Skip to main content
Open Access Publications from the University of California


UCLA Electronic Theses and Dissertations bannerUCLA

When and Why does Learning Profit from the Introduction of Errors?


Generating information from memory not only gives a read out of the contents of memory—it makes those memories stronger than they would have been otherwise (Bjork, 1975). Many researchers have explored the impact of using tests as learning events with a wide range of materials (from word pairs to expository texts) and the effects are largely positive, with tested information faring better than restudied information (Roediger & Karpicke, 2006a). In the current dissertation, I explore a surprising related finding—that guessing answers, even if you are always wrong, can also aid memory when the correct answer is studied afterward (e.g., Kornell, Hays & Bjork, 2009). These results are counterintuitive, as generating guesses takes time out of study, and the guesses could very well interfere with memory for the correct answer. This phenomenon has implications for pretesting in the classroom, where students are likely to generate many errors.

The goal of my research was to explore possible mechanisms of the errorful generation effect. In Chapter 2, I assessed the role of how changing the way incorrect guesses are made affects learning from subsequent feedback. Results showed that giving semantically related guesses improved learning of related word pairs, but making guesses based on another generation rule—rhyme—was not beneficial. In Chapter 3, I report results of experiments testing whether learners are using their incorrect guesses as mediators, or helpful cues to the correct response. While other correlational evidence suggests that mediation is a plausible mechanism, the current experimental evidence does not support a mediation account. Finally, in Chapter 4, I evaluated recent evidence that suggests errorful generation benefits may be hidden in some cases by interference. Results were mixed, and suggested an evaluation of what types of associative and item information are strengthened by errorful generation.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View