Available thesis topics
We are pleased that you are interested in the topics covered by our chair!
In this section, you will find the current research topics we offer for student theses. You are also welcome to approach us with your own topic idea. Taking into account the current research focus and the available capacity of the chair, we will then assess whether supervision is possible.
Reasoning & Decision-Making:
Understanding Decisions Under Uncertainty Using Machine Learning
Understanding Decisions Under Uncertainty Using Machine Learning
Keywords: Decision theory, interpretable AI, machine learning, behavioral economics, cognitive modeling
Decisions in fields such as economics, finance, or healthcare are often associated with considerable uncertainty. Understanding how and why people make these decisions is one of the main goals of behavioral research. To investigate this question, methods from classical cognitive psychology are increasingly complemented by machine learning techniques. These methods offer high predictive accuracy, sometimes at the cost of interpretability. However, feature importance methods can be used to attempt to understand the underlying decision processes. This topic focuses on analyzing data from behavioral experiments (e.g., lottery decision-making) using these approaches.
Within this topic, classical machine learning methods (e.g., Random Forest, Support Vector Machine) could be applied to an existing dataset on decisions under uncertainty. Subsequently, a feature importance analysis could be performed on these models and the results evaluated. Depending on their interests and background, students can set their own focus—from the mathematical foundations of the models, through practical implementation in Python or R, to the psychological interpretation of the results.
This project offers the opportunity to gain practical experience with modern machine learning techniques and interpretation methods (feature importance).
Requirements:
- Basic knowledge of quantitative methods (e.g., statistics, mathematics, data analysis)
- Ideally, some initial programming experience (Python or R)
- Prior knowledge of machine learning is helpful but NOT required
Contact Person:
Human-AI Collaboration:
Designing and Evaluating Human-AI Collaborative Conversational Agents for Enhanced Decision-Making
Designing and Evaluating Human-AI Collaborative Conversational Agents for Enhanced Decision-Making: A Focus on Information Seeking and Trust
Keywords: Human-AI collaboration; Conversational agents; Information seeking; Trust; Human-Computer Interaction
Goal:
To investigate how AI can be effectively integrated with humans to enhance decision-making in everyday contexts (e.g., healthcare, learning, daily tasks) by ensuring AI systems are reliable, trustworthy, and capable of fostering user confidence and critical thinking in their interactions with AI-based systems.
Description:
The research studies how AI can work effectively with humans to help make better decisions in everyday situations, such as learning, healthcare, or even everyday tasks. It builds on earlier research where an AI tool (like a "Smart Librarian") helped pilots by providing information for emergency situations in the cockpit. The focus is on understanding how AI can be a reliable and trustworthy partner, ensuring people feel confident in its answers. The research also explores whether people's satisfaction with AI tools reflects how well they perform in real-world tasks.
Requirements:
This is an experimental study of user interactions with AI-based systems. Fundamental knowledge of human-AI collaboration is essential, and familiarity with experimental design and basic statistical analysis techniques (e.g., ANOVA) is recommended.
Expected Outcomes:
- Empirical evidence linking user perception-based measures to AI system performance. Conduct a controlled user study to investigate whether user satisfaction with AI tools (e.g., perceived helpfulness, confidence in AI advice) correlates with real-world task outcomes (e.g., accuracy, task completion time, or error reduction).
- A case study demonstrating human AI collaboration in a real-world task. Apply the research to a specific domain (e.g., healthcare diagnostics, learning platforms, or emergency response systems) and evaluate how human-AI collaboration affects decision quality, user trust, or task efficiency.
Contact Person:
Human-AI Collaboration:
AI as a Collaborative Learning Partner in Student Group Projects
AI as a Collaborative Learning Partner in Student Group Projects
Keywords: Human-AI collaboration; Collaborative learning; Search as learning; Group work; AI tools
Goal:
To design and test AI tools that help teams collaborate more effectively by offering suggestions, organizing tasks, or clarifying ideas.
Description:
This study explores how AI can enhance group work by addressing challenges like communication gaps, uneven participation, and unclear direction. It will design and evaluate tools with features such as suggestion generation, task assignment, and idea clarification. It will test their impact on group communication, learning outcomes, and user trust in AI recommendations, and evaluate adaptability to diverse team dynamics and learning styles.
Requirements:
Knowledge of human-AI collaboration and experimental design is essential. Technical skills in programming or AI development if building custom tools.
Expected Outcomes:
- Empirical data linking user perception (e.g., trust, satisfaction) to AI tool performance (e.g., task efficiency, accuracy).
- A case study applying AI tools to a real-world educational task (e.g., group research) and evaluating their impact on collaboration and learning.
Contact Person:
Human-Centered AI (HCAI):
Adaptive Interface Design for Enhancing Cognitive Capability in Complex Information Seeking
Adaptive Interface Design for Enhancing Cognitive Capability in Complex Information Seeking
Keywords: User interface design; Cognitive ability; Information seeking; Search as learning
Goal:
To design, prototype, and evaluate an adaptive search interface that dynamically responds to users' real-time cognitive states (e.g., cognitive load, cognitive style, anxiety) to enhance learning, critical thinking, and decision-making during complex and exploratory information seeking tasks.
