Carlos Zednik, PhD

My academic research concerns the explanation of natural and artificial intelligence. In my published work, I investigate methodological principles of cognitive psychology and neuroscience, and investigate methods, norms, and applications for explainable AI. I am the director of the Eindhoven Center for Philosophy of AI, a member of the alignAI (ERC) and ROBUST (NWO) research consortia, and an Associate Editor for the not-for-profit open-access journal Philosophy and the Mind Sciences.

My teaching in Eindhoven generally covers philosophical and ethical issues in AI, but also concerns the psychology of judgment and decision-making. More generally, my teaching addresses a wide range of topics across philosophy of mind, philosophy of science, cognitive science, and logic. I am also engaged in innovation projects to integrate the use of generative AI in higher education.

Beyond the university, I am involved in national and international standardization efforts to promote trustworthy artificial intelligence, my primary area of expertise being explainability in machine learning. Nationally, this involves committee work at the German (DIN) and Dutch (NEN) standardization bodies. Internationally, I participate in the ISO/IEC joint technical committee SC42 - Artificial Intelligence.

Before arriving in Eindhoven, I was based at the Philosophy-Neuroscience-Cognition program at the University of Magdeburg, and prior to that, at the Institute of Cognitive Science at the University of Osnabrück. I received my PhD from the Indiana University Cognitive Science program, after receiving a Master's degree in philosophy of mind from the University of Warwick and a Bachelor's degree in computer science and philosophy from Cornell University.

You can find out more about me on Google Scholar, PhilPapers, Web of Science, and LinkedIn.

☞ See my Google Scholar profile for citation metrics.

Zednik, C. & Verreault-Julien, P. (forthcoming)
Can XAI Justify?
Philosophy of Science for Machine Learning. Edited by J. Duran & G. Pozzi. Dordrecht: Springer.
Abstract  Chapter

AI systems are increasingly being used to drive high-stakes decision-making in domains such as medical diagnosis, credit approval, and human resources. Ensuring that AI-driven decisions are properly justified is crucial. Justification implies that decisions align with relevant norms. However, the opacity of many AI systems hinders this kind of justification since it makes it difficult to determine which features influence these systems' outputs and whether they have learned to track spurious correlations rather than relevant patterns. Although explainable AI (XAI) offers potential solutions by uncovering the reasons for AI systems’ outputs, its role in justification remains underexplored. This chapter adopts a framework from philosophy of action to differentiate between the different types of reasons that may be given for an AI system's outputs, assess how these types of reasons contribute to justification, and evaluate the capacity of current XAI methods to identify and describe such reasons. Overall, while the chapter shows how XAI methods could be used to identify reasons with which to justify AI-driven decisions, it also highlights some of these methods' current limitations.

Budding, C. & Zednik, C. (2024)
Does Explainable AI Need Cognitive Models?
Proceedings of the 46th Annual Conference of the Cognitive Science Society (pp. 5244-5250). Edited by L. K. Samuelson, S. L. Frank, M. Toneva, A. Mackey, & E. Hazeltine.
Abstract  Proceedings

Explainable AI (XAI) aims to explain the behavior of opaque AI systems, and in this way, increase their trustworthiness. However, current XAI methods are explanatorily deficient. On the one hand, "top-down" XAI methods allow for global and local prediction, but rarely support targeted internal interventions. On the other hand, "bottom-up" XAI methods may support such interventions, but rarely provide global behavioral predictions. To overcome this limitation, we argue that XAI should follow the lead of cognitive science in developing cognitive models that simultaneously reproduce behavior and support interventions. Indeed, novel methods such as mechanistic interpretability and causal abstraction analysis already reflect cognitive modeling principles that are familiar from the explanation of animal and human intelligence. As these methods might serve as best practices for trustworthy AI, they deserve closer philosophical scrutiny.

Zednik, C. & Boelsen, H. (2022)
Scientific Exploration and Explainable Artificial Intelligence.
Minds and Machines 32: 219-239.
Abstract  Article

Models developed using machine learning are increasingly prevalent in scientific research. At the same time, these models are notoriously opaque. Explainable AI aims to mitigate the impact of opacity by rendering opaque models transparent. More than being just the solution to a problem, however, Explainable AI can also play an invaluable role in scientific exploration. This paper describes how post-hoc analytic techniques from Explainable AI can be used to refine target phenomena in medical science, to identify starting points for future investigations of (potentially) causal relationships, and to generate possible explanations of target phenomena in cognitive science. In this way, this paper describes how Explainable AI—over and above machine learning itself—contributes to the efficiency and scope of data-driven scientific research.

Widdau, C.S. & Zednik, C. (2022)
Opake Systeme künstlicher Intelligenz und das Problem der Verantwortungslücke.
Algorithmen und Autonomie: Interdisziplinäre Perspektiven auf das Verhältnis von Selbstbestimmung und Datenpraktiken (pp. 121-134). Edited by D. Verständig, C. Kast & A. Nürnberger. Verlag Barbara Budrich.
Abstract  Article

Zednik, C. (2021)
Solving the Black Box Problem: A Normative Framework for Explainable Artificial Intelligence.
Philosophy & Technology 34: 265-288.
Abstract  Article  Preprint

Many of the computing systems programmed using Machine Learning are opaque: it is difficult to know why they do what they do or how they work. Explainable Artificial Intelligence aims to develop analytic techniques that render opaque computing systems transparent, but lacks a normative framework with which to evaluate these techniques' explanatory successes. The aim of the present discussion is to develop such a framework, paying particular attention to different stakeholders' distinct explanatory requirements. Building on an analysis of "opacity" from philosophy of science, this framework is modeled after accounts of explanation in cognitive science. The framework distinguishes between the explanation-seeking questions that are likely to be asked by different stakeholders, and specifies the general ways in which these questions should be answered so as to allow these stakeholders to perform their roles in the Machine Learning ecosystem. By applying the normative framework to recently developed techniques such as input heatmapping, feature-detector visualization, and diagnostic classification, it is possible to determine whether and to what extent techniques from Explainable Artificial Intelligence can be used to render opaque computing systems transparent and, thus, whether they can be used to solve the Black Box Problem.

Zednik, C. & Boelsen, H. (2020)
The Exploratory Role of Explainable Artificial Intelligence.
Preprint volume for Philosophy Science Association 27th Biennial Meeting. Chicago, IL: Philosophy of Science Association.
Abstract  Proceedings

Models developed using machine learning (ML) are increasingly prevalent in scientific research. Because many of these models are opaque, techniques from Explainable AI (XAI) have been developed to render them transparent. But XAI is more than just the solution to the problems that opacity poses—it also plays an invaluable exploratory role. In this paper, we demonstrate that current XAI techniques can be used to (1) better understand what an ML model is a model of, (2) engage in causal inference over high-dimensional nonlinear systems, and (3) generate algorithmic-level hypotheses in cognitive science.

Zednik, C. (2019)
Models and Mechanisms in Network Neuroscience.
Philosophical Psychology 32(1): 23-51.
Abstract  Article  Preprint

This paper considers the way mathematical and computational models are used in network neuroscience to deliver mechanistic explanations. Two case studies are considered: Recent work on klinotaxis by Caenorhabditis elegans, and a long-standing research effort on the network basis of schizophrenia in humans. These case studies illustrate the various ways in which network, simulation, and dynamical models contribute to the aim of representing and understanding network mechanisms in the brain, and thus, of delivering mechanistic explanations. After outlining this mechanistic construal of network neuroscience, two concerns are addressed. In response to the concern that functional network models are nonexplanatory, it is argued that functional network models are in fact explanatory mechanism sketches. In response to the concern that models which emphasize a network's organization over its composition do not explain mechanistically, it is argued that this emphasis is both appropriate and consistent with the principles of mechanistic explanation. What emerges is an improved understanding of the ways in which mathematical and computational models are deployed in network neuroscience, as well as an improved conception of mechanistic explanation in general.

Zednik, C. & Jäkel, F. (2018)
Descending Marr's Levels: Standard Observers are no Panacea.
Behavioral and Brain Sciences 41, e249: 43-44 (commentary on target article by D. Rahnev & R.N. Denison).
Abstract  Article  Preprint

According to Marr, explanations of perceptual behavior should address multiple levels of analysis. Rahnev & Denison (R&D) are perhaps overly dismissive of optimality considerations at the computational level. Also, an exclusive reliance on standard observer models may cause neglect of many other plausible hypotheses at the algorithmic level. Therefore, as far as explanation goes, standard observer modeling is no panacea.

Zednik, C. (2018)
Computational Cognitive Neuroscience.
The Routledge Handbook of the Computational Mind (pp. 357-369). Edited by M. Colombo & M. Sprevak. London: Routledge.
Abstract  Chapter  Preprint

This chapter provides an overview of the basic research strategies and analytic techniques deployed in computational cognitive neuroscience. On the one hand, "top-down" (or reverse-engineering) strategies are used to infer, from formal characterizations of behavior and cognition, the computational properties of underlying neural mechanisms. On the other hand, "bottom-up" research strategies are used to identify neural mechanisms and to reconstruct their computational capacities. Both of these strategies rely on experimental techniques familiar from other branches of neuroscience, including functional magnetic resonance imaging, single-cell recording, and electroencephalography. What sets computational cognitive neuroscience apart, however, is the explanatory role of analytic techniques from disciplines as varied as computer science, statistics, machine learning, and mathematical physics. These techniques serve to describe neural mechanisms computationally, but also to drive the process of scientific discovery by influencing which kinds of mechanisms are most likely to be identified. For this reason, understanding the nature and unique appeal of computational cognitive neuroscience requires not just an understanding of the basic research strategies that are involved, but also of the formal methods and tools that are being deployed, including those of probability theory, dynamical systems theory, and graph theory.

Zednik, C. (2018)
Will machine learning yield machine intelligence?
Philosophy and Theory of Artificial Intelligence III (pp. 225-227). Edited by V. Müller. Berlin: SAPERE Springer.
Abstract  Proceedings  Preprint

One way of going beyond the Turing Test is to consider whether Machine Learning (ML) methods will eventually yield algorithms that resemble the ones that are implemented in human brains. Unfortunately, the well-known Black Box Problem makes it difficult to know which algorithms these methods actually yield. Recent attempts to solve this problem within Explainable AI may therefore deliver an a posteriori answer to the philosophical question of machine intelligence. That said, a priori considerations also suggest that eventually, ML-programmed computers will not only behave like humans, but will also acquire algorithms that are similar to the ones that are implemented in human brains.

Zednik, C. (2017)
Mechanisms in Cognitive Science.
The Routledge Handbook of Mechanisms and Mechanical Philosophy (pp. 389-400). Edited by S. Glennan & P. Illari. London: Routledge.
Abstract  Chapter  Preprint

This chapter subsumes David Marr's levels of analysis account of explanation in cognitive science under the framework of mechanistic explanation: Answering the questions that define each one of Marr's three levels is tantamount to describing the component parts and operations of mechanisms, as well as their organization, behavior, and environmental context. By explicating these questions and showing how they are answered in several different cognitive science research programs, this chapter resolves some of the ambiguities that remain in Marr's account, and shows that many different areas and traditions of cognitive scientific research can be unified under the mechanistic framework.

Fink, S.B. & Zednik, C. (2017)
Meeting in the Dark Room: Bayesian Rational Analysis and Hierarchical Predictive Coding.
Philosophy and Predictive Processing. Edited by T. Metzinger & W. Wiese. Frankfurt am Main: MIND Group.
Abstract  Chapter

At least two distinct modeling frameworks contribute to the view that mind and brain are Bayesian: Bayesian Rational Analysis (BRA) and Hierarchical Predictive Coding (HPC). What is the relative contribution of each, and how exactly do they relate? In order to answer this question, we compare the way in which these two modeling frameworks address different levels of analysis within Marr's tripartite conception of explanation in cognitive science. Whereas BRA answers questions at the computational level only, many HPC-theorists answer questions at the computational, algorithmic, and implementational levels simultaneously. Given that all three levels of analysis need to be addressed in order to explain a behavioral or cognitive phenomenon, HPC seems to deliver more complete explanations. Nevertheless, BRA is well-suited for providing a solution to the dark room problem, a major theoretical obstacle for HPC. A combination of the two approaches also combines the benefits of an embodied-externalistic approach to resolving the dark room problem with the idea of a persisting evidentiary border beyond which matters are out of cognitive reach. For this reason, the development of explanations spanning all three Marrian levels within the general Bayesian approach may require combining the BRA and HPC modeling frameworks.

Zednik, C. & Jäkel, F. (2016)
Bayesian reverse-engineering considered as a research strategy for cognitive science.
Synthese 193: 3951-3985.
Abstract  Article  Preprint

Bayesian reverse-engineering is a research strategy for developing three-level explanations of behavior and cognition. Starting from a computational-level analysis of behavior and cognition as optimal probabilistic inference, Bayesian reverse-engineers apply numerous tweaks and heuristics to formulate testable hypotheses at the algorithmic and implementational levels. In so doing, they exploit recent technological advances in Bayesian artificial intelligence, machine learning, and statistics, but also consider established principles from cognitive psychology and neuroscience. Although these tweaks and heuristics are highly pragmatic in character and are often deployed unsystematically, Bayesian reverse-engineering avoids several important worries that have been raised about the explanatory credentials of Bayesian cognitive science: the worry that the lower levels of analysis are being ignored altogether; the challenge that the mathematical models being developed are unfalsifiable; and the charge that the terms 'optimal' and 'rational' have lost their customary normative force. But while Bayesian reverse-engineering is therefore a viable and productive research strategy, it is also no fool-proof recipe for explanatory success.

Zednik, C. (2015)
Heuristics, Descriptions, and the Scope of Mechanistic Explanation.
Explanation in Biology: An Enquiry into the Diversity of Explanatory Patterns in the Life Sciences (pp. 295-318). Edited by C. Malaterre & P-A. Braillard. Dordrecht: Springer.
Abstract  Chapter  Preprint

The philosophical conception of mechanistic explanation is grounded on a limited number of canonical examples. These examples provide an overly narrow view of contemporary scientific practice, because they do not reflect the extent to which the heuristic strategies and descriptive practices that contribute to mechanistic explanation have evolved beyond the well-known methods of decomposition, localization, and pictorial representation. Recent examples from evolutionary robotics and network approaches to biology and neuroscience demonstrate the increasingly important role played by computer simulations and mathematical representations in the epistemic practices of mechanism discovery and mechanism description. These examples also indicate that the scope of mechanistic explanation must be re-examined: With new and increasingly powerful methods of discovery and description comes the possibility of describing mechanisms far more complex than traditionally assumed.

Zednik, C. (2014)
Are Systems Neuroscience Explanations Mechanistic?
Preprint volume for Philosophy Science Association 24th Biennial Meeting. Chicago, IL: Philosophy of Science Association.
Abstract  Proceedings

Whereas most branches of neuroscience are thought to provide mechanistic explanations, systems neuroscience is not. Two reasons are traditionally cited in support of this conclusion. First, systems neuroscientists rarely, if ever, rely on the dual strategies of decomposition and localization. Second, they typically emphasize organizational properties over the properties of individual components. In this paper, I argue that neither reason is conclusive: researchers might rely on alternative strategies for mechanism discovery, and focusing on organization is often appropriate and consistent with the norms of mechanistic explanation. Thus, many explanations in systems neuroscience can also be viewed as mechanistic explanations.

Landy, D., Allen, C. & Zednik, C. (2014)
A perceptual account of symbolic reasoning.
Frontiers in Psychology 5(275): doi: 10.3389/fpsyg.2014.00275.
Abstract  Article

People can be taught to manipulate symbols according to formal mathematical and logical rules. Cognitive scientists have traditionally viewed this capacity—the capacity for symbolic reasoning—as grounded in the ability to internally represent numbers, logical relationships, and mathematical rules in an abstract, amodal fashion. We present an alternative view, portraying symbolic reasoning as a special kind of embodied reasoning in which arithmetic and logical formulae, externally represented as notations, serve as targets for powerful perceptual and sensorimotor systems. Although symbolic reasoning often conforms to abstract mathematical principles, it is typically implemented by perceptual and sensorimotor engagement with concrete environmental structures.

Zednik, C. & Jäkel, F. (2014)
How does Bayesian reverse-engineering work?
Proceedings of the 36th Annual Meeting of the Cognitive Science Society (pp. 666-671). Edited by P. Bello, M. Guarini, M. McShane, & B. Scassellati. Austin, TX: Cognitive Science Society.
Abstract  Proceedings

Bayesian models of cognition and behavior are particularly promising when they are used in reverse-engineering explanations: explanations that descend from the computational level of analysis to the algorithmic and implementation levels. Unfortunately, it remains unclear exactly how Bayesian models constrain and influence these lower levels of analysis. In this paper, we review and reject two widespread views of Bayesian reverse-engineering, and propose an alternative view according to which Bayesian models at the computational level impose pragmatic constraints that facilitate the generation of testable hypotheses at the algorithmic and implementation levels.

Zednik, C. (2011)
The Nature of Dynamical Explanation.
Philosophy of Science 78(2): 238-263.
Abstract  Article  Preprint

The received view of dynamical explanation is that dynamical cognitive science seeks to provide covering-law explanations of cognitive phenomena. By analyzing three prominent examples of dynamicist research, I show that the received view is misleading: some dynamical explanations are mechanistic explanations and in this way resemble computational and connectionist explanations. Interestingly, these dynamical explanations invoke the mathematical framework of dynamical systems theory to describe mechanisms far more complex and distributed than the ones typically considered by philosophers. Therefore, contemporary dynamicist research reveals the need for a more sophisticated account of mechanistic explanation.

Spivey, M., Richardson, D. & Zednik, C. (2010)
Language is spatial, not special: Using space for language and memory.
Spatial Foundations of Cognition and Language (pp. 16-40). Edited by L. Smith, K. Mix, & M. Gasser. Oxford: Oxford University Press.
Abstract  Chapter

Cognitive science has made as much progress as possible with theories of discrete amodal symbolic computation that too coarsely approximate the neural processes underlying cognition. We describe a collection of studies indicating that internal cognitive processes are often constructed of analog spatial formats of representation, not unlike the topographic maps that populate so much of mammalian cortex. These findings point to a view of language in which, far from being a specialized module performing computations on discrete logical symbols, linguistic ability is an emergent property that opportunistically draws from the existing topographic representational formats of perceptual and motor processes.

Zednik, C. (2009)
The Varieties of Dynamicism.
Proceedings of the 31st Annual Meeting of the Cognitive Science Society (pp. 2298-2303). Edited by N.A. Taatgen & H. van Rijn. Austin, TX: Cognitive Science Society.
Abstract  Proceedings

The dynamical approach to cognition is often considered to be 'revolutionary'. In contrast to the well-established approaches of computationalism and connectionism, dynamicism is typically thought to be anti-representational, holistic, phenomenological and law-based. In this paper, I will argue that this way of thinking about dynamicism is too restrictive: it fails to capture the heterogeneous nature of dynamicist research. Although all dynamicist research projects share a commitment to the mathematical methods, tools, and concepts of dynamical systems theory, they frequently disagree with respect to the truth or falsity of representationalism, the role of holistic phenomenological modeling, and in general, the nature of dynamical explanation.

Zednik, C. (2008)
Dynamical Models & Mechanistic Explanations.
Proceedings of the 30th Annual Meeting of the Cognitive Science Society (pp. 1454-1459). Edited by B.C. Love, K. McRae, & V.M. Sloutsky. Austin, TX: Cognitive Science Society.
Abstract  Proceedings

Philosophers of science increasingly believe that cognitivephenomena are explained by describing the mechanisms thatproduce those phenomena, rather than by subsuming them under laws. With respect to the increasingly influential dynamical approach to cognition, however, it has been claimed that only certain kinds of dynamical models (in particular, connectionist dynamical models) satisfy the requirements for mechanistic explanation. This article evaluates this claim. In particular, here it is argued that in the field of evolutionary robotics, several non-connectionist dynamical models have been developed that provide mechanistic explanations of "minimally cognitive" phenomena. Accordingly, the dynamical approach to cognition is conceived as a scientific approach unified by the common employment of a certain body of mathematical tools and concepts, rather than by a common explanatory method.

Upcoming Talks
2024
11.2024	Jülich	Cognitive Explainability for Deep Alignment	Workshop	Philosophy of Deep Learning
11.2024	Eindhoven	Deep Explainability and Deep Alignment	Conference	AI Summit Brainport
09.2024	Bremerhaven	Explanation, Interpretation, and Justification in XAI	Colloquium	Deutsches Zentrum für Luft- und Raumfahrt
07.2024	Rotterdam	Does Explainable AI Need Cognitive Models?	Conference	46th Annual Meeting of the Cognitive Science Society
03.2024	Münster	Explanation, Justification, and Trustworthy AI	Workshop	Ethics of AI (Un)Explainability
03.2024	Eindhoven	Re-Aligning Higher Education in the Age of Generative AI (With Gunter Bombaerts)	Workshop	AI in Education: Ethics and Epistemology
2023
12.2023	Erlangen	Does Explainable AI Need Cognitive Models? (poster presentation with Celine Budding)	Conference	Philosophy of AI 2023
11.2023	Stuttgart	Explanation, Justification, and Responsibility	Keynote	Science and Art of Simulation: Reliability or Trustworthiness
10.2023	Osnabrück	Why Explainable AI Needs Cognitive Models	Keynote	Computational Cognition Conference
10.2023	Umea	Explanation, Justification, and Responsibility	Colloquium	FrAIday at the Center for Transdisciplinary AI
09.2023	Belgrade	Justification and Explainable AI (with Philippe Verreault-Julien)	Conference	Biannual Conference of the European Society for Philosophy of Science
03.2023	Darmstadt	Disentangling XAI Concepts: Explanation, Interpretation, and Justification	Workshop	WhiteBox Symposium on Explainability
2022
12.2022	Köln	Social Integration as Intelligence Attribution	Workshop	GeSiMEx Work-in-Progress Workshop
11.2022	Dresden	Moral Responsibility and Explainable AI	Workshop	Workshop des Schauffler Kollegs
10.2022	Berlin	Toward a Cognitive Science of AI: Methods and Norms	Workshop	GeSiMEx Final Symposium
09.2022	Berlin	Explainable AI: Norms, Standards, and Theories	Workshop	Workshop der Jungen Akademie
07.2022	Hannover	Moral Responsibility and Explainable AI	Workshop	Leibniz Workshop on Digital Ethics
03.2022	Osnabrück	Toward a Cognitive Science of AI	Conference	Cognitive Science Alumni Conference
03.2022	Manchester	Moral Responsibility and Explainable AI	Colloquium	Neurotech Consortium
2021
11.2021	Baltimore	The Exploratory Role of Explainable AI (with Hannes Boelsen)	Conference	Philosophy of Science Association 24th Biennial Meeting
11.2021	Tübingen	Explainable AI as a Tool for Scientific Exploration	Workshop	Philosophy of Science Meets Machine Learning
10.2021	Bern	How should Explainable AI explain?	Colloquium	Center for Artificial Intelligence in Medicine
10.2021	Southampton	How can Explainable AI help Scientific Explanation	Colloquium	Research Network: Artificial Intelligence and Augmented Intelligence for Automated Investigations for Scientific Discovery
07.2021	Hamburg	Moral Responsibility and Explainable AI (with Christoph Widdau)	Conference	International Association for Computing and Philosophy 2021
06.2021	München	Explainable AI and the Problems of Philosophy	Colloquium	Cognition, Values & Behaviour Research Group Meeting
04.2021	Lugano	Explainable AI as a Tool for Scientific Exploration	Colloquium	Philosophy and AI
04.2021	London	Overcoming Opacity in Machine Learning (with Hannes Boelsen)	Conference	Annual Convention of the Society for Artificial Intelligence and the Simulation of Behavior
2020
12.2020	Kassel	25 years after Mind as Motion: What is left of the dynamical revolution? (Online)	Colloquium	Situated Cognition Research Seminar
12.2020	Aachen	Explainable AI and the Problems of Philosophy (Online)	Colloquium	Theory of Science and Technology
11.2020	Bochum	Moralische Verantwortung und erklärbare künstliche Intelligenz (Online)	Workshop	CAIS NRW: Philosophische Digitalisierungsforschung
06.2020	Magdeburg	Über den Begriff der Verantwortung im digitalen Zeitalter (Online, mit Christoph Widdau)	Colloquium	Ringvorlesung: Autonomie im Digitalen Zeitalter
05.2020	Osnabrück	Toward a Cognitive Science of AI (Online)	Colloquium	Colloquium of the Institute of Cognitive Science
01.2020	Warsaw	25 years after Mind as Motion: What is left of the dynamical revolution?	Workshop	Dynamical Models in Neurophenomenology
2019
12.2019	Darmstadt	Toward a Cognitive Science of AI: Aims, Norms and Methods	Colloquium	Cognitive Science Colloquium
11.2019	Leiden	Network Mechanisms in the Brain: Organization and Correlation	Workshop	What is Translation? Exploring the Missing Link Between Neuroscience and Psychiatry
10.2019	Hannover	The Black Box Problem in Artificial Intelligence: Philosophical Reflections on Explainable AI	Colloquium	Kolloquium Philosophie und Wissenschaftsreflexion
09.2019	Geneve	Solving the Black Box Problem: From Machine Learning to Marr (Poster)	Conference	7th Biennial Conference of the European Philosophy of Science Association
08.2019	Köln	Explanation in Cognitive Science: Computation, Algorithm, and Mechanism	Workshop	Computational approaches in language and music cognition research
02.2019	Köln	The Black Box Problem and the Norms of Explainable AI	Conference	3. Kongress der Gesellschaft für Wissenschaftsphilosophie, Symposium: Deep Learning and the Philosophy of AI
2018
12.2018	Köln	Marr's Levels and Deep Neural Networks	Workshop	Causation and Computation in Neuroscience
11.2018	Stuttgart	Opacity, Marr, and the Norms of Explainable AI	Workshop	Science and Art of Simulation: Epistemic opacity in computer simulation and machine learning
10.2018	Magdeburg	Vom maschinellen Lernen zur maschinellen Intelligenz	Colloquium	Themenabend der evangelischen Studierendengemeinschaft
10.2018	Praha	The Black Box Problem and the Norms of Explainable AI	Workshop	Machine Learning and Explanation in Cognitive Science
