Learning autonomy, autonomous learning, and self-directed learning are central in contemporary labour markets, and in many higher education institutions they are key learning goals. It is also a facet of broader personal development. This text outlines the meta-didactics of autonomy and autonomous learning from an Activity Theory point of view. Activity [Activity: Swedish: “Verksamhet”, German: Tägigkeit] Theory is a trans-disciplinary approach for the study of development, learning and change.

Meta-didactics refer to the more abstract and managerial principles which guide the construction of didactic principles, which in turn are principles that structure a specific teaching moment, behaviour of a digital tool, or communication with a learner.

Autonomous learning spans both practices where learners learn in autonomous mode, as well as hybrid forms of learning, as well as preparatory activity (learning to learn on one’s own) in school, higher education, workplace learning and life-long learning institutions.

In order to make clear what difference it does to think about learning autonomy in Activity Theory terms, Boud’s writings on autonomy is taken as a reference, mainstream version of autonomy. What does it mean when individuals’ autonomy increases or decreases in the context of the larger activity systems in which they participate? The main outcome of the text is the prioritization matrix in the Conclusions section, which is a list of thinking tools to make capabilities in autonomous learning to cross boundaries between modes of practice. The intended use of the text is in pedagogical activity in general as well as when infrastructuring and designing educational technology. Though this text is rather technical in nature, the meta-didactic principles are also important for learner-led learning activity, i.e. that the concepts in this text are grasped by learners.

Often Information Systems research have justified their existence by trying to accumulate knowledge that prevents failures in IT development through identifying and eliminating the causes of failures. These causes have been shown to be hard to capture, possibly because the very concept of causes is not properly understood. Some research schools have a more elaborated notion of the concept of causes than most reported research, Activity Theory being one of them. When Activity Theory-inspired researchers put forward their results, the premises for their predictions and coherence with their alleged activity theoretical position are often implicit, however. In this paper, I make a critical review of classical works commonly drawn upon in Information Systems Activity Theory, in order to systematize their view of predictions. I describe some very general principles related to prediction, namely the principle of determination, the laws of dialectics, the tension between free will and prediction, and the stance towards statistical regularity. From that, the range of possible classes of and some alternatives to predictions are discussed. The last part of the paper reviews how contemporary high quality work in activity theoretical IT research relates to predictions.