[Cognitive advantages of the third age: a neural network model of brain aging]

M P Karpenko,L M Kachalova,E V Budilova,A T Terekhin
Abstract:We consider a neural network model of age-related cognitive changes in aging brain based on Hopfield network with a sigmoid function of neuron activation. Age is included in the activation function as a parameter in the form of exponential rate denominator, which makes it possible to take into account the weakening of interneuronal links really observed in the aging brain. Analysis of properties of the Lyapunov function associated with the network shows that, with increasing parameter of age, its relief becomes smoother and the number of local minima (network attractors) decreases. As a result, the network gets less frequently stuck in the nearest local minima of the Lyapunov function and reaches a global minimum corresponding to the most effective solution of the cognitive task. It is reasonable to assume that similar changes really occur in the aging brain. Phenomenologically, these changes can be manifested as emergence in aged people of a cognitive quality such as wisdom i.e. ability to find optimal decisions in difficult controversial situations, to distract from secondary aspects and to see the problem as a whole.
What problem does this paper attempt to address?