Simulation and thought experiment
8-9 June 2017, University of Geneva
Uni Bastions – room B214
Rue de Candolle 5, 1205 Genève

    • Margherita Arcangeli, Humboldt Universität zu Berlin

Mental simulation as the hidden link between real, thought and numerical experiments

Several parallelisms between thought experiments and computer simulations, or numerical experiments, can be drawn and a comparative analysis can shed light on both topics at once. Some authors have commented in passing on the parallelism between thought experimentation and numerical experimentation, while others have suggested that numerical experiments can be seen as a type of thought experiments and will even replace the latter. But the “trading zone” between thought experiments and numerical experiments has been sparsely considered by current works on either thought or numerical experiments. Most works have primarily focused their attention on the links between these two scientific tools, and real experiments. Moreover, it has been suggested that the discussions on thought experiments and numerical experiments are limited, because they pay too much attention to the outputs of these scientific tools, more than to their processes. Mental simulation has been seen as a key process that would show that between thought experiments and numerical experiments there is an essential similarity, to such a degree as to allow the latter to replace the former. More should be said about the relevant notion of mental simulation, however. The literature in philosophy of mind recognizes at least two different types of mental simulation, namely mental models and imaginings. The aim of this paper is to deepen our understanding of the relationships between real experiments, thought experiments and numerical experiments by means of an analysis of the role played by mental simulation, in both its guises, in their processes. It will emerge that thought experiments and numerical experiments stand in a closer relationship to each other than to real experiments. Nevertheless, the upshot will be that all three are fundamental to scientific research, in opposition to the provocative view that numerical experimentation will replace thought experimentation.

    • Anouk Barberousse, Sorbonne Université – Sciences, Normes, Décision

On the respective powers of computer simulations and imagination

Are computer simulations thought experiments in silico? I argue they are much more than that. Prima facie, it seems that computational constraints of various sorts impose limits on what can be simulated, whereas the power of imagination appears to escape this kind of limits. However, computed results can take us much further than our imagination capacities.

    • Claus Beisbart, Universität Bern

What if we want to know more about counterfactuals? How experiments, thought experiments and computer simulations help gain knowledge about counterfactual conditionals

Experiments, thought experiments and computer simulations are often used to study counterfactual scenarios. The respective results may thus be summarized in terms of counterfactual conditionals, which become at best the object of new knowledge. But how exactly do these methods help establish knowledge of counterfactuals? To answer this question, the talk starts from a few examples in which scientists have claimed to obtain knowledge of counterfactuals by means of the methods. Cotenability views of conditionals and a possible-world semantics for counterfactuals are used to explain what counterfactuals mean. The question then is how experiments, thought experiments and computer simulations can at least make a case for believing such counterfactuals. Assuming that experiment is the least controversial method among the three, it is briefly explained how experiments can help gain knowledge about counterfactuals. I then use my examples and the semantic considerations as a testing ground for some prominent epistemologies that have been suggested for thought experiments and computer simulations, respectively. A particular focus is on matters of validation. A two-fold benefit is hoped for: First, some accounts of thought experiments or computer simulations may do better than others in view of the counterfactuals. Second, the hope is that this exercise in comparative epistemology allows for a clear appreciation of the similarities and differences between the three scientific methods.

    • Daniel Dohrn, Humboldt-Universität zu Berlin

Schlick on Simulating Experience as a Sinnkriterium for Thought Experiments

Thought experiments pervade intellectual history, but they have not been discussed as a category until the nineteenth century. Around 1900, philosophers like Ernst Mach, intrigued by their lack of direct contact to reality, started to critically inquire into the conditions under which merely imagined scenarios might contribute to solving relevant questions. While Mach’s contributions have been intensely researched, the views of empiricist near-contemporaries like Schlick tend to be somewhat neglected. I discuss Schlick’s ambiguous stance towards thought experiments. On the one hand, he repudiates merely though-ought scenarios as not bearing on reality, on the other hand he uses thought experiments and he insists that philosophers are entitled to do so. The explanation of this ambiguous stance is that he reckoned thought experiments to make sense only under one condition: the questions addressed had to be solvable by imaginatively simulating the experiences which would have confirmed that the scenario considered had been actual. I show how this criterion leads Schlick to a particular stance towards exemplary thought experiments.

