On this page, you can read more about my research. Most of my work is centered around improving and applying cognitive models of memory retrieval in educational and clinical contexts. Such models are computer simulations that aim to capture the dynamic process of human learning and forgetting over time, and allow us learn more about the mechanisms that underly learning and forgetting. In addition, we can use them to estimate the memory dynamics of individual learners in real time, and use that information in memorization tools that are tailored towards the needs of individuals, or to diagnose memory impairments in clinical settings. 

Below, you can read more about the main research themes I am working on.

Speech-informed models of memory retrieval

Digital adaptive learning applications use memory models to keep track of the memory processes of individual learners. Typically, they do this by providing retrieval practice questions to the learners, and using the accuracy and latency of typed responses to these questions to inform a memory model that can predict at what point in time the learner will have forgotten the information. They use this information to provide appropriate feedback to the learner or create optimal item repetition schedules. Recent developments in speech technology allow for the transition of typing-based systems to speech-based systems. 

In this paper and in this paper, we showed that benefits of adaptive learning generalize from typing-based learning to speech-based learning, and that spoken responses to retrieval practice questions are more informative of memory strength compared to typed responses.

In another project, we explored the possibility of improving models of memory retrieval by exploiting information present in speech signals. We examined prosodic speech features, such as intonation, rhythm and stress. We demonstrated that prosodic speech features are associated to response times and accuracy scores for retrieval attempts, and that they can be used to predict the extent to which a learner has successfully memorized an item. You can read more about this project here.

I also demonstrated that next to information about the accuracy of a response, it is possible to infer a speaker’s subjective confidence in a response from the speech signal. We found evidence for the idea that the objective memory strength of a response is mainly reflected in a speaker's loudness, whereas the metacognitive process of judging one's certainty or doubt about a response is mainly reflected in a speakers' pitch and speaking speed. 

Individual and group differences

Every learner is unique. Learners differ in experience, prior knowledge, and abilities. Yet, most fundamental learning and memory research is based on aggregate scores for (neuro)typical learners . In my recent work, I use keyword-networks and bibliometric analyses to inspect nearly 23,000 studies in the fields of retrieval practice and specific learning disabilities. Despite a shift from fundamental to applied research in the field of retrieval practice, most research focuses on typically developing learners, and studies into the benefits of retrieval practice for learners with specific learning disabilities is lacking.

Accordingly, a central theme in my work is the investigation of the effects of differences between learners on the accuracy and efficiency of models of memory retrieval in educational contexts. In this project, which was awarded a computational modeling award by the cognitive science society, I compared the performance of learners with developmental dyslexia to the performance of typical learners in an adaptive retrieval practice task using both typing-based and speech-based response conditions. We found that typical learners outperform learners with dyslexia when they are asked to respond by typing, but that this difference disappears when learners respond by speech. Using a mathematical model to decompose response times, we demonstrated that this typing-specific disadvantage in learners with dyslexia is mainly a consequence of processing delays, rather than poorer memory performance. 

This study (preprint) investigates how combining pretesting with posttesting affects learning under realistic digital conditions. Four experiments show that pretesting improves retrieval accuracy and reduces response times during retrieval practice for country outlines, but its benefits diminish after multiple repetitions. However, pretesting can enhance overall learning efficiency by identifying prior knowledge, allowing the exclusion of known items, which benefits learners with moderate to high prior knowledge without disadvantaging those with less prior knowledge.

Finally, I am involved in this project, which aims to develop a personalized, adaptive, and inclusive learning and assessment system to better support neurodivergent and low-performing students. It is a collaboration between MemoryLab, ETS EMEA, and the University of Groningen. Over three years, the team will design, validate, and implement a system tailored to the needs of vulnerable learners, involving northern Dutch schools, teachers, and stakeholders. 

Models of memory retrieval in clinical contexts 

I work as a research scientist for MemoryLab Health, where I contribute to the development of the Seattle-Groningen Memory Assessment (SGMA), a cognitive model-driven memory test suitable for frequent use in research and clinical settings. You can read more about my research on using cognitive models of memory retrieval in clinical settings here.