Data & consent
Last updated 6 September 2026
AffectLearn is a research platform as well as a learning one. It exists to test whether a system that notices when a learner is struggling, and responds, helps them more than one that does not. This page explains what that means for you.
What is being studied
Three questions. Which signs of confusion or drifting attention can actually be detected from a webcam and from how someone works. Whether combining those signals beats using the better one alone. And whether responding to them improves what a learner takes away from a lesson.
What taking part involves
- Working through course material as you normally would.
- Occasionally being asked how a section felt. You can skip any of these, and some are deliberately not shown at all so we can tell whether being asked changes how people behave.
- A short questionnaire before you start and a short survey at the end.
- Optionally, having the camera on. Declining changes nothing about your access.
Some material is written to be harder going than the rest, and some to be deliberately plain. That is part of the design rather than a flaw in the course, and there is a debrief at the end explaining which was which.
What the system cannot do
Stated plainly, because it is easy to imply more than is true and we would rather not.
- It detects two states, not four. Confusion, from how you work; drifting attention, from the camera. Frustration and boredom-as-an-emotion have no detector here that we would stand behind, so the system does not claim to see them.
- It is often wrong. The detectors score well below certainty. That is why help is offered rather than imposed, why the system waits for a signal to persist before acting, and why it stays quiet when it is unsure.
- The reasoning behind a hint is currently a set of rules. A fine-tuned language model was trained for this role and evaluated offline, but the production service runs without the hardware to host it, so a deterministic rule map chooses the response instead.
- Whether any of this improves learning is unmeasured. The trial that would answer it has not been run. Nothing here should be read as a claim that it does.
Your rights
- Taking part is voluntary, and declining costs you nothing.
- You can withdraw at any point without giving a reason, including after finishing.
- You can delete your data yourself, from your profile, and see exactly what was removed. See the privacy policy for what that covers.
- You can ask what has been recorded about you and receive a copy.
How records are kept
Research records carry an opaque identifier rather than your name or email, and identifying fields are never included in an export. This is pseudonymisation rather than anonymisation — an administrator with access to both tables could join them — and we describe it accurately rather than calling it anonymous.
Ethical review
This project was carried out as an individual final-year research project and its design, including the webcam analysis and the interaction logging, was reviewed and authorised by the project supervisor. No separate ethics committee reference was issued. We record that here rather than leaving it to be assumed, because the work processes facial imagery of student participants and the basis on which it proceeds should be legible to you.
Contact
Questions, or a request to withdraw: privacy@affectlearn.tech.