Research master thesis | Psychology (research) (MSc)
closed access
Prediction-based learning is an effective teaching method for building factual knowledge, i.e., semantic learning. Its effectiveness likely depends on its potential to elicit surprise in learners....Show morePrediction-based learning is an effective teaching method for building factual knowledge, i.e., semantic learning. Its effectiveness likely depends on its potential to elicit surprise in learners. Only a few studies tested this hypothesis using a prediction-based learning framework comparable to semantic learning in the classroom. Most of these studies used physiological measures of surprise. However, the link between prediction-based semantic learning and learners' metacognitive surprise remains to be investigated. Using mixed models, we tested and explored to what degree participants' (N = 41; Mage = 21.9 years, SD = 1.5, 73% female) metacognitive surprise about the learning material (numerical trivia facts) explained how well participants learned (continuous metric) and recalled (binary metric) this material during a numerical-fact learning task designed to resemble classroomlike prediction-based learning. In line with our hypothesis, preregistered analyses showed that the more surprising participants found a fact, the more they learned from it. Extending previous work, we found that this link remained when controlling for a) between-fact differences in learning potential and b) facts already known to the participants and when c) participants failed to recall a fact correctly. Further extending previous work, our exploratory analyses suggested that learning also improved when participants perceived the facts as nonsurprising. So, the link between metacognitive surprise and learning may be u-shaped rather than linear. Altogether, these findings hint that learners'surprise about the learning material is one of the factors explaining to what degree learners learn from their prediction mistakes to update their factual knowledge. We forgo conclusions about the link between metacognitive surprise and recall accuracy since the confirmatory and exploratory results were ambiguous and negligibly small.Show less
Generative learning strategies (GLSs) are activities that enable learners to develop an understanding that goes beyond the information presented by the instructor, depending on how they try to make...Show moreGenerative learning strategies (GLSs) are activities that enable learners to develop an understanding that goes beyond the information presented by the instructor, depending on how they try to make sense of the information presented to them. GLSs have become more popular in recent years. Prediction-based learning is one of the most popular and widely used GLSs. Despite its widespread use, much knowledge still needs to be gained about the moderators by which generating a prediction may improve learning. The present study investigated the effects of gender and executive functioning on the effectiveness of prediction-based learning and the underlying mechanisms of prediction-based learning by combining behavioral assessments with neuroactivity measures (fMRI). 62 native Dutch-speaking young adolescents (38 girls), aged 11 to 13 years old, performed a numerical trivia fact learning task, measuring the test performances of generating predictions, following direct instructions, and learning. A questionnaire regarding executive functioning, critical thinking, and curiosity was also administered. Results indicate that making predictions shows beneficial outcomes in learning among young adolescents, especially among boys. This is likely due to their different approach to learning compared to that of girls. Among one pilot participant, neural activation of a frontoparietal network and frontal, temporal and occipital regions were observed. There are likely many other (external) factors that could potentially affect the effectiveness of prediction-based learning. Further research is recommended to determine these factors in order the gain more knowledge about the mechanisms by which generating a prediction may improve learning.Show less
Research master thesis | Psychology (research) (MSc)
closed access
When learning new information, generating a prediction before receiving the information strongly improves the amount of learning. To this date, the mechanisms underlying why generating predictions...Show moreWhen learning new information, generating a prediction before receiving the information strongly improves the amount of learning. To this date, the mechanisms underlying why generating predictions increases learning are poorly understood. One potential factor that influences this effect is surprise: more surprising information has been demonstrated to deepen information processing when the information differs from our expectations, which occurs when we cannot explain the new information through our belief systems. However, too much surprise may be adverse to learning, as the information may be flagged as implausible and consequently be rejected from our belief network (Munnich & Ranney, 2019). In this study, we investigated the influence of surprise and plausibility on a numerical fact-learning task using three different kinds of learning outcomes: recall, recognition, and memory updating. Using multilevel modelling, we found a nonlinear influence of surprise on immediate recall, as well as a linear influence of surprise on the updating of beliefs. We did not find a significant association between surprise and delayed recognition, nor any significant effects of plausibility on the three measures of learning, although there appeared to be a trend effect of plausibility on the updating of beliefs. Future research should further investigate when newly presented information gets rejected from belief systems, and the role that implausibility of information plays in this phenomenon.Show less
This study compares the effect of the learning strategies generating predictions and passive learning on learning number facts, with the between group factor of order of training and covariates of...Show moreThis study compares the effect of the learning strategies generating predictions and passive learning on learning number facts, with the between group factor of order of training and covariates of executive function and critical thinking skills. Participants took part in a training session, learning 35 number facts by passive learning and another 35 by generating predictions. Half the participants started with passive learning; the other half started with making predictions. After a distractor task, in the form of a self-evaluation questionnaire on executive function, curiosity, and critical thinking skills, participants were tested on how many number facts they remembered correctly. There was an overall significant result of learning strategy, with participants scoring higher when learning from making predictions. Also, a significant interaction result was found for order of training and learning strategy, pointing to a possible carry-over effect for the participants starting the training with generating predictions. The covariates of critical thinking and executive function showed no significant interaction with the learning strategies. These results possibly point to the potential benefit learning from making predictions when learning number facts could have for all students, regardless of individual characteristics.Show less
