Our approach involves:

  • (i) understanding the creative problem-solving process in natural cognitive systems (humans);
  • (ii) enabling artificial cognitive systems to deploy creative problem solving;
  • (iii) studying the types of knowledge and processes which enable creative problem-solving in natural and artificial cognitive systems.

We believe in the interdisciplinary core of cognitive science, and think cognitive computational modelling and artificial intelligence can strongly support each other.

In this spirit, we think artificial intelligence systems can learn from research on cognition and cognitive processes, especially when it comes to emulating or supporting tasks at which humans are very good, like creativity, creative problem solving, inductive reasoning, and so on. Also, the study of human cognition can benefit from the insights provided by systems attempting to implement cognitive processes and solve tasks cognitively, and from the computational tools thus created.

Many of the systems we have designed have comparability features - that is their performance and processes can be compared to that of humans, or they can be used to implement cognitive theories when exploring human processes, or generate new types of tasks that can be used to study cognition.

An example of this is comRAT-C - a cognitive system aimed initially at solving the compound Remote Associates Test (Olteteanu and Falomir, 2014). comRAT-C's performance showed an interesting correlation with human data (Olteteanu and Falomir, 2015). This prompted us to turn comRAT into a tool for creating new compound RAT queries, which would help cognitive psychologists control different variables (Olteteanu, Schultheis and Dyer, forthcoming). Also, interesting new visual queries were created based on similar principles, thus allowing the study of the differences between cognitive processing with visual and language based queries (Olteteanu, Gautam and Falomir, 2015). Prompted by the correlation with human performance, and with the help of our new tool comRAT-G, we further explored the roles probability and frequency have when humans solve such queries (Olteteanu and Schultheis, forthcoming).


Oltețeanu, Ana-Maria and Falomir, Zoe (2014) – Towards a Compound Remote Associate Test Solver based on Language Data. In Proceedings of the Catalan Conference of Artificial Intelligence, Frontiers in Artificial Intelligence and Applications, IOS Press. DOI: 10.3233/978-1-61499-452-7-249

Oltețeanu, Ana-Maria and Falomir, Zoe (2015) - comRAT-C: A Computational Compound Remote Associate Test Solver based on Language Data and its Comparison to Human Performance. Pattern Recognition Letters, vol. 67, pp. 81-90, doi:10.1016/j.patrec.2015.05.015 -- url

Oltețeanu, Ana-Maria; Schultheis, Holger and Dyer, Jonathan B. - Constructing a repository of compound Remote Associates Test items in American English with comRAT-G, in: Behavior Research Methods, submitted July 2016

Oltețeanu, Ana-Maria, Gautam Bibek and Falomir, Zoe (2015) - Towards a Visual Remote Associates Test and its Computational Solver, in: Proceedings of the International Workshop of Artificial Intelligence and Cognition, AIC 2015, CEUR-Ws Vol. 1510.

Oltețeanu, Ana-Maria and Schultheis, Holger - The Influence of Frequency and Probability in Solving the compound Remote Associates Test, forthcoming.