Christoph D. Dahl
- Animal and human cognitive processes
- Computational ethology
- Computational modelling
- Machine learning
- Social cognition
- Embodied cognition
Computational Cognition Lab
Research programmes
1 – Formal computational cognitive science
This research programme is situated within formal computational cognitive science, an interdisciplinary framework that seeks to understand cognition through explicit mathematical, computational, and algorithmic descriptions. Rather than treating cognition as a collection of loosely defined capacities, this approach focuses on identifying the formal structures and computational principles that give rise to perception, learning, decision-making, and social interaction across systems. By combining behavioural data with computational modelling, information-theoretic analysis, and machine learning, the programme aims to characterise cognition in terms of transformations of information and constraints imposed by limited resources. Within this framework, empirical work in humans and non-human animals is tightly coupled to mechanistic models that allow hypotheses to be tested at the level of representations, algorithms, and dynamics. Studying cognition in systems ranging from small-brained animals to artificial agents provides a comparative perspective on how complex behaviour can emerge from relatively simple components. Formal computational cognitive science therefore offers a unifying language for integrating computational ethology, mathematical models of cognition, and biologically grounded experimentation into a coherent research strategy.
2 – Human cognition
Within this framework, the human cognition project focuses on understanding how complex cognitive abilities emerge from structured information processing under uncertainty and variability. Using controlled behavioural experiments, psychophysics, and computational modelling, the project investigates core functions such as perception, categorisation, learning, and decision-making. Particular emphasis is placed on how humans achieve robust generalisation despite noisy, incomplete, or ambiguous sensory input, and how internal representations are shaped by experience, task demands, and prior knowledge. Formal models, including neural network architectures, information-theoretic analyses, and neurosymbolic approaches, are used to link behavioural performance to underlying computational mechanisms. By comparing human data with model behaviour, the project aims to identify minimal yet sufficient computational principles that account for human-level flexibility and efficiency. This approach allows human cognition to be studied not only descriptively, but as a formally constrained system whose strengths and limits can be precisely characterised and compared to those observed in other biological and artificial agents.
3 – Animal cognition
The animal cognition project investigates cognitive processes across a diverse range of species, including fish, dogs, and small-brained animals, with the aim of identifying general principles of cognition that transcend specific taxa. Rather than focusing on species-specific abilities in isolation, the project adopts a comparative approach that examines how perception, learning, decision-making, and social interaction are shaped by ecological demands, sensory constraints, and neural architecture. This perspective allows cognitive capacities to be understood as adaptive solutions to environmental problems rather than as fixed, hierarchical traits. Using computational ethology and quantitative behavioural analysis, the project studies cognition in naturalistic and social contexts, capturing fine-grained behavioural dynamics in freely moving animals. In fish, this includes numerosity-based decision-making, social learning, and group dynamics; in dogs, social cognition, communication with humans, and inter-breed variability; and in small-brained animals, the minimal neural circuitry required to support internal representations and higher-order cognitive functions. Across systems, computational models are used to link observed behaviour to underlying information-processing mechanisms. By integrating behavioural data with formal modelling, the animal cognition project aims to uncover shared computational principles that explain how diverse nervous systems generate flexible, adaptive behaviour.
Dogs: This research programme investigates dog cognition as a product of long-term co-evolution with humans, focusing on the mental and behavioural processes that support social interaction and communication. We examine canine social cognition, including how dogs perceive and interpret signals from both conspecifics and humans. A second focus is communication, assessing dogs’ abilities to understand and use gestures, vocal commands, and other communicative cues. Finally, we address inter-breed variability, particularly differences in olfactory cognition linked to skull morphology, connecting cognitive mechanisms to neural structure and applied welfare issues arising from selective breeding.
Fish: This research programme investigates fish cognition as a rich model for understanding decision-making, learning, and social behaviour in dynamic environments. It focuses on cognitive capacities such as numerosity, spatial processing, memory, and social learning, highlighting how ecological pressures shape adaptive tradeoffs between risk, reward, and efficiency. Using zebrafish as a model, the work examines how numerical information interacts with other perceptual cues in context-sensitive decisions. A further emphasis is on group dynamics, employing quantitative, data-driven approaches to characterise social interaction, learning, and information flow within shoals, and to link collective behaviour to underlying neural mechanisms.
Small-brained animals: This research programme adopts a computational ethology approach, an interdisciplinary framework that addresses limitations of traditional ethological methods by integrating tools from mathematics, computer science, and artificial intelligence. Computational ethology aims both to automatically detect predefined behaviours in freely moving animals and to discover novel behavioural patterns from data-driven analyses. Within this framework, we investigate higher-order cognition in small-brained animals by comparing similarities and differences in their sensory systems and examining how these support more complex cognitive processes. By studying the neuronal circuitry underlying internal representations in both virtual and naturalistic environments, including social interactions, this work leverages comparatively simple nervous systems to gain fundamental insights into the mechanisms and computations that give rise to cognition.
