Lorentz Center - Understanding and the Aims of Science from 31 May 2010 through 4 Jun 2010
  Current Workshop  |   Overview   Back  |   Home   |   Search   |     

    Understanding and the Aims of Science
    from 31 May 2010 through 4 Jun 2010




The pictorial languages of theoretical physics.

Sander Bais


The sciences have over time developed numerous visual systems to encode their key messages. These can be thought of as symbolic languages, each with their own iconographic vocabulary and syntaxis, somehow reflecting in a very effective and transparant way the contents the field of study. They facilitate not only the conceptualisation as with the use of space-time diagrams in the theory of relativity but serve often also serve to systematize calculational schemes as with the use of Feynman diagrams in quantum field theory. Many branches of mathematics use pictorial language, geometry to begin with but you also find many diagrammatic tools in group theory or topology; in fact all of  mathematics concerns in one way or another a symbolic representation of objects. Likewise the physical sciences from biology to electric engineering and computer science feature a hierarchies of such languages.  

In the talk I will highlight some specific examples to illustrate the power of images in physics -- supporting the view that an image can say more than a thousand words or even  a bunch of dense formulae. For that reason pictorial language is at the heart of scientific communication. 



Visualizing Complex Dynamic Mechanisms in Biology

William Bechtel

Department of Philosophy, Science Studies Program, and Center for Chronobiology University of California, San Diego


Visualizations play fundamental roles in scientists’ reasoning about mechanisms. Diagrams enable scientists to keep track of the myriad parts and operations identified within a mechanism and how they are organized. Often investigators simulate the behavior of a proposed mechanism, either in their heads or mathematically, starting from the diagram. But visualizations are still more central as scientists confront the complex dynamics exhibited in endogenously active biological mechanisms. Using examples involving recent research on the mechanisms responsible for circadian behavior, I will illustrate the critical role visualization plays in determining the phenomenon to be explained, gaining experimental evidence about the functioning of the mechanism, and understanding how the mechanism produces stable oscillations that can be synchronized to each other and entrained by a variety of Zeitgebers.



What Do We See in a Thought Experiment?

James Robert Brown

Philosophy, Toronto


The Lorentz contraction of a rapidly moving object was thought to be something one would see.  It turns out that rapidly moving objects do not appear contracted, but rather they appear rotated.  This was a surprising discovery, still not widely known.  Many thought experiments in Special Relativity assume that we see the contraction; it is essential to sorting out various conceptual problems.  Indeed, in the thought experiment, it is a contraction we see.  What this means is that appearances in a thought experiment are not like appearances in reality.  This flies in the face of the so-called continuity thesis of Sorenson and others, who claim that thought experiments and real experiments form a continuum.  It also illustrates something quite dramatic about the distinction between appearance and reality.  What we see in a thought experiment is the view from nowhere. Only the thought experiment gets it right, since the rotation we would experience in a real experiment is a kind of illusion, whereas Lorentz contraction is an objective feature of reality.



Mechanistic Understanding in the Life Sciences:  Cases from Molecular and Evolutionary Biology

Lindley Darden

Department of Philosophy, University of Maryland


Complex relations exist between understanding, explanation, intelligibility, predictability, picturability, manipulability, diagnosability (in the light of failure), and fixability (to remove failure). This paper takes one path through these nested concepts, based on the Machamer, Darden, Craver (MDC) view of mechanisms in biology, which are characterized by entities, activities and their organization.  Mechanistic explanations show how the mechanism works to produce a phenomenon. Some mechanistic explanations, particularly those in the experimental science of molecular biology, are picturable and facilitate predictability, manipulability, diagnosability, and fixability.  Diagnosis and repair of molecular biological mechanisms are especially important in medicine.  Other mechanistic explanations lack some of these additional features, while still providing understanding. Examples include explanations in evolutionary biology based on the mechanism of natural selection.



MDC: Machamer, Peter, Lindley Darden, and Carl F. Craver (2000), "Thinking About Mechanisms," Philosophy of Science 67: 1-25.Reprinted in first chapter of

Darden, Lindley (2006; paperback 2009), Reasoning in Biological Discoveries: Mechanisms, Interfield Relations, and Anomaly Resolution. New York: Cambridge University Press.



Understanding and explanation in physics

Dennis Dieks

Utrecht University


In this talk I shall argue that understanding in physics is sometimes achieved via the creation of  conceptual schemes that facilitate qualitative arguments but that are false, as judged from the perspective of accepted theory, when taken literally. This falsity is often an acknowledged fact; still, awareness of the falsity of the picture does usually not lead to its rejection for the purpose of providing understanding.

In the talk I shall review a number of examples of this taken from modern physics, in particular quantum physics.

The broader thesis that I shall defend on the basis of an analysis of this situation is that understanding and explanation are relatively autonomous subjects: one may plausibly require that good explanations should be in complete accordance with accepted theory, but this does not hold for understanding. As I shall argue, understanding is the more flexible concept, which not only makes use of accepted physical theory but combines such theory with available “conceptual toolkits”, thereby often deviating from the exact content of the theory. 