Description:
This research explores how AI can assist students in group work, such as research projects, presentations, or creative tasks. The research will investigate how these tools impact group communication, learning outcomes, and whether students trust the AI's recommendations. It will also examine if AI can adapt to different team dynamics and learning styles, making collaboration easier and more productive.
Requirements:
Familiarity with system design and cognitive modelling is essential. Technical skills in developing adaptive are recommended.
Expected Outcomes:
- A functional prototype of an adaptive search interface, incorporating real-time adjustments based on user cognitive states.
- Empirical evidence from user studies demonstrating the interface’s impact on learning outcomes, cognitive efficiency and user experience.
Contact Person:
Human-Centered AI (HCAI):
Human-Centered AI for Augmenting Information and Knowledge Extraction from Complex Survey Data
Human-Centered AI for Augmenting Information and Knowledge Extraction from Complex Survey Data
Keywords: Human-Centered AI; Survey data; AI tools; Topic extraction; Thematic analysis
Goal:
To develop and evaluate Human-Centered Artificial Intelligence (HCAI) tools that support researchers in efficiently analyzing complex mixed-method survey data, with a focus on identifying patterns in qualitative responses and their connections to quantitative data.
Description:
This study addresses the challenges of analyzing mixed-method survey data by creating HCAI tools that combine automated techniques with human oversight to extract, interpret, and synthesize findings. The research emphasizes usability, transparency, and collaboration between AI systems and researchers, ensuring tools are designed to domain-specific needs and ethical considerations. It explores how these tools can enhance the analysis of qualitative comments, improve the identification of themes, and strengthen the integration of qualitative and quantitative findings in survey research.
Requirements:
Ability to manage and analyze large, complex survey datasets is essential. Familiarity with survey methodology, thematic analysis, and mixed-methods research design is recommended.
Expected Outcomes:
- A functional HCAI tool that automates or assists in tasks like topic extraction, thematic coding, and linking qualitative findings to quantitative results.
- Empirical validation that demonstrates the tool’s effectiveness in improving accuracy, efficiency, or interpretability of survey data.
Contact Person:
Human Problem Solving:
Investigating Human Understanding of Relationships in Systems
Investigating Human Understanding of Relationships in Systems
Keywords: Human problem solving, implicit learning, cognitive modeling, web experiments
In everyday life, people are often implicitly required to grasp (mathematical) relationships: ranging from simple cases like ratios in cooking recipes to complex relationships in financial planning, the interplay of controls in software (e.g., image editing), or physical phenomena. In doing so, we need to form a mental representation of the system to make predictions about how the system would behave under different parameters. This ultimately requires the successful interplay of two phases: an exploration or learning phase, in which a mental representation is built, and a mental simulation phase, in which the final expectation is formed. Naturally, the complexity of the system plays a role here — which is why it is hardly surprising that many studies use complexity as a primary measure.
Complexity is often created by the number of variables. However, the goal of this topic is the systematic investigation of complexity arising from the relationships themselves: for example, exponential relationships are known to be substantially more difficult for people to grasp than linear ones. Within this topic, a scenario should be developed and implemented as a web experiment that systematically examines the effects of different mathematical relationships on exploration and mental simulation. The data obtained will then be used to develop a model that can predict performance in working with a system based on the complexity of its underlying relationships.
This topic thus offers the opportunity for independent and complete scientific work—from study design to the final development of a model.
Requirements:
- Knowledge of HTML/JavaScript for developing a web experiment
- Ideally, skills in Python for data analysis
- Initial experience in study planning
- Preferably knowledge in cognitive science or psychology
Contact Person:
Modeling in Information Processing:
Modeling Human Reasoning Processes
Modeling Human Reasoning Processes
Keywords: Cognitive modeling, human reasoning, machine learning, data science
The ability to draw conclusions from given information is one of the fundamental components of human intelligence. However, people often deviate significantly from logic, and it is precisely these deviations that provide insights into the underlying processes. Cognitive science research has already identified a variety of processes, heuristics, and assumptions underlying reasoning.
Usually, however, only a single type of reasoning task has been considered at a time, so it remains unclear whether and to what extent common processes exist (a "general reasoning ability").
This topic aims to approach this question—with a dataset of reasoning tasks from various domains available. The goal is to investigate the transferability of information between different types of reasoning—using methods from data analysis and machine learning, as well as studies based on cognitive science theories and cognitive modeling methods. Depending on background and interest, the focus can be chosen from the entire range of modeling approaches. In any case, knowledge of Python is essential, as existing analyses and models will need to be used and a new implementation developed.
Requirements:
- Proficiency in Python
- Initial experience working with data (e.g., preprocessing for machine learning, statistical analysis, etc.)
- Ideally, background in cognitive science, with a focus on reasoning
Contact Person:
