What counts as scientific understanding in cognitive science?

What counts as scientific understanding in cognitive science is a surprisingly unsettled and unexplored question. Its relevance is amplified not only by the advances in artificial intelligence, neuroscience, but also by the rise of interdisciplinary research (Weston et al., 2020) and the expansion of interdisciplinary programs in higher education (Rana et.al.2025). Although the central question of the conference targets cognitive science especially, we cover interdisciplinary science as well.

Competing scientific frameworks (symbolic, connectionist, representationalist, enactive, embodied and others) usually offer divergent accounts of “cognition”. This makes the issue of understanding more pressing now that AI models produce outputs that suggest passing False Belief Tasks, Empathy scales, being used for therapy (therapy bots), solving scientific problems, echoing classic philosophical challenges and renewing debates about whether performance reflects genuine understanding. Here are the conference panels with the main topics:

Scientific understanding in cognitive science

• Can cognitive science produce a scientific explanation of understanding? Do we expect a mechanistic one, a computational (representational) one, one based on dynamic systems, or should one better aim for a Bayesian approach?
• Do we need a unifying theory about the mind in cognitive science, or should we settle for pluralism (at the level of explanations, models and scientific practices)?
• What does a grand unified theory in cognitive science afford that pluralism does not
• Should we strive for a unified explanation (one that should account for cognitive, neural, phenomenological and behavioral aspects alike)?
• Do we want integration at the level of explanations? Should we also integrate at the level of models (is that even possible)? Do we need explanations or models to account for phenomenal aspects of understanding? If so, what does explanatory pluralism imply for the phenomenology of understanding? What use for a unifying theory when finer-grained, multilevel, partial analyses might be more explanatory? 
• What epistemic desiderata do cognitive-scientific models meet - approximate truth, explanatory or predictive power, simplicity, empirical adequacy, others? Which such desiderata matter more in which cognitive-scientific contexts?
• If different ensuing models impact different branches of cognitive science differently, how does this bear on the field's interdisciplinary unity?

Scientific understanding and interdisciplinarity

• Does integrating multiple levels of analysis require new forms of explanation and, if so, which? What are the limits of integration? Is integration desirable whenever achievable?
• What roles do models play in interdisciplinary understanding? How do these models function when integrating assumptions from multiple domains (with different ontologies)?
• How can experts communicate their understanding to an audience of non-experts? Does “translation” between multiple disciplines affect understanding? Are there aspects or nuances/features that get lost or transformed when concepts “migrate” between fields?
• If interactional expertise is required for interdisciplinary approaches, does it grant contributory abilities? Is it sufficient for researchers in an interdisciplinary community of experts to be spectatorial cognizers? Is scientific understanding something individuals possess when part of an interdisciplinary effort, or is understanding distributed across research teams, maybe even split between specific research fields?
• Are epistemic standards transferable between fields in interdisciplinary studies, or are they  bound to specific fields?
• Does interdisciplinary research require new epistemic virtues (tolerance for ambiguity, transferable and translatable knowledge) or norms?
• Can understanding at one level of analysis substitute for another level of analysis? If so, in what circumstances?
• What is the epistemic value of interdisciplinarity? Does combining models from multiple fields increase scientific understanding,  or does it sometimes obscure it?

Benchmarking scientific understanding

• How can scientific understanding be operationalized?
• Is (scientific) understanding (just) a peak performance? Can we benchmark (scientific) understanding and if so, should we include AI systems? If AI systems understand, does AI understanding bear on how we conceive of human understanding?
• What distinguishes understanding from mere predictive success?
• What role does explainability play in benchmarking?
• Can human and AI understanding be compared? If any, which shared metrics would apply across biological and artificial entities?
• Can interdisciplinary scientific understanding be benchmarked? How could it be evaluated?
• Do different models strike different trade-offs? If so, how does impact benchmarking model-based understanding? 

 The conference will take place on the 22nd, 23rd and 24th of May, 2026 at the University of Bucharest, Faculty of Philosophy. Regular presentations will be 30 minutes long, followed by 20 minutes Q&A.