Schedule for Semester 2, 2025

29 July

Isis Urgell (Monash University)

Vicious BERT: An LLM to identify epistemic vices on polarized social media

There is growing evidence that indicates that individuals with conservative ideology are more likely to believe in conspiracy theories, endorse and spread misinformation and have lower trust on science. The asymmetry thesis proposes that conservatives present distinct cognitive features that makes them more susceptible to politically motivated reasoning and hence to anti science attitudes. In this work I analyse this problem through a virtue/vice epistemology framework. I contain that from the perspective of epistemic vices and virtues there is also an ideological asymmetry on how conservatives build their epistemic networks and gather, process and transmit information as a result. Conservative networks are structurally and individually vicious. Structurally they are organized around malicious agents that preclude the exercise of epistemic virtues while fostering ill epistemic practices. Individually, conservative agents tend to be more epistemically vicious which in turn enables a feedback loop with the structural properties of their networks. To explore these points, I focus on social media behaviour, particularly X (Twitter), with posts related to the recent COVID-19 pandemic. I fine-tuned an LLM from the BERT family to classify tweets according to their degree of epistemic viciousness. I propose then to apply this model to a corpus of tweets from different political communities on Twitter USA. I present evidence obtained from the result of network analysis on these communities to account for the structural claim of conservative vs liberal communities. I propose to use the BERT model to compare the presence of epistemic vices at individual level on conservative networks.

---

5 August

Tyler Paytas (Australian Catholic University)

If it Pleases the Gods: The Socratic Piety of Epictetus

Epictetus’ Stoicism is distinguished by the centrality of Socrates, who serves as both role model and authorizer of the core teachings in the Discourses. A second distinctive feature is Epictetus’ unique conception of piety and the central importance he assigns to this virtue. The primary aim of this paper is to elucidate Epictetian piety and its Socratic origins. Epictetian piety comprises three attitudes inspired by the life and teachings of Socrates: optimism, gratitude, and obedience. After outlining the nature and interrelations of these attitudes, I explain how Epictetian piety can be manifested independent of religious belief. I argue that this form of piety is a neglected virtue that can promote human flourishing and ethical living in a modern secular context.

---

12 August

Wendy Lipworth & Anson Fehross (Macquarie University)

Knowledge-when: Patient perspectives of epistemic authority

The epistemology of Evidence-Based Medicine (EBM) has traditionally privileged two forms of knowing: the propositional ‘knowledge-that’ derived from systematic reviews and trial data, and the clinician’s practical ‘knowledge-how’. While patient values have been appended to this model, they are typically framed as ethical preferences, serving to guide our practical responses. Our qualitative research into experimental stem-cell therapies, however, suggests that some members of the public harbour an alternative view. A sizeable minority of our sample appeal to a third, unrecognised form of purported knowledge. We term this ‘knowledge-when’: a capacity to determine when it the evidence is sufficient to move from clinical trials to common practice. In our interviews, participants grounded this claim not in an understanding of statistical evidence, nor in terms of being willing to accept risk, but in the lived experience of their condition—a form of knowing we call ‘nosomathia’. This represents a claim not just to know when it is worth taking a particular risk, but also when there is evidence of risk. Our presentation will attempt to take this claim seriously, noting that doing so would require reconfiguring EBM’s epistemic architecture to accommodate ‘knowledge-when’ alongside ‘knowledge-that’ and ‘knowledge-how’. This would position the patient as an arbiter of epistemic standards, alongside their expanded role as a part of designing the practicalities of research programmes. As we will show, even if one is not inclined to accept the particular epistemology that serves to undergird their responses, such a view does raise interesting questions as to how evidentiary standards ought to be understood in this domain.

---

19 August

Simine Vazire (University of Melbourne)

Journal prestige can and should be earned

With prominent threats to trust in science around the world, it is especially important to make clear why scientific institutions, including journals, are worthy of trust. Too often, journals’ reputations are unearned—based on flawed metrics such as impact factors, or simply the inertia of prestige. But journal prestige can and should be earned. Journals that invest in and facilitate both pre- and post-publication quality checks, error detection, and correction are the ones that deserve the most trust, and the most prestige.

---

26 August

Adam Hochman (Macquarie University)

Is “Race” Change Possible?

This talk explores the metaphysical possibility of “transraciality.” Most philosophers working on this topic have adopted a social form of what Charles Mills calls objectivism about race. They have argued that transraciality is currently impossible on this basis. I argue that—as Mills himself maintained—social objectivism actually allows for “transraciality,” but that objectivists will be unable to determine the true “race” of many individuals. I compare social objectivist approaches to “transraciality” to the approach entailed by reconstructionist anti-realism about race. For reconstructionists, transraciality is impossible. However, one can racialize oneself differently at different points throughout one’s life, and in different contexts, and some—intentionally or not—end up racialized differently by others. I make the case that, when it comes to the topic of “transraciality,” anti-realist reconstructionism is more explanatory from a theoretical perspective and more desirable from an ethical perspective than social objectivism about race.

