Writing

This research introduces a Dynamic Field Theory-based process model that perceptually grounds complex, nested linguistic phrases within visual scenes. By autonomously generating neural activation sequences to resolve recursive relational structures, the model provides a neurally plausible solution to the binding problem and demonstrates how combinatorial flexibility and structure-dependent reasoning can emerge from continuous neural dynamics.

Sabinasz, D. (2026). Doctoral thesis, Ruhr-Universität Bochum.

We propose an empirically grounded framework for modeling language understanding, where sentence comprehension emerges from the composition of neural concept representations guided by syntactic structure. This framework integrates insights from cognitive linguistics and computational neuroscience to describe how conceptual structures inform the dynamic integration of meaning.

Sabinasz, D. (2025). Interdisciplinary College

This research introduces a neural process model of visual analogical mapping that executes structural correspondence matching through coupled integro-differential equations. The model provides a dynamic, neurally plausible account of how featural and relational similarities influence mapping, offering an explanation for developmental shifts in performance as a result of varying attentional and inhibitory control.

Kang, M., Sabinasz, D., & Schöner, G. (2024). In L. K. Samuelson, Frank, S. L., Toneva, M., Mackey, A., & Hazeltine, E. (Eds.), Proceedings of the 46th Annual Conference of the Cognitive Science Society.

We present a neural dynamic architecture that models the processing of negated relational phrases within visual scenes. By grounding linguistic input in perceptual processes, the model replicates key psychological phenomena, specifically the negation effect and the polarity-by-truth-value interaction, providing a neurally plausible account of how humans evaluate the truth of negated statements.

Kati, L., Sabinasz, D., Schöner, G., & Kaup, B. (2024). In L. K. Samuelson, Frank, S. L., Toneva, M., Mackey, A., & Hazeltine, E. (Eds.), Proceedings of the 46th Annual Conference of the Cognitive Science Society.

This research utilizes Dynamic Field Theory to provide a neural process account of higher cognition, demonstrating how hallmarks like productivity and systematicity emerge from sensory-motor grounding. By employing neural gain fields and shared index dimensions, the architecture resolves the "massive binding problem" in complex, nested relational phrases.

Sabinasz, D., Richter, M., & Schöner, G. (2023). Cognitive Neurodynamics.

This research introduces a neural dynamic architecture that perceptually grounds nested noun phrases by organizing active visual searches. By utilizing interacting neural populations, the model addresses critical theoretical hurdles, including the massive binding problem and the maintenance of complex, nested structures within short-term memory.

Sabinasz, D., & Schöner, G. (2022). Topics in Cognitive Science.

We present a dynamic neural architecture that detects analogies between objects in visual scenes by utilizing interacting neural populations rather than ungrounded symbolic computation. This model provides a neurally plausible implementation of Structure Mapping Theory, linking high-level analogical reasoning directly to perceptual and attentional processes.

Hesse, M., Sabinasz, D., & Schöner, G. (2022). In J. Culbertson, Perfors, A., Rabagliati, H., & Ramenzoni, V. (Eds.), Proceedings of the 44th Annual Conference of the Cognitive Science Society.

This research introduces a neural dynamic architecture designed to perceptually ground nested noun phrases by organizing visual search processes. The model addresses core theoretical challenges, such as the binding problem and the maintenance of nested structures in short-term memory, providing a biologically plausible solution for resolving complex descriptive language within visual scenes.

Sabinasz, D., Richter, M., Lins, J., & Schöner, G. (2022). In J. Culbertson, Perfors, A., Rabagliati, H., & Ramenzoni, V. (Eds.), Proceedings of the 44th Annual Conference of the Cognitive Science Society. Cognitive Science Society.

A historical trend in the cognitive sciences has been to understand the brain as nature’s way of implementing a computer, a view often termed the Classical Computational Theory of Mind (CCTM). [...]

This paper introduces a Dynamic Field Theory-based neural architecture capable of grounding complex sentences involving nested spatial relations in perceptual input. By utilizing highly recurrent connectivity to autonomously generate processing sequences, the model demonstrates the ability to resolve relational roles and perform hypothesis testing to identify correct grounding outcomes in continuous time.

Sabinasz, D., Richter, M., Lins, J., & Schöner, G. (2020). In S. Denison, Mack, M., Xu, Y., & Armstrong, B. C. (Eds.), Proceedings of the 42nd Annual Conference of the Cognitive Science Society (pp. 620–626). Cognitive Science Society.

