Disruption in the neural network has been observed in the clinical studies on dementia. This is investigated here, theoretically, via the macroscopic thermodynamic properties of the Hopfield model to determine whether such a disruption in the network is possible. The analysis is carried out using a mean field theory. The results show a bifurcation in the network in the absence of direct energy input. This is seen when the average connective energy of the neuron becomes equal to or less than its thermal energy. The number of firing neurons that exceed the inactive neurons is then zero and the behavior of the neurons is random. The model further suggests that direct energy input will postpone the degradation of the neural network, suggesting that external mental stimuli will slow this disruption. However, if the neuronal connections are already weak, then improvements will not be marked.