In learning, errors are both ubiquitous and inevitable. It is widely understood that these errors may provide a clue about a person's misconceptions. In this article we propose and investigate a model that aims to identify misconceptions from observed errors. We apply the method to single digit multiplication; a domain that is very suitable for the method, is well-studied, and allowed us to analyze over 25,000 error responses from 335 actual learners. The model, derived from the Ising model popular in physics, makes use of a bigraph that links possible errors to possible misconceptions. The error responses were taken from Math Garden, a computerized adaptive practice environment for arithmetic that is widely used in The Netherlands. The results show that the model outperforms a random selection from the observed errors' possible causes, and correctly predicts the possible cause of a person's subsequent error up to over 75% of the time. Finally, we discuss the model, the findings, and the implications.