ANNe - Artificial Neural Network with Emotion
ANNe stands for Artificial Neural Network with Emotion and unlike standard neural network contains one-dimensional model of emotion and is used for simulation of emotional disorders (e.g. depression and schizophrenia). The presented model with source code below is very primitive but still it shows that for explaining the mentioned disorders we may not need disfunctions at molecular or cell level, but we need "only" synaptic plasticity and the existence of emotion related structures in the central nervous system. Assumptions of the simulation are as follows: 1.The neural network is generated in random way, the emotion related structure is created as the first as group of neurons, every other neuron is connected to a randomly selected existing neuron using synaptic connection with random weight. 2.The simulation runs as generation of nerve pulses of random strength into random neurons, at randomly distant moments. The output is graph emotion versus time. 3.The weight of every synaptic connection drops exponentially with time with the koeficient Kwd (koeficient of weight dropping). 4.Every generated global emotion is decreasing exponentially with time with the koefficient Ked (koeficient of emotion dropping). The dropping of emotion is significantly faster (minutes) than dropping of the synaptic connection weight (days, weeks and more). 5.Run of nerve pulse causes reinforcement of the synaptic connection, this weight increase is proportional to the strength of the nerve pulze. The mechanisms 3. and 5. work against each other. If the reinforcement process prevails after exceeding certain threshold, the neural network can enter the state of "emotional disorder" characterized by permanently high value of emotion almost independent on what random pulse runs into the network. Because of negative back-coupling this is a stable state of the neural network. Outputs of the Simulation: Emotion versus time graph of the network in "healthy" state, peaks correspond to moments when problems appeared and there was temporary increase of the global emotion value:
Emotion versus time graph of the network in starting emotional disorder:
ANNe.cpp - source code of the simulation.