The article considers the problem of actions–beliefs link. As author shows, the widespread approach in social science, those origins can be traced back to Hume and Mill and which tries to reveal the causal relations between beliefs and actions, is mistaken. It is mistaken because it proposes that, firstly, beliefs and actions are distinct and separately identifiable social phenomena and, secondly, causal connection consists in constant conjunction. MacIntyre, instead, proposes, taking as a starting point the distinction between physical movement and human action, to consider the actions–beliefs link in terms of the descriptions which the action should correspond to. If we, on being asked for an explanation of what we have done, refer it to an antecedent condition of a Humean kind, we precisely remove it from the class of actions and assign it to, most probably, the class of physical movements. To explain behavior as a genuinely human action, an explanation must refer to the customarily recognized rules of a particular social order. This presupposes that (1) action must fall under some description which is socially recognizable as the description of an action; (2) an action must fall under a description and my actions under a description available to me; and (3) agent can do only what he/she can describe. As an illustration of his approach, author examines the role of Stalin’s philosophical work “Dialectical and historical materialism” in the process of the ideological “closing” of Soviet society.