Inspired by the recent literature on aggregation theory, we aim at relating the long range correlation of the stocks return volatility to the heterogeneity of the investors' expectations about the level of the future volatility. Based on a semi-parametric model of investors' anticipations, we make the connection between the distributional properties of the heterogeneity parameters and the auto-covariance/auto-correlation functions of the realized volatility. We report different behaviors, or change of convention, whose observation depends on the market phase under consideration. In particular, we report and justify the fact that the volatility exhibits significantly longer memory during the phases of speculative bubble than during the phase of recovery following the collapse of a speculative bubble.