d i j . With w i j = 1 if there exists a pattern such that neurons i and j are co-activated, w i j = 0 otherwise ; and d i j is drawn randomly in {0,1} being 1 with probability d≪1.
To compute the capacity we follow the procedure described in the ‘Methods’ section. We first set the network in a state corresponding to one of the patterns \(\boldsymbol {\Xi }^{\mu _{0}}\), and ask whether this a fixed point of Eq. (1). This is done by computing \(\mathbb {P}_{ne}\), the probability that all the fields are on the right side of the activation threshold. Similarly to Eq. (13),
Using the Eqs. (52), (54) that describe the distributions of inputs and the fact that \(\mathbb {P}(h_{i}\leq fN\theta \, \vert \, \xi _{i}^{\mu _{0}}=1)\simeq \mathbb {P}(h_{i}= fN\theta \, \vert \, \xi _{i}^{\mu _{0}}=1)\) (see (Dubreuil et al. 2014)) we can write
and
where q is the fraction of synapses on i, w i j , that are 1 after the learning phase, which can be expressed as
For \(\mathbb {P}_{ne}\) to go 1 in the large N limit, Equations (17), (18) have to be satisfied. Saturating these inequalities leads to a choice of activation threshold θ → d and a coding level with
In the specific case we are studying the general expression for the storage capacity (20) can be written
Using the two Eqs. (75)–(76), it can be expressed more simply
A maximal storage capacity I = 0.26 is reached at q = 0.24.
Note that for modular networks we can not get such a closed form for I since the \(\mathbb {P}(h_{i}=fN\theta )\) needs to be estimated numerically.
Rights and permissions
About this article
Cite this article
Dubreuil, A.M., Brunel, N. Storing structured sparse memories in a multi-modular cortical network model. J Comput Neurosci 40, 157–175 (2016). https://doi.org/10.1007/s10827-016-0590-z
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10827-016-0590-z