UC Berkeley

Conceptual change represents a crucial, challenging, learning component. This study hypothesized and observed evidence for two parallel forms of learning within the Numerically-Driven Inferencing (NDI) paradigm's rather minimalist intervention of providing direct feedback regarding a numerical estimate-feedback that yields remarkably robust cognitive alterations. The present experiment probed the nature of learning apropos recall or estimation improvements observed after participants (a) provided estimates, (b) received feedback, and (c) re-estimated after waiting for one day. The results show that improved estimation/recall was predicted by two independent elements-surprise at feedback and an explicit sense of episodic recall upon testing. This suggests at least two learning processes: (1) an explicit (though perhaps approximate) recollection of a quantity's magnitude and (2) a non-episodic semantic restructuring that correlates with surprise. Thus, even for concise, factual information, we educators might consider students' "unknown knowns" - knowledge that learners gain without any explicit understanding that they have done so.