Trust in Epistemic Networks (14-15 Nov)

Description

This event will explore the nature and importance of trust in the transmission of information within epistemic networks, particularly in contexts where non-experts rely on experts for knowledge, as in the case of scientific communication. Bringing together a range of interdisciplinary perspectives (from social and formal epistemology, cognitive science, data science, or anthropology), the workshop will examine the mechanisms by which trust is established, maintained, and undermined in epistemic networks, especially in digital environments where misinformation, filter bubbles, and echo chambers are prevalent. By investigating how trust operates across various types of epistemic networks, we aim to deepen our understanding of how these networks can either support or hinder informed decision-making. The discussions will focus on strategies for building more robust and trustworthy epistemic environments that resist the epistemic harms so prevalent today.

Location

This is a hybrid event (all talks are also on Zoom).

In person, at Makerspace M2, Zellescher Weg 17, 01069 Dresden. Here is the room location.

Digitally, via Zoom, upon registration.

Program

Day 1: Thursday, 14th November 2024

09:00 – 09:30 > Welcome Coffee (30 min)

09:30 – 10:40 > Session 1 (40 min talk + 30 min Q&A) Erik Olsson (Lund University) – Learning from Ranters: The Effect of Information Resistance on the Epistemic Quality of Social Network Deliberation

10:40 – 11:10 > Coffee Break (30 min)

11:10 – 12:20 > Session 2 (40 min talk + 30 min Q&A) Ulrike Hahn (Birkbeck, University of London) – What Rational Agent Based Models Tell Us About Social Networks (via ZOOM) 

12:20 – 13:20 > Lunch Break (60 min)

13:20 – 14:30 > Session 3 (40 min talk + 30 min Q&A) Patricia Rich (University of Bayreuth) and Sergiu Spatan (TU Dresden) – Trust Updating in Strongly Tied Networks

14:30 – 15:00 > Coffee Break (30 min)

15:00 – 16:10 > Session 4 (40 min talk + 30 min Q&A) Carline Klijnman (University College Dublin) – Who Knows? The Epistemic Crisis as Public Credibility Dysfunction

16:10 – 16:40 > Coffee Break (30 min)

16:40 – 17:50 > Session 5 (40 min talk + 30 min Q&A) Emily Sholfield (Johns Hopkins University) – Science Communication for Earning Thick Public Trust (via ZOOM)

19:00 – 21:00 > Dinner at L'Osteria (ATTENTION, at Wilsdurffer Straße 14)

.............................................................

Day 2: Friday, 15th November 2024

09:00 – 09:30 > Morning Coffee (30 min)

09:30 – 10:40 > Session 6 (40 min talk + 30 min Q&A) Silvia Fierăscu (West University of Timișoara) – Building Resilient Communities through Transfer of Trust: The Case of Epistemic Networks in Culture (via ZOOM)

10:40 – 11:10 > Coffee Break (30 min)

11:10 – 12:20 > Session 7 (40 min talk + 30 min Q&A) Radu Umbreș (SNSPA, Bucharest) – From Familial Benevolence to Partnership Competence. The Cultural Evolution of Interpersonal Trust in a Romanian Village

12:20 – 13:20 > Lunch Break (60 min)

13:20 – 14:30 > Session 8 (40 min talk + 30 min Q&A) Hein Duijf (Utrecht University) – Towards Empirical Robustness in Network Epistemology

14:30 – 15:00 > Coffee Break (30 min)

15:00 – 16:10 > Session 9 (40 min talk + 30 min Q&A) Laura Engel (University of Hamburg) – Group Deliberation in Source Networks 

16:10 – 16:40 > Coffee Break (30 min)

16:40 – 17:50 > Session 10 (40 min talk + 30 min Q&A) Rico Hauswald (TU Dresden) – Epistemic Networks: Which Ones Should You Trust?

19:00 – 21:00 > Dinner

Abstracts

Session 1. Erik Olsson – Learning from Ranters: The Effect of Information Resistance on the Epistemic Quality of Social Network Deliberation

People who spread misinformation in public debates expose others to the risk of forming false beliefs. Excluding them from participation can limit this exposure, but fact-checking takes up resources of time and money, and censorship violates social and political norms. Here, computer simulations of Bayesian learning in social networks suggest that, in some contexts anyway, the epistemic benefits of excluding sources of misinformation might be small or nonexistent, and not worth associated costs. It is shown more specifically that, under certain conditions, open-minded agents in a network can learn just as well in the presence of false ranters: information resistant agents that repeatedly broadcast falsity within the network. Relevant conditions are that the open-minded agents can keep track of their social sources and maintain appropriate levels of trust in them, and that some sufficiently reliable sources introduce truth into the network.