    • Rawad El Skaf, Université Paris 1 Panthéon-Sorbonne, IHPST

The Structure of Scientific Thought Experiments, or an Inconsistency Revealers and Eliminators Account

In this talk, I propose and defend a novel “non-reductive”, “non-restrictive” epistemic account of scientific thought experiments (hereafter TEs), compatible with empiricism and built on case studies from the history of physics. In it, I characterise TEs as inconsistency revealers and eliminators and argue that they share a common general structure. This structure appraises TEs as a sui generis scientific tool (thus non-reductive) and accounts for unrealisable in principle TEs (thus non-restrictive). First, I briefly outline the current epistemic debate on TEs. Second, I formulate this account’s two main claims; i.e. (i) TEs are inconsistency revealers and eliminators, and (ii) TEs share a common general structure. Third, I provide several illustrations from the history of physics. Finally, I expose why TEs bring new knowledge and describe the ways they could fail.
This analysis indicates that the structure of TEs, at least in some cases, differ from other scientific tools such as computer simulations: TEs could involve scenarios whose nomological possibility (i.e. possibility under a certain theory, law or principle) remains indeterminate. In the sense that the TEer, in some cases, is free to imagine under-described, fictive particulars (e.g. a demon) and their dynamics (e.g. the demon opens and closes a massless door), in such a way that they give rise to under-described, fictive processes (e.g. the demonic process that separates fast from slow molecules), concerning which we are not in a position to say whether they are nomologically possible or not, even under our best theories. In addition, for the purpose of these TEs, the scenario’s nomological possibility could be ignored.

    • Stephan Hartmann, Ludwig-Maximilians-Universität München, MCMP

Confirmation via Analogue Simulation: A Bayesian Analysis

Analogue simulation is a novel mode of scientific inference found increasingly within modern physics, and yet all but neglected in the philosophical literature. Experiments conducted upon a table-top “source system” are taken to provide insight into features of an inaccessible “target system”, based upon a syntactic isomorphism between the relevant modelling frameworks. An important example is the use of acoustic “dumb hole” systems to simulate gravitational black holes. In a recent work Dardashti, Thebault and Winsberg argued that there exists circumstances in which confirmation via analogue simulation can obtain; in particular when the robustness of the isomorphism is established via universality arguments. This talk reviews and extends these claims via an analysis in terms of Bayesian confirmation theory. The talk is based on joint work with Radin Dardashti, Karim Thebault and Eric Winsberg.

    • Nenad Miscevic, Central European University, Budapest

Simulation in political thought experiments

Some famous political thought experiments, like the Veil-of-ignorance, or Habermas’s dialogical ones invite the reader to simulate the decision process of imagined characters. The whole Kantian tradition comes close to this kind of invitation. The paper discusses the promises and problems of such „simulationist“ approaches to political thought experimenting, tentatively arguing that promises are worth pursuing, in spite of undeniable problems along the way.

    • Gualtiero Piccinini, University of Missouri – St. Louis

Mental Representation, Simulation, and Thought Experiments

I distinguish between two types of representation, natural and nonnatural. I argue that nonnatural representation is necessary to explain intentionality. I also argue that traditional accounts of the semantic content of mental representations are insufficient to explain nonnatural representation and, therefore, intentionality. To remedy this, I sketch an account of nonnatural representation in terms of natural representation plus offline simulation of nonactual environments plus tracking the ways in which a simulation departs from the actual environment. This is a step towards a naturalistic, mechanistic, neurocomputational account of intentionality. A consequence of this account is that thought experiments are mental simulations.

    • Michael Stuart, London School of Economics, CPNSS

Thought Experiments and Computer Simulations are Metaphorical Experiments

“Both thought experiments and simulation experiments apparently belong to the family of experiments, though they are somewhat special members because they work without intervention into the natural world” (Lenhard 2017, 484). What is an experiment that does not intervene, and how do we learn from it? This has been a major focus in the epistemologies of thought experiments and computer simulations. One attractive solution is to say that an experiment without intervention is a metaphorical experiment. But not all aspects of thought experiments and computer simulations are metaphorical. So which are the metaphorical aspects, and how exactly do they gain their metaphorical status? If we can answer these questions, we can begin to address the more difficult issue, which concerns how scientists learn about real things from metaphors. In aesthetics, it is widely accepted that imagination is the cognitive ability that underpins our usage of metaphor; so we can assume that it will also be necessary for an account of metaphorical experiments.

With the help of Nancy Nersessian, I recently completed an ethnographic study of scientific imagination in a computational systems biology laboratory. In this talk, I will present the various ways that we observed scientists using their imaginations to create, use and interpret computer simulations. I will compare these to the uses of imagination in thought experiments, and argue that imagination must be a central notion for any epistemological account of metaphorical experiments. I close by considering what kind of account of imagination we need, and what it would allow us to do.

Lenhard, J. 2017. “Thought Experiments and Simulation Experiments: Exploring Hypothetical Worlds,” in M. Stuart et al. The Routledge Companion to Thought Experiments. London: Routledge.