Prof. Constantin A. Rothkopf, PhD

Constantin Rothkopf ist W3 Professor am Institut für Psychologie am Fachbereich Humanwissenschaften und Zweitmitglied im Fachbereich Informatik sowie Gründungsdirektor des Centre for Cognitive Science an der Technischen Universität Darmstadt. Er ist darüber hinaus Gründungsmitglied des Hessischen Zentrums für Künstliche Intelligenz (hessian.ai), Mitglied der ELLIS Unit Darmstadt, Mitglied der DAAD Konrad Zuse Schools of Excellence in Artificial Intelligence ELIZA und Co-Sprecher des Clusterprojekts The Adaptive Mind und des LOEWE Schwerpunkts Whitebox. Nach der Promotion in Gehirn- und Kognitionswissenschaften sowie in Informatik an der University of Rochester, NY war er ab 2009 als Postdoc am Frankfurt Institute for Advanced Studies (FIAS) in der theoretische Neurowissenschaften Gruppe tätig. 2009 begann er an der Goethe Universität, Frankfurt zu lehren und von 2010 bis 2012 war er Principal Investigator der Forschungsgruppe “beliefs, representations, and actions” am FIAS. Nach einem Jahr als Vertretungsprofessor am Institut für Kognitionswissenschaft an der Universität Osnabrück, hat er 2013 die W2 Professur „Psychologie der Informationsverarbeitung“ an der TU Darmstadt angetreten. Während des Wintersemesters 2017 war er Gastprofessor im Department of Cognitive Science an der Central European University, Budapest. 2022 erhielt er vom European Research Council einen ERC Consolidator Grant für sein Projekt 'ACTOR'. Im Sommersemester 2023 war er Gastprofessor am Zuckerman Institute, Columbia University, New York, USA.

Constantin Rothkopf's research interests revolve around the distinction between 'looking' and 'seeing' and how this distinction relates to vision in goal directed behavior in an ambiguous, uncertain, and variable world. The aim is to better understand the interrelationship between perception and action in humans, i.e. how we use our perceptual systems actively during natural extended behavior to guide decisions and actions with our bodies. This leads to the study of how we use sensory input, specifically vision, form beliefs about the world by carrying out computations on the basis of our cognitive representations, and then employ decision making processes to act in goal directed behavior. To achieve this goal he uses experimental studies in humans as well as computational modeling involving methods from statistical and machine learning. His current focus is on:

• behavioral studies involving eye tracking of human eye movements during complex naturalistic tasks in natural and virtual environments,
• developing inverse optimal control and inverse optimal decision-making models that allow recovering individual subjects' subjective beliefs, intrinsic costs, and internal models,
• building computational models of tasks and developing algorithms that learn how to solve these tasks in virtual agents,
• developing models for the representation and quantification of extended sequential human behavior,
• simulation of learning algorithms in naturalistic virtual environments,
• developing learning algorithms with an emphasis on the learning of sensory representations for actions.

Constantin Rothkopf's research interests revolve around the distinction between 'looking' and 'seeing' and how this distinction relates to vision in goal directed behavior in an ambiguous, uncertain, and variable world. The aim is to better understand the interrelationship between perception and action in humans, i.e. how we use our perceptual systems actively during natural extended behavior to guide decisions and actions with our bodies. This leads to the study of how we use sensory input, specifically vision, form beliefs about the world by carrying out computations on the basis of our cognitive representations, and then employ decision making processes to act in goal directed behavior. To achieve this goal he uses experimental studies in humans as well as computational modeling involving methods from statistical and machine learning. His current focus is on:

• behavioral studies involving eye tracking of human eye movements during complex naturalistic tasks in natural and virtual environments,
• developing inverse optimal control and inverse optimal decision-making models that allow recovering individual subjects' subjective beliefs, intrinsic costs, and internal models,
• building computational models of tasks and developing algorithms that learn how to solve these tasks in virtual agents,
• developing models for the representation and quantification of extended sequential human behavior,
• simulation of learning algorithms in naturalistic virtual environments,
• developing learning algorithms with an emphasis on the learning of sensory representations for actions.

T. Thomas, D. Straub, F. Tatai, M. Shene, T. Tosik, K. Kersting & C. A. Rothkopf. Modelling dataset bias in machine-learned theories of economic decision-making. Nature Human Behaviour, 2024, [https://doi.org/10.1038/s41562-023-01784-6]

D. Straub*, M. Schultheis*, H. Koeppl, C. A. Rothkopf. Probabilistic inverse optimal control for non-linear partially observable systems disentangles perceptual uncertainty and behavioral costs. Advances in neural information processing systems (NeurIPS), 2023

D. Straub, C. A. Rothkopf. Putting perception into action with inverse optimal control for continuous psychophysics, eLife, 11:e76635, 2022, [https://doi.org/10.7554/eLife.76635]

M. Schultheis*, D. Straub*, C. A. Rothkopf. Inverse stochastic optimal control adapted to the noise characteristics of the sensorimotor system. Advances in neural information processing systems (NeurIPS), 2021, [https://proceedings.neurips.cc/paper/2021/hash/4e55139e019a58e0084f194f758ffdea-Abstract.html]

D. Hoppe, C. A Rothkopf. Multi-step planning of eye movements in visual search. Scientific Reports, 9(1):144, 2019, [https://doi.org/10.1038/s41598-018-37536-0]

D. Hoppe, S. Helfmann, C. A. Rothkopf. Humans quickly learn to blink strategically in response to environmental task demands. Proceedings of the National Academy of Sciences (PNAS), 2018, [https://doi.org/10.1073/pnas.1714220115]

B. Belousov, G. Neumann, C. A. Rothkopf, J. Peters. Catching heuristics are optimal control policies. Proceedings of the Annual Conference on Neural Information Processing Systems (NIPS), 2016, [https://proceedings.neurips.cc/paper/2016/hash/43fa7f58b7eac7ac872209342e62e8f1-Abstract.html]

D. Hoppe, C. A. Rothkopf. Learning rational temporal eye movement strategies. Proceedings of the National Academy of Sciences (PNAS), 113(29), 8332-8337, 2016, [https://doi.org/10.1073/pnas.1601305113]

C. A. Rothkopf, D. H. Ballard. Modular inverse reinforcement learning for visuomotor behavior. Biological Cybernetics, 107(4), 477-490, 2013, [https://doi.org/10.1007/s00422-013-0562-6]

C. Dimitrakakis, C. A. Rothkopf. Bayesian multitask inverse reinforcement learning. European Workshop on Reinforcemnt Learning (EWRL), September 9–11, 2011, [https://doi.org/10.1007/978-3-642-29946-9_27]

M. M. Hayhoe, C. A. Rothkopf. Vision in the natural world. Wiley Interdisciplinary Reviews: Cognitive Science, Wiley, 2010, [https://doi.org/10.1002/wcs.113]

C. A. Rothkopf, D. H. Ballard, M.M. Hayhoe. Task and context determine where you look. Journal of Vision, 7(14):16, 1-20, 2007, [https://doi.org/10.1167/7.14.16]