We all need a certain degree of agency – the freedom to make our own decisions. When children are learning, they need agency to feel in control and enjoy the learning process, but at the same time, they struggle to choose effective learning strategies. Knowing which learning strategies will be helpful for a particular task requires metacognition – understanding and regulating one’s own thoughts, a capacity which develops through childhood. These conflicting needs play out in various learning contexts, but perhaps most strongly in educational technologies (EdTech).

More on agency
How can children exercise their agency?

Based on our research, we’ve seen that EdTech tends to go to one extreme or the other. Some learning apps are designed to use data in a sophisticated way to assign ‘optimal’ tasks to learners, eliminating choices. Others give learners free rein to explore, sometimes offering gaming possibilities – children have agency, but their learning may be undermined if they fail to make the best choices. Given the conflict between children’s developing metacognition and the need to foster their agency, is a balance possible?

To find out, we first reviewed various perspectives on agency. The philosophical view holds that agency is always present to some degree, and plays out differently in different contexts. The educational perspective emphasises how teachers can support children’s agency, for example by providing activities that foster agency. A psychological perspective focuses on tailoring the degree of agency in an app to a learner’s characteristics. 

“When children are learning, they need agency to feel in control and enjoy the learning process.”

Based on what we have learned, we presented a framework that makes balanced agency possible. An EdTech app can adapt the degree of agency based on information it has gathered on learner characteristics, such as a child’s prior knowledge of the subject or ability to employ certain learning strategies. This information can be used to determine how much choice a child is given when using the app. This should be done in an adaptive manner – in a loop – so that the degree of agency changes with changes in the child. For example, if children make rapid progress towards a learning goal, the app could give them more agency, allowing them to choose a different activity next.

Through this approach, EdTech strikes a balance between allowing learners to freely choose learning content and assigning ‘optimal’ tasks. App designers need to individualise content according to each child’s needs, while simultaneously expanding the range of content presented to the child.

More on educational technology
Using edtech to personalize learning

What might this look like in practice? Inspired by the philosophical view, we think of agency in terms of four dimensions that should be considered when designing EdTech. First is what can be chosen, such as the content presented, progress through a game, or the appearance of an activity. Second is when choices can be made – continually throughout the activity or only at the beginning. The third is where these choices can be made, for example in a dashboard or within a task. The fourth dimension, finally, relates to who makes the choices – the child, the EdTech, the teacher, or a combination. All of these dimensions are important in designing EdTech.

“EdTech needs to adjust the degree to which learners have control as they progress in their learning.”

Thus young learners who do not yet have the knowledge and skills to effectively self-direct their learning would be given easy choices. They might choose the appearance of the app, while the app chooses the content. Older learners might have the skills to select their own learning path. Considering these four dimensions can help EdTech designers to strike an optimal balance between allowing children to make free choices and ensuring that they are assigned optimal content.

It is important to give children the right amount of agency. This means that EdTech needs to adjust the degree to which learners have control as they progress in their learning, and monitor how such adjustments help or hinder their learning and development. We hope that this approach will be the starting point for future research aimed at helping EdTech to support learner variability and deliver on its promise to be truly educational.


A paper describing this work in detail has been published with our colleagues in Educational Psychology Review. It includes examples of existing EdTech apps that differ in how they provide agency to learners.

Keep up to date with the BOLD newsletter