A mathematical model of changing the amount of information in the abstract human memory is proposed in the presence of the subsequent "external discrete" training (filling the information). Under this model, the amount of information is a solution of impulsive differential equation with fixed moments of impulsive effects and variable structure. Sufficient conditions are proposed related to the moments and magnitudes of the impulsive effects (i.e., to the moments of discrete training and the volume of the received information), where the quantities of information in two different models of learning can be compared.