Zachar István

István Zachar is a theoretical evolutionary biologist researching problems related to the major evolutionary transitions, like the origin of life, eukaryotic cells, human cognition and language. He graduated at the Eötvös University, Budapest, in 2004, and is mostly working there ever since. He got his PhD in 2011 on his work on replicator theory, aiming to provide a sound, more-than-philosophical basis for the replicator as the unit of evolution. He also worked on human insight problem solving at the Queen Mary University of London, performing experiments with human subjects, and also was an informal participant in the EU FET-open project “INSIGHT“. He is interested in abstract, theoretical problems that lend themselves nicely for computer modelling, as he is an avid programmer. At present, he is working on modelling Darwinian neurodynamics and also develops in his free time a software to study the evolutionary sound-changes of (artificial) languages, as a homage to J. R. R. Tolkien.

Research at iASK

Insight problems are generally considered hard problems of cognitive psychology, relying on unconscious search for completely novel representations. The brain is known to perform in a massively parallel way, being serialized only at the conscious level. Parallel search and the generation of new variability for selection are the hallmarks of Darwinian search. Our research at the “Cooperation and Conflict in Evolutionary Systems” centre aims to consolidate via modelling and experimentation the Neural Replicator Hypothesis which claims that Darwinian neurodynamics is responsible for certain cognitive functions in humans, and is particularly relevant to theory exploration and generation. My aim is 1) to find a suitable fitness landscape representation of a human-solvable insight problem; 2) to prove that insight problem solution can be represented as evolutionary search over such a fitness landscape, and 3) to prove that evolutionary dynamics coupled with neural networks are able to solve insight problems qualitatively similar to human solvers.