Philosophy and History of Science Colloquium – Winter Semester 2025/26 (Hybrid)
We are pleased to announce the Frontiers in Philosophy and History of Science and Technology Research Colloquium for the Winter Semester 2025/26.
📍 Time: Tuesdays, 15:30–17:15 (CET)
📍 Venue: 1st Floor, Augustenstrasse 40, Munich
📍 Format: Hybrid (in person and via Zoom – the link will be provided upon registration)
If you plan to attend in person, please let Oksana Bondar know in advance (oksana.bondar@tum.de) so that building access can be arranged.
To register for the seminar and to receive the Zoom link, please email desantila.hysa@tum.de.
Programme:
• 14 October – Online Workshop: Philosophy of Science in Practice
• 21 October – Dr. Aysel Görkan (EPSA Fellow, Turkey) & Nathanael Sheehan (TUM/Exeter University)
• 28 October – Luis Lopez (LMU)
• 04 November – Emma Cavazzoni (TUM)
• 11 November – David Colaço (LMU)
• 18 November – Hugh Williamson (TUM/Exeter University)
• 25 November – Maria Volkova (Exeter University)
• 02 December – Sven Nyholm (LMU)
• 09 December – Silvia Milano (TUM)
• 16 December – Paolo Leone (Nova SBE-Barcelona)
• 13 January – Rena Alcalay (TUM)
• 20 January – Michael Klenk (University of Delft)
• 27 January – Hybrid Workshop on Democracy and Expertise
• 03 February – Joyce Koranteng-Acquah (TUM)
21 OCTOBER Aysel Görkan | EPSA Fellow, Turkey
Science–Society Interaction and the Problem of Levels of
Selection
In this study, I aim to clarify the relationship between certain misconceptions about
evolution and their reflection in society. Through a theoretical and epistemological
analysis, I examine the dynamic interplay between science and society. I focus
particularly on reductionist approaches in biology, such as typological thinking and
gene-centrism. These approaches not only shape scientific debates on the levels
of selection but also influence how evolutionary concepts are understood
and utilized in broader social contexts. In particular, I argue that gene-centrism
and reductionism generate significant epistemological challenges: these
approaches reinforce the deterministic perspective that grants causal primacy to
genes. Such approaches contribute to both the emergence of limitations in
scientific practice and the oversimplification of the evolutionary role of other levels
of organization. By addressing these issues, I aim to shed light on the
epistemological problems that arise in clarifying and understanding scientific
concepts related to evolutionary theory.21 OCTOBER Nathanael Sheehan | TUM/ Exeter University
Methodologies: Towards a Situated Metascience
This chapter sets out the methodological commitments of the thesis by
positioning them in relation to, and in critique of, contemporary metascience.
Current approaches to metascience typically cluster around three domains: open
science, the science of science, and methodological activism. I begin with a
critical review of this landscape, highlighting in particular the tendency of
metascience to neglect insights from the history, philosophy, and sociology of
science (HPS). In response, I develop an alternative research design, which I
call situated metascience. This design reorients methodological attention toward
openness as situated practice, scientific activism as a form of care, and HPS as a
critical methodological resource. Building on this research design, I then specify
the topics, methods, and values that ground the present thesis. The chapter
concludes by showing how situated metascience is operationalized through three
interlinked methods drawn from philosophy of science in practice: case
studies that situate inquiry in concrete research contexts; concept cartography as
a tool for mapping philosophical and infrastructural commitments; and reflexive
analysis of the research process itself.28 OCTOBER Luis Lopez | LMU
Making Translational (Mis)alignment Auditable
I propose a formal framework for assessing translational alignment in biomedicine.
This framework takes as its conceptual starting point Lara Keuck’s notion of scope
validity, defined as the matching between the target as operationalized in
experimental settings and in application contexts. The problem I address is how to
make those operationalizations—and their matching—explicit, open to scrutiny,
and amenable to computation. I do so by representing each practice with formally
constrained diagrammatic specifications of types, functional relations, and
declared equivalences, and by relating them through a common-ground schema
that records alignment maps for measurements and target-defining features.