Session 2. Ulrike Hahn – What Rational Agent Based Models Tell Us About Social Networks

TBA

Session 3. Patricia Rich and Sergiu Spatan – Trust Updating in Strongly Tied Networks

In this talk, we present interim results from our modeling of trust-updating in epistemic networks given the relational strength of the ties within those networks. In previous work, we explored how people’s propensity to holding bad beliefs – i.e. beliefs contradicted by clear, strong evidence – might be influenced by their reliance on unreliable strong-tie contacts (family members, friends etc.). People’s strong-tie contacts, we claim, are instrumental in determining the epistemic position of an individual within a network. Our proposal opposes the account of Emily Sullivan and colleagues, who analyze an individual’s epistemic position within a network based on the number, independence, and diversity of their information sources. Our contention is that an individual's epistemic position within a network on a topic, t, should instead be defined in terms of (i) the reliability of their trusted sources on t and (ii) the reliability of their higher-order informants in tracking reliable sources on t, which quite often are represented by their strong-tie contacts. In this presentation, we offer interim results of the comparison of the two frameworks using agent based modeling. Moreover, we draw analogies between our proposal and the existing trust-updating framework developed by Erik Olsson, Ulrike Hahn, and others, by looking at how our model predicts belief polarization. We end with remarks about future work for further developing our framework.

Session 4. Carline Klijnman – Who Knows? The Epistemic Crisis as Public Credibility Dysfunction

We appear to live in an ‘epistemic crisis', characterized by epistemic bubbles and echo chambers, fake news, conspiracy theories, and other forms of misinformation and disinformation. The literature tends to equate the epistemic harms of this crisis with false beliefs and ignorance, i.e. a lack of true justified beliefs. I argue that there is something more fundamentally at stake in the epistemic crisis that affects alternative criteria for knowledge. This perspective emerges when we analyse the epistemic crisis as a dysfunction in testimonial practices.

The epistemic quality of our testimonial-based beliefs depends in part on features of our epistemic environment. Over the last few decades, developments in communication and information technologies have drastically changed the way we receive, consume and share information - especially the introduction of the internet, the smart phone, social media, and more recently generative AI.  As I will illustrate, features of this online epistemic environment exacerbate existing problems and introduce additional challenges for gathering knowledge through testimony. I argue that various epistemic challenges identified in the epistemic crisis literature can be understood as factors that disrupt important social mechanisms of credibility appraisals. I term the resulting state in which we can no longer reliably count on social credibility clues for locating trustworthy testifiers Public Credibility Dysfunction. I further argue that Public Credibility Dysfunction undermines our ability to gather genuine knowledge, even if we can still gather true justified beliefs, because it undermines alternative conditions for knowledge such as reliability.  Importantly, conceptualizing the epistemic crisis as a failure of social mechanisms for credibly appraisals, underpins how it is at its core a social problem. Accordingly, effective remedies need to include concern for this social dimension and focus on testimonial trust relations.

Session 5. Emily Sholfield – Science Communication for Earning Thick Public Trust

Increasing political polarization, emerging observable impacts of climate change, the Covid-19 pandemic, and the rapid rise of computing technologies have built a world in which people are forced to rely upon technology and research which they are not capable of fully understanding on their own, without specialized training. This places everyone, including scientists outside of their disciplines, in a position of trusting the structures, standards, and expertise of scientific institutions with central and intimate parts of their lives. I will argue that this calls upon scientific institutions to make active commitments to science communication aimed at earning the public’s trust in a respectful, meaningful, resilient way. This is not only because the work done by scientific institutions is typically trustworthy and essential to forward progress — which I do believe — but also because the lay public deserves to be able to sincerely trust the institutions producing the information that influences so much of their daily lives.

In this presentation, I will utilize philosophical work on trust to develop a new model of science communication, aimed at earning meaningful, resilient public trust in scientific institutions. I will first evaluate current models of science communication. Second, I will review different forms of trust, introducing my own concept of thick institutional trust. Third, I will evaluate the benefits — for both scientists and the public — of science earning the public’s trust. Finally, I will propose a new model of science communication: informative public engagement, which aims to earn the public’s trust in a meaningful, resilient way by prioritizing epistemic respect and bidirectional engagement between science and the public.

Session 6. Silvia Fierăscu - Building Resilient Communities through Transfer of Trust: The Case of Epistemic Networks in Culture

This presentation explores the role of epistemic communities in building resilient cultural ecosystems, focusing on the experience of the city of Timisoara, European Capital of Culture (ECoC) in 2023. Highlighting the design, implementation, and evaluation phases of the cultural program, I outline a mixed-methods approach that integrates quantitative, qualitative, and participatory data to understand the network dynamics underpinning the cultural landscape formed around the title. This approach reveals the mechanisms and principles necessary for sustainable, trust-based collaborations among cultural stakeholders and decision-makers, contributing to a robust legacy program for 2024 and beyond. By examining these dynamics, I offer insights into how epistemic networks in culture can foster resilience, cohesion, and long-term urban sustainability, with implications for other cultural and policy initiatives in Europe.

