Selective Propagation Of Pulse Packets In A Feedforward Network By Including Resonance Pair

The brain is always selective in signal routing and transmission.The ways our brain propagate signals and process information selectively in complex environment have always been a hot and difficult topic of research in the fields of computational neuroscience and artificial intelligence.Signals are propagated between different regions in the cerebral cortex through sparse and weak long-range connections.Oscillations can overcome the weak connectivity between networks,synchronizing and amplifying signals,and then promote signal propagation in the sparsely connected feedforward network.At present,several mechanisms of signal propagation based on oscillations have been proposed:communication through coherence,communication through resonance,and communication through resonance in the resonance pair network.The appearance of the resonance pair significantly improves the propagation of feedforward network and reduces the threshold of signal propagation.In this paper,we studied the selective propagation of pulse packet signal in the resonance pair network based on the feedforward network model with resonance pair.In addition,the effect of resonance pair on synchronous signal propagation was discussed by regulating the position,number and connectivity of resonance pair in the network.The main research results of this paper are as follows:Firstly,we simulated multiple resonance pair networks with different activity states and found that all of them responded selectively to a certain input signal,which confirmed that the resonance pair network propagate signals selectively.Secondly,the successful conditions for selective propagatin of pulse packet were studied and the key for successful propagation is that transmission delays between the resonance pair match the network frequency.Finally,in order to control the network more flexibly for selective propagation,we selected background input,connection strength and connection probability as control parameters and found qualitative relationship between the control parameters and the network frequency.In addition,we adjusted the position,number and connectivity of the resonance pair by adjusting the position,number and connectivity of feedback connections,investigating the effect of resonance pair on signal propagation.On the one hand,modulating the position and quantity of the resonance pair could promote signal propagation to a certain extent,on the other hand,it resulted highly synchronized network activities and spontaneous oscillation,which is not conducive to signal propagation.The effect of inhibitory feedback connections on signal propagation depends on the proportion of connections to projecting neurons.When inhibitory feedback connections mainly projected to projecting neurons,signal failed to be propagated.When the inhibitory connections were connected to the projection neuron partly,the signal propagation could be promoted and the network frequency became different.