09.2018	Köln	Will machine learning produce machine intelligence?	Conference	10. Kongress der Gesellschaft für Analytische Philosophie
06.2018	Warszawa	From Machine Learning to Machine Intelligence	Conference	International Association for Computing and Philosophy 2018
06.2018	Magdeburg	Künstliche Intelligenz als Evolutionsstufe: Übermenschlich oder doch menschenähnlich?	Colloquium	Ringvorlesung "Von Hominiden zu Humanoiden"
04.2018	Edinburgh	From Machine Learning to Machine Intelligence	Colloquium	Mind Network Meeting
2017
11.2017	Leeds	From Machine Learning to Machine Intelligence (Poster)	Conference	Philosophy and Theory of Artificial Intelligence 2017
10.2017	Witten	Mechanistic Explanation in Network Neuroscience: A Reply to Objections	Workshop	Models of Mechanisms in Computational Neuroscience
06.2017	Magdeburg	Explanation in cognitive science: Marr and mechanisms	Colloquium	Magdeburg-Augustana Neurophilosophy Summer School
04.2017	Dubrovnik	Mechanistic Explanation in Systems Neuroscience: A Reply to Objections	Conference	Annual Philosophy of Science Conference
2016
06.2016	Warszawa	Explanation in cognitive science: Marr and mechanisms	Workshop	Mechanistic Integration and Unification in Cognitive Science
03.2016	Düsseldorf	Viewing Marr as a Mechanist	Conference	2. Kongress der Gesellschaft für Wissenschaftsphilosophie
01.2016	Witten	Models and Mechanisms in Systems Neuroscience	Colloquium	Philosophy of Science Colloquium
2015
10.2015	Magdeburg	Systems Neuroscience: Models and Mechanisms	Colloquium	Philosophisches Institutskolloquium
2014
11.2014	Chicago	Are explanations in systems neuroscience mechanistic?	Conference	Philosophy of Science Association 24th Biennial Meeting
10.2014	Berlin	Bayesian reverse-engineering considered as a research strategy in cognitive science	Colloquium	Current Issues Seminar, Berlin School of Mind and Brain
09.2014	Köln	Systems, networks, and mechanistic explanations in neuroscience	Workshop	Causality and Complexity in the Sciences
07.2014	Quebec	How does Bayesian reverse-engineering work?	Conference	36th Annual Meeting of the Cognitive Science Society
05.2014	Leiden	Panel discussion: The computational and physical realizability of probabilistic minds and brains	Workshop	Perspectives on Human Probabilistic Inference
05.2014	Leiden	How does Bayesian reverse-engineering work?	Workshop	Perspectives on Human Probabilistic Inference
2013
12.2013	Bochum	Mechanistic explanation in cognitive science: Top-down and bottom-up	Colloquium	Philosophy meets Cognitive Science
08.2013	Helsinki	Are explanations in network neuroscience mechanistic?	Conference	4th Meeting of the European Philosophy of Science Association.
08.2013	Berlin	Bayesian modeling and heuristic strategies for model-development	Conference	35th Annual Meeting of the Cognitive Science Society, Symposium: Constraints on Bayesian Explanation
07.2013	Cortina d'Ampezzo	The role of Bayesian rational analysis in cognitive scientific explanation	Conference	12th Annual Summer Interdisciplinary Conference
05.2013	Magdeburg	Mathematics and Mechanistic Explanations of Mind	Colloquium	Philosophisches Institutskolloquium
03.2013	Hannover	Heuristics of mechanism-discovery and the limits of mechanistic explanation	Conference	1. Kongress der Gesellschaft für Wissenschaftsphilosophi
03.2013	Hannover	Mechanistic explanation and cognitive science	Conference	Netzwerktagung der Alexander von Humboldt Stiftung
2012
12.2012	Tilburg	The heuristic and justificatory role of Bayesian task-analysis in mechanistic explanation	Workshop	Models & Mechanisms
09.2012	Konstanz	From Extended Mechanisms to Extended Minds	Conference	8. Kongress der Gesellschaft für Analytische Philosophie
08.2012	London	Marr's Levels and Mechanistic Explanation: Why ask Why?	Conference	20th Annual Conference of the European Society for Philosophy and Psychology
06.2012	Osnabrück	Dynamical Systems Theory as a Conceptual Framework for Affective Science	Workshop	Giovanna Colombetti's "The Feeling Body: Affective Science meets the Enactive Mind"
2011
10.2011	Osnabrück	Cognitive Science and the Changing Face of Mechanistic Explanation	Colloquium	Kolloquium des Instituts für Kognitionswissenschaft
07.2011	Montreal	Cognitive Mechanisms: Soft-assembled, complex, and large	Conference	37th Annual Meeting of the Society for Philosophy and Psychology
04.2011	Bloomington	The Changing Face of Mechanistic Explanation (Poster)	Colloquium	2011 IGERT Research Showcase
01.2011	Wien	Complex Systems & Mechanistic Explanation	Colloquium	Jour fixe: Simulation komplexer Systeme
2010
08.2010	Bochum	Mental Mechanisms & the Extended Mind	Conference	18th Annual Conference of the European Society for Philosophy and Psychology
04.2010	Bloomington	Situated, Embodied, Dynamical Mechanisms (Poster)	Colloquium	2010 IGERT Research Showcase
2009
12.2009	Bloomington	The Nature of Dynamical Explanation	Colloquium	Indiana University Cognitive Lunch Series
07.2009	Amsterdam	The Varieties of Dynamicism	Conference	31st Annual Meeting of the Cognitive Science Society
06.2009	Bloomington	The Varieties of Dynamicism	Conference	35th Annual Meeting of the Society for Philosophy and psychology
2008
07.2008	Washington, DC	Dynamical Models & Mechanistic Explanations (Poster)	Conference	30th Annual Meeting of the Cognitive Science Society
07.2008	Bloomington	Computation & Dynamics: Alternative frameworks for describing mental mechanisms?	Conference	North American Conference on Computing and Philosophy
06.2008	Philadelphia	Dynamical Models & Mechanistic Explanations	Conference	34th Annual Meeting of the Society for Philosophy and Psychology