Session 7. Radu Umbreș – From Familial Benevolence to Partnership Competence. The Cultural Evolution of Interpersonal Trust in a Romanian Village

Analyses of trust often distinguish between evaluations of the benevolence vs the competence of others. Based on ethnographic fieldwork in Sateni, an NE Romanian village, I present how and why these two facets of trust have changed over the past century in a peasant society shaped by communism and the transition to market economy and democracy. Over time, moral trust in family and relatives was paramount and remained much higher than in generic individuals and political leaders. Social alliances based on kinship, friendship or group identity dynamically responded to key moments such as forced collectivisation in the 1950s or the blundered restoration of property rights in the 1990s. Meanwhile, the importance of epistemic trust increased in response to long-term processes such as increased division of labour, education, urbanisation, and integration in the bureaucratic nation-state. This paper argues that trust evolves in response to the scale and scope of cooperation in various forms of social organisation, from family structures and economic modes of production to local and national politics. Evolved mental mechanisms for epistemic vigilance and signalling intersect with the social ecology of knowledge in Sateni, leading to enduring cultural representations of suspicion and secrecy.

Session 8. Hein Duijf – Towards Empirical Robustness in Network Epistemology

One of the central papers in simulation studies of science argues that less communication often leads to higher reliability. More generally, simulation studies have been used to explore which communication networks enhance group reliability. However, this literature has largely concentrated on relatively simple network structures (e.g., cycles, wheels, full graphs), which bear little resemblance to real social networks. Does less communication often lead to higher reliability also in real social networks? In this talk, we provide some results concerning the empirical robustness of these results with respect to real social networks.

We develop novel methods to perform this empirical robustness analysis. First, we use citation graphs to approximate real social networks commonly discussed in the literature as examples of lagging discovery—i.e., one concerning the bacterial causes of peptic ulcers and another concerning the prolonged history of the perceptron. Second, we develop methods to generate a sample of "similar" networks. Third, we work out the group reliability of these communication networks by running simulations. Finally, we analyse the simulation data to determine which outcomes can be expected in real networks and which network properties (e.g., connectivity, degree inequality) strongly affect group reliability. 

Session 9. Laura Engel – Group Deliberation in Source Networks 

Democratic group deliberation is generally taken to possess epistemic qualities that improve collective decision-making outcomes. One mechanism by which deliberation is argued to do so is by neutralizing epistemic effects of over- or undercounting of unequally distributed evidence. However, deliberation can only fulfill this promise if it does not itself lead to overcounting of evidence that is held by multiple deliberators. I argue that to avoid overcounting, it is necessary that individuals are aware of the dependency structures between their evidence. However, this condition cannot generally be assumed to be satisfied. On the contrary, most deliberators get informed in networked epistemic systems whose structures are often opaque and untraceable for the individual. I therefore analyse the epistemic quality of group deliberation in a model that assumes epistemic source networks. Within this framework, it can be shown that while deliberation tends to produce correct beliefs, simple majority voting without deliberation performs equally well in a majority of cases. On top of that, deliberation under source correlations can distort individual credences, such that on average, individuals tend to hold credences closer to the evidentially justified credence before group deliberation than after. 

Session 10. Rico Hauswald – Epistemic Networks: Which Ones Should You Trust?

Networks of experts can be epistemic sources for laypeople, just as individual experts can be. In many cases, they are even superior sources. But what does it mean to epistemically trust a network? And what are the challenges that laypeople face in doing so? This talk proposes an account according to which trusting a network with respect to a proposition p amounts to taking some appropriate fact about the network (such as the fact that its members have reached a consensus about p) as a proxy for the truth of p. This poses two challenges for laypeople: determining which fact (among possible others) is the best (or at least a sufficiently good) proxy, and determining whether the fact in question actually obtains in a particular case. Two additional challenges arise from the observation that networks vary widely in their epistemic trustworthiness. For example, consensus in one network may be a good proxy for the truth of a proposition, while consensus in another network may be a bad proxy. This raises the question of what characteristics make a network epistemically trustworthy. And it raises the question of how laypeople can determine whether a network exhibits these characteristics.

Organization

Organization team:

• Moritz Schulz
• Sergiu Spatan

Part of the project „Gefangen im Netz? Chancen und Perspektiven von epistemischer Netzwerkanalyse“, financed by Sächsische Aufbaubank, and based at Technische Universität Dresden.