---

2 September

Marc Cheong (University of Melbourne)

From (red) pandas* to vibe coding: building software for/by/with philosophers

Software packages to build networks, analyse corpora, and make informative graphics are no longer in the exclusive domain of data scientists. From experimental philosophy, to social epistemology, to philosophy of fame-and-celebrity, philosophers and cross-disciplinary researchers increasingly find utility in these packages. Making software open source is common practice, allowing people to build on existing studies, for the public good. In this talk, I will discuss my experiences building, maintaining, testing, and deploying these packages with colleagues at the Digital Trust Lab. This ranges from the exciting (launching a Python package any philosopher can install in less than 20 seconds), to the mind-numbing (having to create a myriad of test cases and documentation for adding one single piece of functionality!), to the pedantic (not updating a code base without following a series of audits). More importantly, I will discuss challenges and considerations on different approaches to take, tradeoffs between complexity/explainability and accuracy, and the kettle of fish that is Generative AI and “vibe coding”. * pandas is a common Python library, which, sadly, leaves me red in the face every time I use the wrong syntax.

---

9 September

Elena Walsh (University of Wollongong)

Chatbot companions and the new ‘talking cure’

As large language models (LLMs) like ChatGPT become deeply embedded in everyday life, users are increasingly turning to them for emotional support. This talk explores this development. It compares the function of chatbot-based therapies against traditional models of psychotherapy, and situates their appeal in context of a global mental health crisis, unequal access to mental healthcare, rising social isolation, and lives increasingly dominated by virtual modes of interaction. It focuses on a key criterion for therapeutic value: the capacity to balance empathic mirroring with, when necessary, a critical or confrontational stance. While LLMs excel at the former, they appear structurally limited in performing the latter, reflecting a broader and well-documented tendency for LLMs to reinforce the status quo, rather than question it. The talk concludes with initial reflections on the potential benefits and risks of these systems in the context of mental healthcare, the kinds of regulatory oversight they may require, and their capacity to both support and disrupt existing and dominant modes of interpersonal engagement.

---

16 September

Caroline West (University of Sydney)

Happiness as a psychological good

Theories of happiness generally divide into those that analyse ‘happiness’ as a purely descriptive psychological term, and those that take ‘happiness’ to be an evaluative term roughly synonymous with ‘leading a good life.’ I argue that happiness is most accurately and usefully analysed as a fusion of these. Specifically, happiness is the psychological state, whatever it is, that we ought to seek for its own sake. Substantive first-order debates about the nature of happiness are then best understood as implicit disputes about exactly which of a range of semantically eligible states individuals have most prudential reason to pursue. I argue that this may vary (somewhat) from individual to individual and hence that happiness is differentially realizable: the nature of happiness may differ (somewhat) from person to person. This contrasts with orthodox ‘circumscriptionist’ analyses, according to which there is just one way to be happy.

---

23 September

[No meeting - Mid-semester break]

---

30 September

[No meeting - Mid-semester break]

---

7 October

Christopher Lean (Macquarie University)

TBA

TBA

---

14 October

Cristian Larroulet Philippi (University of Melbourne)

Quantifying the human

Quantitative measurement in the human sciences remains both widespread and controversial. Are depression scales, intelligence tests, etc. valid measurement instruments? Do they deliver quantitative or merely ordinal information? I discuss two approaches for understanding practices of quantitative measurement of theoretical attributes. One uses causal notions to characterize dispositional attributes and to understand how they relate to measurement indications. It aims at standard epistemic desiderata in science (discovery, explanation, prediction) and offers good answers to traditional worries about human attributes (namely, are they really quantitative?) and about their measurement instruments (namely, are they valid?). A second approach uses the notion of value(as in Dan Hausman’s 2015 Valuing Health) to make sense of quantification practices. This approach does not resemble what scientists think of their measurement practices: it is not designed for the testing of tentative concepts but rather to standardize political decision making. Yet, I argue, this approach is the most plausible candidate for making sense of some human sciences’ measurement practices as quantifying anything. Such is the case for measurements that (i) combine distinct dimensions of the phenomena at stake and (ii) for which we don’t observe serious efforts aiming at embedding such measurements in predictive and explanatory networks. I illustrate with two examples: depression severity (HAMD) and the Human Development Index (HDI).

---

21 October

Annie Sandrussi (Macquarie University)

Moral Patienthood and the Question of the Polis

TBA

---

28 October

Wendy Higgins (Melbourne University)

Rethinking Psychological Measurement Validity

Valid measurement is critical for empirical research. Yet, a growing body of literature suggests that validation practices in the psychological sciences are highly inadequate. A lack of clarity about the concept of validity itself has been proposed as a contributing factor to poor validation practices. In this talk, I introduce a concept of measurement validity with three key features that I argue can facilitate improved validation practices in the psychological sciences. The first feature of the proposed concept of validity is that it is restricted to the question of what a test measures. This contrasts with influential guidelines from the American Psychological Association, which state that validity should be ascribed to interpretations of test scores for different test uses. The second feature is that it is based on a necessary, and in principle, sufficient causal condition for valid measurement; and the third feature is that it makes an explicit distinction between the validity potential of measurement procedures in abstracta (e.g., tests) and the realised validity of concrete measurement outcomes (e.g., specific sets of test scores). Considering each feature in turn, I describe the potential benefits and challenges of adopting this concept of validity in the context of psychological measurement.

4 November

Joanne Faulkner (Macquarie University)

TBA

TBA