Evolutionary psychology is an approach to understand human behavior that combines insights gained from evolutionary biology, the computational sciences and the study of ancestral living conditions. [...]

This thesis presents a Dynamic Field Theory (DFT) based neural architecture for grounding complex combinatorial concepts in perceptual input. By demonstrating that semantic compositionality can emerge from neural dynamics without symbolic computation, the work provides a neurally plausible framework for resolving intricate descriptive phrases and addresses key cognitive benchmarks like productivity and systematicity.

Sabinasz, D. (2019). Master’s thesis, Ruhr University Bochum, Germany.

This research presents a neural dynamic architecture that provides a biologically plausible alternative to symbolic computation for higher cognition. By grounding combinatorial concepts directly in perceptual systems, we demonstrate how complex linguistic and cognitive structures can emerge from neural dynamics without the need for traditional amodal symbol systems.

Sabinasz, D., Richter, M. Lins, J. & Schöner, G. (2019). In Proceedings of the 41st Annual Conference of the Cognitive Science Society.

A series of tutorials in which we learn how to build a content-based search engine that retrieves multimedia objects based on their content rather than based on keywords, title or meta description.

Unbalanced training data leads to misleading accuracy metrics. Learn practical techniques to overcome class imbalance and optimize your model for real-world use.

A series of tutorials in which we explore methods for robot localization: the problem of tracking the location of a robot over time with noisy sensors and noisy motors, which is an important task for every autonomous robot, including self-driving cars.

This text deals with arguments against the possibility of so-called strong artificial intelligence, with a particular focus on the Chinese Room Argument devised by philosopher John Searle.

This text gives an overview of Gödel’s Incompleteness Theorem and its implications for artificial intelligence. Specifically, we deal with the question whether Gödel’s Incompleteness Theorem shows that human intelligence could not be recreated by a traditional computer.

A series of tutorials, in which we develop the mathematical and algorithmic underpinnings of deep neural networks from scratch and implement our own neural network library in Python.

This research addresses the computational intensity of the Earth Mover’s Distance (EMD) by introducing signature compression approximations designed for large-scale datasets. We provide a theoretical analysis of this approximation-based approach and demonstrate through extensive experimentation that it maintains high retrieval accuracy while significantly reducing computational costs, prioritizing real-time user responsiveness.

Uysal, M. S., Sabinasz, D., & Seidl, T. (2016). In International Conference on Database Systems for Advanced Applications (pp. 165–180). Springer.

This study provides a comprehensive comparative analysis of distance-based similarity measures—including EMD and IM-Sig—for large-scale, near-duplicate video retrieval. Through experiments across eight diverse databases, we demonstrate that IM-Sig offers high retrieval accuracy, consistently matching or exceeding the performance of EMD while outperforming other state-of-the-art methods in handling various video edit tasks.

Uysal, M. S., Beecks, C., Sabinasz, D., & Seidl, T. (2015). In 2015 IEEE International Symposium on Multimedia (ISM) (pp. 254–257). IEEE.

This work provides the first comprehensive evaluation of lower-bounding filter distance functions for the Earth Mover’s Distance (EMD) within signature-based video databases. By testing techniques like Independent Minimization for Signatures (IM-Sig) on datasets of up to 350,000 videos, we demonstrate that IM-Sig achieves superior retrieval effectiveness and efficiency, outperforming existing state-of-the-art methods.

Uysal, M. S., Beecks, C., Sabinasz, D., & Seidl, T. (2015). In Proceedings of the 2015 Workshop on Large-Scale and Distributed System for Information Retrieval (pp. 3–8).

This paper introduces FELICITY, an interactive system for content-based video retrieval that utilizes video signatures and the Earth Mover’s Distance (EMD). We implement a multi-step query processing algorithm using lower-bounding filters to ensure efficiency, while providing a visualization interface that allows users to intuitively explore EMD flows and tune parameters for optimal similarity results.

Uysal, M. S., Beecks, C., Sabinasz, D., & Seidl, T. (2015). In International Conference on Similarity Search and Applications (pp. 347–350). Springer.

This research introduces a novel video retrieval method using "feature signatures"—compressed representations derived from clustered feature vectors. By evaluating the Earth Mover’s Distance and Signature Quadratic Form Distance alongside optimized indexing techniques, we demonstrate significant improvements in retrieval effectiveness compared to existing state-of-the-art approaches.

Sabinasz, D. (2015). Bachelor’s thesis, RWTH Aachen University, Germany.