Measurement claims are captured in a model-based manner (after Tal), detailing
what is measured, how it is produced, and the calibration and uncertainty
involved. This turns comparability into an explicit, auditable mapping rather than a
tacit analogy, supports simple quantitative indicators of matching, and does so
without erasing local context.04 NOVEMBER Emma Cavazzoni | TUM
Bugs and Pears: Data Models Discriminating among Differences
beyond Statistics
This presentation examines the reasoning behind selecting biological parameters
and mathematical variables for data models, as well as the considerations and
dynamics shaping their construction. Challenging traditional accounts that
interpret data models solely in statistical terms, we argue that data models in
biology cannot be separated from what Suppes (1962) calls ceteris
partibus conditions; and that such conditions include not only experimental
settings for data collection and analysis but also computational constraints of
large-scale AI systems. We ground our reflections on two case studies from the
agricultural project Haly.Id: pear classification via tissue differentiation and insect
monitoring through drones. In the first part of the presentation, we advance four
related claims: 1) biological data models involve processes of differentiation and
identification; 2) these processes are interdependent, though researchers’ roles
differ; 3) each phase produces a distinct model contributing to the final one; 4)
model development is constrained by biological and technical factors. In the
second part, we show how reliance on complex and partly black-boxed
technologies like those common in the age of genAI may facilitate some modelling
tasks but strongly limit researchers’ ability to adapt models to their specific goals.11 NOVEMBER David Colaço | LMU
The Generative Conceptual Conflict in Science
The past decade has seen new philosophical explorations of scientific concepts,
the building blocks of scientific theorization, with accounts from Feest, Arabatzis,
and Haueis. While most philosophers accept that concepts can proliferate and
change, what remains unaddressed is how we account for conceptual conflict.
This occurs when there is disagreement over the intension and extension of a
concept, resulting in inconsistent conceptualizations of the same ostensible
targets of investigation. In this talk, I account for conceptual conflict, focusing on
cases in cognitive science. I address how this conflict can be generative,
informing new conceptual and empirical advances. My account rests on the posit
that concepts can be treated as conjectures. They are hypotheses that pick out
phenomena in their extensional spaces, where these phenomena serve as data
against which we test these hypotheses. This transforms conceptual conflict into a
form of rival hypothesis testing.18 NOVEMBER Hugh Williamson | TUM
The Field of Indicators: Quantitative Genetic Repertoires in
Animal and Plant Breeding
Quantitative genetics is a statistical approach to genetics, distinct from classical
Mendelian and molecular genetics, that has shaped and been shaped by attempts
to intervene in the biology of agricultural plants and animals at the population level
throughout the twentieth and twenty-first centuries. This seminar explores the
‘repertoires’ (Ankeny and Leonelli 2016) of materials, skills and practices that
comprise quantitative genetics in plant and animal breeding, focusing in particular
on practices of producing and using statistical indicators in breeding programmes
(namely heritability, breeding values, and genetic gain). I will analyse three critical
issues that arise from the use of quantitative genetic indicators in breeding and
agriculture: 1. Contestations over expertise, between data-driven approaches and
skilled judgement; 2. Indicators as tools of biopower, especially over animals; and
3. The ambiguous epistemic and political role of environmental factors in
quantitative genetics. I will also show how the movement of particular repertoires
back and forth between animal breeding and plant breeding has brought these
different issues into prominence.25 NOVEMBER Maria Volkova | Exeter University
Making up marriage: officials and AI checking the genuineness
of the relationship
This paper presents findings from an ethnographic study of how the UK state
seeks to detect ‘sham’ (fraudulent) marriages through immigration control. It
advances two central findings. First, states are often assumed to govern through
clear-cut categories that define who people are and how they should act. But in
the case of ‘sham’ marriage detection, the categories used by the Home Office are
anything but clear. Precisely because couples cannot know what counts as a
‘genuine relationship’, they go to great lengths to present themselves in ways that
pre-empt possible doubts. Many turn to online communities, where they
collectively develop shared norms and informal conventions about how to appear
as a ‘real’ couple in the eyes of the state. In this way, ambiguous classifications
shape couples’ self-presentation. Second, although automation and AI are often
expected to reduce human judgment, my findings suggest the opposite. In a
context of opaque infrastructures and vague criteria, these technologies can
intensify reliance on intuition. Frontline officials reported feeling more confident in