Current Teaching
Data Science Ethics	Eindhoven 2021-current	Lecture
Decisions Under Risk & Uncertainty Project Course	Eindhoven 2021-current	Tutorial
Descriptive Decision Theory	Eindhoven 2020-current	Lecture
Philosophy & Ethics of AI	Eindhoven 2021-current	Lecture
Previous Teaching, by Topic
Philosophy and Ethics of Artificial Intelligence
Can Machines Think?*	Magdeburg 2015	B Seminar
Ethics of AI*	Magdeburg 2019	B Seminar"
Philosophy of AI	Magdeburg 2017	B Seminar
Philosophy of AI (Advanced)*	Magdeburg 2018	M Seminar
Philosophy of Mind
Computation and the Brain*	Magdeburg 2016	B Seminar
Embodied Cognition*	Magdeburg 2016, 2019	B Seminar
Introduction to Philosophy of Mind	Osnabrück 2012	B Seminar
—	Osnabrück 2015	Lecture
Philosophical Foundations of Cognitive Science	Indiana 2008	Tutorial
—	Indiana 2009	Lecture
Ryle: The Concept of Mind	Osnabrück 2015	M Seminar
Sellars: Empiricism and the Philosophy of Mind*	Magdeburg 2018	B Seminar
Philosophy of Science
Cognition as a Natural Kind	Magdeburg 2020	B Seminar
Does Psychology Reduce to Neuroscience?	Magdeburg 2016	B Seminar
Explaining Perceptual Choice	Magdeburg 2017	M Seminar"
Explanation in Cognitive Science	Osnabrück 2014	B Seminar"
—	Osnabrück 2015	Lecture"
Philosophy and Ethics of Science	Magdeburg 2018	Lecture
Theory and Methodology of Cognitive Neuroscience*	Magdeburg 2016	M Seminar"
History of Philosophy
British Empiricism*	Magdeburg 2016	B Seminar
Descartes: Meditations*	Magdeburg 2015	B Seminar
Introduction to Theoretical Philosophy*	Magdeburg 2019	Lecture
Wittgenstein: Philosphical Investigations*	Magdeburg 2017	B Seminar
Wittgenstein: Tractatus Logico-Philosophicus*	Magdeburg 2019	B Seminar
Methods Courses
Argumentation Theory*	Magdeburg 2017	B Seminar
Introduction to Formal Logic	Osnabrück 2014	Lecture
—*	Magdeburg 2020	Lecture
Thinking and Reasoning	Indiana 2007	Tutorial

