That is the central question driving the NIAS-Lorentz Theme Group, Abstraction, Broad Generalisation, and Composition: The ABCs of Analogy, winner of the NIAS-Lorentz Call 2026/2027.
Using computational models and large language models as experimental testbeds, the group examines analogical reasoning through three lenses: abstraction, generalisation, and composition. Their work aims to shed new light on how this fundamental cognitive capacity develops in children, how language shapes it, and what human reasoning can teach us about AI — and vice versa.
The group brings together Lucia Donatelli (linguistics, coordinator), Martha Lewis (computer science / neurosymbolic AI), and Claire Stevenson (psychology), combining three disciplines that each capture a different dimension of how analogy works in minds and machines.
More information about the group and its members can be found on the NIAS website.
Lucia Donnatelli Claire Stevenson Martha Lewis
