Lu Suojun, Fang Jian'an, Guo Aike, Peng Yueqing
College of Information Science & Technology, Donghua University, 2999 North Renmin Road, Songjiang District, Shanghai 201620, China.
Neural Netw. 2009 Jan;22(1):30-40. doi: 10.1016/j.neunet.2008.09.012. Epub 2008 Oct 9.
The dynamical behaviors of a neural system are strongly influenced by its network structure. The present study investigated how the network structure influences decision-making behaviors in the brain. We considered a recurrent network model with four different topologies, namely, regular, random, small-world and scale-free. We found that the small-world network has the best performance in decision-making for low noise, whereas the random network is most robust when noise is strong. The four networks also exhibit different behaviors in the case of neuronal damage. The performances of the regular and the small-world networks are severely degraded in distributed damage, but not in clustered damage. The random and the scale-free networks are, on the other hand, quite robust to both types of damage. Furthermore, the small-world network has the best performance in strong distributed damage.
神经系统的动力学行为受到其网络结构的强烈影响。本研究调查了网络结构如何影响大脑中的决策行为。我们考虑了具有四种不同拓扑结构的循环网络模型，即规则、随机、小世界和无标度。我们发现，小世界网络在低噪声决策方面表现最佳，而随机网络在噪声较强时最为稳健。在神经元损伤的情况下，这四种网络也表现出不同的行为。规则网络和小世界网络在分布式损伤中性能严重下降，但在集群损伤中则不然。另一方面，随机网络和无标度网络对这两种类型的损伤都相当稳健。此外，小世界网络在强烈的分布式损伤中表现最佳。