*in German "co-taught

February 2022, Köln
The Nature of Physical Computation
3rd Workshop (10 participants) of the GeSiMEx Project.

April 2021, London
Overcoming Opacity in Machine Learning
Symposium (10 speakers) at the Annual Convention of the Society for the Study of Artificial Intelligence and the Simulation of Behavior.

March 2020, Magdeburg
Machine Learning and Mechanistic Models in Neuroscience
2nd Workshop (6 participants) of the GeSiMEx Project.

March 2019, Köln
Deep Learning and the Philosophy of Artificial Intelligence
GAP-GWP Symposium (4 speakers) at the Congress of the German Philosophy of Science Association.

October 2017, Witten
Models of Mechanisms in Computational Neuroscience
Kick-off Workshop (15 participants) for the DFG-Funded German-Polish International Collaboration Project, Models of Mechanisms in Computational Neuroscience.

Innovation Origins (August 2024): Rethinking intelligence: Is AI making us dumber? (also available in Dutch)

TÜV SÜD (June 2023): WalkShow #7: ChatGPT und wir: Wie KI unser Leben verändert (in German)

Cursor (June 2023): The future of AI: Make the system itself responsible (also available in Dutch)

TÜV SÜD (April 2023): WalkShow #6: Ist die KI wirklich intelligent oder tut sie nur so? (in German)

Current Projects

alignAI: Aligning LLM Technologies with Societal Values
Horizon Europe MSCA Doctoral Network (ca. €300K of €6M)
Consortium Member, August 2024 - July 2028
The alignAI Doctoral Network will train 17 doctoral candidates (DCs) to work in the international and highly interdisciplinary field of LLM research and development. The core of the project focuses on the alignment of LLMs with human values, identifying relevant values and methods for alignment implementation. Funds one PhD student in Eindhoven (current vacancy) to work on a philosophical investigation of XAI methods to promote the explainability of LLMs.
☞ Project Website
☞ Funding Scheme

Generative AI for Academic Writing in Engineering Education
BOOST! & 4TU CEE (€175K)
Principal Investigator, August 2023 - September 2025
An interdisciplinary research project to develop and evaluate teaching activities and assessment methods that incorporate generative AI in engineering education, with a focus on academic writing. Funds one postdoctoral researcher (Manuel Barbosa de Oliveira) to conduct empirical research on the students' use of generative AI in writing-intensive university courses.
☞ Project Website
☞ Funding Scheme

ROBUST: Trustworthy AI-based Systems for Sustainable Growth
NWO Long Term Program (ca. €250K of €87M)
Consortium Member, March 2023 - February 2028
A 10-year, 54-partner consortium spanning academia, industry, and civil society aims to identify and develop talent with the skills to develop robust, reliable, transparent, and trustworthy AI. The project funds one PhD student (Michela Ghezzi) within the AI FAIR lab to work on the evaluation of explainable AI methods to promote trustworthy machine vision in the context of self-driving cars.
☞ Project Website
☞ Funding Scheme

Cognitive Models as Surrogate Models for Explainable AI
EAISI Startup Grant (€250K)
Principal Investigator, August 2021 - July 2025
This project will fund one PhD student (Céline Budding), to investigate the utility of cognitive modeling methods for the purposes of explaining the behavior of "black box" AI systems such as large language models.
☞ Project Website
☞ Funding Scheme

---

Previous Projects

Technical Contributions to ISO/IEC 6254
StandICT Fellowship (€10K)
Fellow, August 2023 - February 2024
The StandICT fellowship allowed me to contribute to the effort to develop an international standard for explainability in artificial intelligence, ultimately resulting in the development of ISO/IEC TS 6254 - Objectives and approaches for explainability and interpretability of ML models and AI systems.
☞ Project Website
☞ Funding Scheme

popAI: A European Positive Sum Approach towards AI tools in support of Law Enforcement and safeguarding privacy and fundamental rights
Horizon 2020 CSA (ca. €40K of €1.6M)
Consortium Member, May 2022 - September 2023
This project aims to evaluate and guide the use of AI in European policing. A multinational consortium investigates current technological state of the art and public perception to issue policy recommendations and ethical guidance.
☞ Project Website
☞ Funding Scheme

Artificial Intelligence - Quality Requirements and Life Cycle Management for AI Modules
DIN SPEC (PAS) 92001
Consortium member, May 2018 - December 2022
This consortium of industry, government, and university representatives works under the guidance of the German Standardization Agency (DIN) to specify norms and practices for the development of artificial intelligence. My own contribution focuses on questions of the explainability of machine learning technologies.
☞ Project website.
☞ Technical Specification (PDF)

GeSiMEx: Model-Development in Neuroscience - Simplicity and Generalizability in Mechanistic Explanations
DFG Sachbeihilfe (ca. €300K of €700K)
Principal Investigator, April 2019 - March 2022
The aim of this project is to investigate the ways in which criteria of simplicity and generalizability are applied in the development of mechanistic explanations in neuroscience. It funds one PhD position in Magdeburg (Hannes Boelsen), as well as a student assistant and a replacement of Jens Harbecke's position in Witten.
☞ Project website
☞ Funding Scheme

ULPEC: Ultra Low-Power Event-Based Camera
Horizon 2020 Action on Smart System Integration
Ethics Advisor, November 2018 - March 2021
The aim of this project is to implement spiking neural networks on neuromorphic hardware for the purposes of building an event-based camera for self-driving cars. My role is to oversee the ethical conduct of scientific research, as well as to advise on the technology's possible social impact with respect to e.g. data privacy and non-civilian use.
☞ Project website

Comprehensability in Machine Learning
neurocat GmbH
Research Strategy Consultant, January 2020 - December 2020
neurocat was developing a product to bring explainable AI tools to market. I was helping them do that.
☞ neurocat website

Models of Mechanisms in Computational Neuroscience
DFG Internationale Kooperationsanbahnung (€18K)
Principal Investigator, September 2017 - December 2018
The aim of this project is to foster collaboration and initiate joint projects between researchers in Magdeburg, Witten, and Warsaw who work on the norms and practices of mechanistic explanation in computational neuroscience. It funds a series of workshops and research visits
☞ Funding Scheme