Providing implicit formative feedback to learners by combining self-generated and instructional explanations

Formative feedback for a learner typically uses human or artificial intelligence to draw an inference about a learner’s knowledge state from the learner’s actions, and select a learner-directed response. To tackle cases when such intelligence is not easily available, we are exploring ways of providing implicit formative feedback: A learner’s action is to respond to an explanation prompt, and the learner-directed response is to provide an instructional explanation. We consider explanations for correct examples to mathematics exercises, but the exciting implications will be for less well-defined domains that are challenging for cognitive tutors to model. To motivate learners to explain and to increase implicit feedback, we also explore prompts to compare the self-generated and instructional explanations.