flagging couples based on ‘gut feeling’, assuming that any error would be
corrected by an objective system, unaware that the system itself draws on their
own discretionary inputs.02 DECEMBER Sven Nyholm | LMU
The Ethics (and History) of Defining Artificial Intelligence
I will be discussing whether it matters – and if so, why it matters – how we define
artificial intelligence. Like other controversial concepts with a history, like freedom
or equality, the concept of artificial intelligence is an evolving concept, and new
definitions of this idea are suggested on a fairly regular basis. Moreover, according
to some authors, how we should define artificial intelligence is not just a purely
descriptive question of conceptual analysis or semantics – instead, there are
ethical reasons for favoring certain definitions over others. To approach this topic
in at least a somewhat systematic way, I will start with an incomplete history of
attempts to define artificial intelligence. I will then ask whether we should choose
one definition or whether we should take an “inclusive” approach that
incorporates elements from several of the suggested definitions we will consider. I
will argue that we should adopt what I will call a “broadly inclusive” definition of
artificial intelligence, since, among other reasons, this helps to explain why the
notion of artificial intelligence is not only philosophically interesting, but also a
topic that raises many different kinds of ethical questions.16 DECEMBER Paolo Leone | NOVA SBE, Lisbon
Decentralizing science: Market and commons pathways
The organization of science is being reconfigured by the emergence of initiatives
that decentralize the governance of scientific knowledge production, evaluation,
and dissemination. These initiatives respond to growing concerns about the
centralization of authority in publishers and journals, which can impede the
efficient production of scientific knowledge by limiting the disclosure and reuse of
data and research materials, create imbalances in evaluation by relying heavily on
the labor of uncompensated reviewers, and constrain innovation in scholarly
dissemination by preserving outdated publication formats. Drawing on qualitative
methods, including in-depth observations, interviews, and document analysis, this
paper investigates two such initiatives—ResearchHub and Evidence—to examine
how they enact decentralization and with what implications. The analysis shows
that these initiatives developed distinct governance mechanisms, which underpin
different governance forms. ResearchHub introduced a “market-based”
governance system in which reviewers are compensated with research coins that
can be exchanged for currency or used to request preprint reviews, pose
specialized questions, or crowdfund research proposals. This token-based model
aligns effort with reward, addressing exploitative dynamics in traditional evaluation
systems. In contrast, Evidence fostered a “commons-based” governance system
built around an open research ecosystem that supports in-depth engagement with
scientific work, allowing scientists to reuse data and materials, interrogate figures,
and reproduce analyses before and irrespective of journal publication. By
comparing these cases, the paper theorizes alternative governance systems for
decentralized science, clarifying the trade-offs between market-based and
commons-based approaches.13 JANUARY Rena Alcalay | TUM
The Violence of Knowing: Epistemic Harm as a Condition, Not
an Exception
I will examine how epistemic harm evolves not only from flawed ethical or political
systems, but also from within institutional vulnerabilities—what might be described
as an epistemic ‘state of nature’ in the Hobbesian sense. Imagining even minimally
structured environments, individuals may remain vulnerable to misrecognition,
exclusion, and exploitation, despite the presence of shared norms, formal
protections, or epistemic justice frameworks. In such a state, harms are not merely
deliberate or accidental, intentional or incidental; they are constitutive of the very
condition under which knowledge is produced. By exploring this idea alongside
questions such as whether all harms are ethically valanced, I seek to contribute to
a more robust and philosophically grounded taxonomy of epistemic harm.27 JANUARY Richard Williams | TUM, Hybrid Workshop on
Democracy and Expertise
The Feasibility Power of Experts
In the philosophy of science, the values in science debate largely explores how
the value judgments of experts may influence politics in unacceptable ways. In
contrast, I will foreground the underexplored “feasibility-power” of experts. In
politics, experts often make feasibility judgments about what is possible and what
is necessary. In practice, methodological choices rather than moral choices often
shape the feasibility judgments of experts. So, I will argue that methodological
pluralism is a critical check on the fallible feasibility judgments of experts.
However, I will argue that no particular individual or institution must practice
methodological pluralism. In practice, individuals inside institutions may need
consensus on methodological questions. On a much bigger scale, I will argue that
the research ecosystem as a whole should cultivate methodological pluralism. In
practice, different institutions should seek contestation on methodological
questions as a critical check on each other.