Understanding and Interpreting Quantum Mechanics

Jan Faye

Section of Philosophy, Department of Media, Cognition and Communication, University of Copenhagen


Several generations of physicists and philosophers have seen quantum mechanics as the most intriguing physical theory which nobody really understood. Already shortly after its formulation around 1925 physicists began wonder how they could make sense out of the formalism. They were confronted with non-commuting variables which was quite new for a physical theory. Also they were confronted with two different formalisms which involved different mathematical expressions. Some of these expressions seem pointed to quantum object as if they were particles, whereas other expressions made sense as if these objects were waves. The upshot became the Copenhagen interpretation which its notions of complementarity, indeterminacy, and non-visualizability. The centre of the disputes was how one should understand the state vector or wave function, and even today it still is.


In the period of classical physics a concept like visualization played the main role for understanding physical phenomena. A physicist couldn’t claim to have understood a physical problem unless he was able to give it a visual solution in terms of picturable mechanisms and processes. It was thought that a physicist was successful in providing a visual solution if could describe the development of a physical system as a continuous process in space and time.


This is a view which currently has its proponents. Perhaps even many. The late James Cushing defended such a view in his book on quantum mechanics. He made a distinction between empirical adequacy, explanation, and understanding. Quantum mechanics is an empirically adequate theory. It also gives us some form of explanation but no understanding in the perspective of the Copenhagen interpretation. He then concluded: “My concern has been that, for physical processes, a satisfying understand may be possible for us only when we are able to tell a local, causal picture story of those processes, If this is so, then we are, by definition, doomed, according to the standard (Copenhagen) interpretation of quantum mechanics, to be unable to understand quantum phenomena (and quantum mechanics) (p.20). Instead Cushing defended the Bohmian interpretation of quantum mechanics which supplies quantum mechanics with an ontology according to which an atomic object is seen as a classical particle. The object’s trajectory, however, is constantly under the influence of a non-local quantum field potential.


Other physicists were also dissatisfied with the standard interpretation because they saw it as incoherent by not giving us a physical understanding of the ontological basis of the quantum mechanics and the measuring process. Hugh Everett therefore introduced the many worlds interpretation according to which every possible measuring result contained in the state vector description of an observable is realized in its own world different from the actual world. Apparently, the idea here is that we can maintain a kind of causal understanding of quantum processes as long we are open for introducing physical worlds besides the actual world.


The bottom-line seems to be this: The indeterminacy contained the Copenhagen interpretation prohibits any ontological understanding of quantum phenomena. In my contribution, however, I will try to argue that there is no particular reason to associate understanding with causation and visualization if the epistemic costs are too high. Neither Bohm’s non-local quantum potentials nor Everett’s many worlds are accessible to observation or empirical test. Rather these interpretations seem to be epistemic ad hoc. Bohm just added by hand an ontology to save a description in terms of a classical trajectory, whereas Everett merely “read off” a physical representation from the abstract Hilbert-space-formulation without putting any epistemological constraints on whether such a representation should be in accordance to what else we know about the world. Based on this discussion I shall argue that the requirements of understanding in quantum physics are rather complex and contextual.



A pragmatic theory of scientific explanation

Raoul Gervais


The project aims to provide an account of explanation by starting from an analysis of explanatory practice, and to identify some common features of scientific explanation which seem to apply regardless of the particular discipline. Three such features stand out:


1) Taking our cue from Van Fraasen en Kitcher's work, explanations can be seen as answers to why-questions.


2) The formulation of these why question, and consequently the form the explanation takes, depend partly on the epistemic interests of the particular researchers involved, i.e. what they hope to achieve in answering a particular question.


3) Taking into account these epistemic interests, one can say something about the relative explanatory power of different kinds of explanation.


Though not the first in choosing a pragmatic approach, our project differs from traditional approaches in that we attempt to encompass not just the explanatory practices of a few 'model' sciences, but a variety of interrelated scientific disciplines, from physics and biology, to history and sociology. My particular contribution to the project will be in the sphere of psychology and biology.



Unification and Understanding
Victor Gijsbers

Explanation is a matter of determination of the particular explanandum, not of unification of the explanandum and as many other facts as possible. I will review some of the arguments for this conclusion, and then ask why we nevertheless have the intuition that unifying our world view leads to a greater understanding of the world. In recent years, writers such as Peter Lipton and Henk de Regt have explored the possibility that understanding and explanation diverge--that not all understanding is by way of explanation. I will explore the possibility that it is in these areas of divergence that unification plays its understanding-enhancing role.



Agent-Based Simulation and Sociological Understanding

Peter Hedström & Petri Ylikoski


The presentation will discuss the use of agent-based simulations as a tool of theoretical exploration in sociology. We will argue that simulations are an indispensable tool for a development of mechanism-based explanations of many social phenomena. The simulations increase explanatory understanding both by allowing more systematic study of consequences theoretical assumptions and by making theoretical inferences more reliable. However, the simulations can also give rise to illusory understanding, for example, when a researcher attempts to build ”a realistic” simulation of a complex real world phenomenon without properly understanding the inner workings of the simulation, or when the results of a simple theoretically motivated simulation are hastily extrapolated to real world phenomena.



