Griniasty et al. introduced an attractor neural network of the temporal cortex based on the correlationtype associative memory model. In this model, there are parameter regions where the Hopfield attractor and the correlated attractor coexist. We study a method of distinguishing these two attractors. For this purpose, we examine the relaxations of neural firing rate fluctuations. In other words, we introduce sublattices and calculate the correlations of firing rate fluctuations in the

more »

... s using a statistical mechanical method and Monte Carlo simulations. As a result, we found that the relaxation time for the correlated attractor is longer than that for the Hopfield attractor. Therefore, two bistable attractors can be distinguished by observing relaxation times.