Annika Richter

My background is in Cognitive Science, completing my B.Sc. at the University of Osnabrück and my M.Sc. in Embodied Cognition at the University of Potsdam. During my bachelor studies, I conducted a research internship with RISE (DAAD) at the Social Brain in Action Lab, Macquarie University. There I worked on developing a Python workflow for generating experimental visual stimuli, and was trained in collecting fNIRS data, which evoked my aspiration not only to be fascinated by the human brain but also to actively pursue my own research in human cognitive neuroscience. Furthermore, I joined the Institute of Cognitive Science (IKW) in Osnabrück under Prof. Elia Bruni, where I contributed to an evaluation framework for intuitive physics in multimodal language models; this work was later published at IJCAI 2024. Beyond academia, I worked for four years as a game and concept artist at the start-up 1000 Orks in Osnabrück, contributing from the studio's early stages through the release of its first game, WARCANA. In my master studies, I aimed to integrate different applications of AI and machine learning into the study of psychological and neural processes. During an internship at the Cognitive Neuroscience Lab of the University of Potsdam with Prof. Milena Rabovsky, I used the Sentence Gestalt Model to predict the N400 component in ECoG activity during language processing, highlighting the importance of computational models in both their design and level of detail. In a second internship at the German Institute for Nutrition with Dr. Meriem Ouni, I applied machine learning to high-dimensional methylation data, using differentially methylated sites to predict brain insulin resistance. Together, these experiences deepened my perspective on machine learning approaches to study the brain in two ways: Conceptually, by modelling neural processes and information-processing hierarchies, and practically, by developing new methods to apply machine learning to multidimensional biological data. My master thesis investigated optimal experience during social interaction with humans and AI chatbots through a study at the Max Planck Institute for Human Development (MPIB), measuring self-report, behavioral and physiological measures. Alongside my master studies, I worked as a machine learning specialist at the Max Planck Institute for Human Development. In my position, I developed automated pipelines for fine-tuning, training, and running simulations using large-language models and recommender systems for research on the cultural and technical evolution of AI.

"Foveated Vision Models of Curiosity-Driven Attention in Children and Machines"

In my research, I investigate the neural and cognitive processes underlying curiosity at the Neural Data Science Lab. Using foveated vision models, my project investigates how curiosity guides visual exploration. By combining computational modeling with cognitive neuroscience, I aim to better understand how curiosity shapes perception, attention, and information-seeking behavior in children.

What fascinates me most is how remarkably little we know about how subjective experience arises from the correlational insight we have about the brain and human information-seeking behavior, and how we can use machine learning and computational models to push beyond that, taking different kinds of information into account.

Jassim, S., Holubar, M., Richter, A., Wolff, C., Ohmer, X., & Bruni, E. (2024). Grasp: A novel benchmark for evaluating language grounding and situated physics understanding in multimodal language models [Conference paper]. In K. Larson (Ed.), Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence (IJCAI-24)(pp. 6297–6305). International Joint Conferences on Artificial Intelligence Organization.

• International Interdisciplinary Computational Cognitive Science Summer School (IICCSSS 2024)
• Berlin Neuroscience Meeting 2024 (BNM 24)
• Helmholtz Reproducibility Workshop (2025)