The epistemology of computer simulations: Three types of understanding

Rafaela Hillerbrand

Institut für Philosophie & Human Technology Center, RWTH Aachen University


Taking hydrodynamics as study case, this paper analyses the epistemic status of computer simulation compared to both theory and experiment. I introduce the

following terminology that distinguishes computer simulations with respect to their

epistemic goals. Type-I simulations aim to provide information on mathematical systems

that cannot (yet) be solved analytically; type-II simulations yield information on systems that are experimentally inaccesible. While type-I simulations may substitute theories, type-II simulations replace experiments. A third type of simulation aims at predicting the behaviour of (generally complex) real world systems for which we lack an accepted theoretical description. On the basis of this distinction I want to reject general claims on computer simulations as ‘opaque thought experiments’ (Di Paolo) or ‘lacking any empirical content’ (Oreskes).



Idealization, Mechanisms, and Understanding in Economics

Frank Hindriks


Traditional accounts of explanation fail to illuminate the explanatory relevance of models that are descriptively false in the sense that the regularities they entail do not obtain. In this paper I propose an account of explanation, which I call ‘explanation by concretization’, that serves to account for the explanatory relevance of such models. The basic idea is that scientists provide causal explanations of why the regularity entailed by an abstract and idealized model _fails to_obtain. They do so by adding complicating factors to that model – i.e. by concretizing it. In order that explanation by concretization may succeed, the original model must represent a basic mechanism. This account of explanation is developed in the context of economics and contrasted to those of Daniel Hausman and Nancy Cartwright. It also provides the basis for an account of how unrealistic models can be used to achieve an understanding of the way mechanisms work.



The Tense Relation between Understanding and Evaluation in the History of Science

 Bart Karstens


Historical understanding of science is mostly achieved by descriptions of the past which explain how and why things went as they went. There is great emphasis on the relations of science to the political, social and cultural realms of society. The aim of science according to these accounts is not a search for knowledge for the sake of knowledge itself but the search for knowledge is always interpreted with respect to other societal factors that explain the relative success of knowledge claims. Historical research has offered many insights in the actual dynamics of science and rightfully questioned interpretations of science as a one sided story of linear progress. Still it seems that understanding past science should also be about evaluating the outcomes of the scientific processes. Understanding and evaluation however form an uneasy couple in historiography of science of the last decades. In my talk I will give reasons for this and explore possibilities to integrate evaluation in the concept of (historical) understanding.



Understanding and the aims of cognitive neuroscience

Rogier Kievit


Cognitive neuroscience commonly involves the simultaneous analysis of behavioral and neurological data, and is one of the most active fields of psychological inquiry. However, the exact purpose, or ultimate goal, of this class of reductive science is often left unstated. Is the ultimate goal to reduce psychological properties to neurological properties, to better explain them, to constrain mechanistic interpretations or to establish the ontological status of psychological constructs?


In this talk I will consider these options from a conceptual perspective, and to see to what extent these aims are realistic, justifiable or sensible. I will argue that the cognitive neuroscience should be seen as a separate roads of insight, and that theoretical explication of certain positions may serve to structure reductive hypotheses. Drawing on psychometrics, neuroscience, genetics and psychology, I will see what the aims of reductive psychological investigation could be, and maybe even should be.



A feeling for patterns in rivers: understanding and reductionistic explanation in earth science

Maarten G. Kleinhans


Earth science seeks to understand phenomena and their history on Earth. Using examples of fluvial research, I will first show that earth scientists gain understanding from unification and reduction approaches. The former includes classical geologic reconstruction of past conditions on (a part of) Earth. The latter includes physical

or chemical explanation, often in the form of numerical models. Furthermore, reductionistic explanations are used as a partial basis of unified explanations. I will then argue that in both approaches understanding is obtained if a theory is intelligible sensu De Regt and Dieks and if theoretical skills are integrated with performative skills sensu Leonelli, despite the large differences between the required skills.



A search for non-reductive unification

Merel Lefevere


Although I believe that unification can play a role in the explanatory process, there are still problems to be solved. Unification can pose a formal constraint on the research process. Mäki opposes this worry by distinguishing derivational and ontological unification. Elaborating further on that point of view, I doubt there is solely derivational explanatory unification. There are always some shared ontic foundations. Any unification process is by necessity ontological, although it can have different origins:  it can originate from the imposed law or it can be based on common grounds that are not necessarily lawlike.  I want to distinguish a top-down unification that meets the expectability requirement of deductive nomological explanation from a bottom-up unification that does not meet those requirements, but still has explanatory virtue. By this approach the worry that unification poses formal constraints is also met and moreover, this approach can broaden the usability of unification as an explanatory virtue.



Understanding Data in the Digital Age: The Experimental Context In Silico

Sabina Leonelli

University of Exeter, UK


The consultation of online data repositories and the related use of computer software to retrieve, display and model such data are fast becoming key components of scientific research. This paper addresses the relation of such in silico research to in vivo experimentation, by examining the conditions under which the scientific significance of data posted online is understood by the researchers who access them. Such an understanding arguably underpins and guides the subsequent use of data to foster new discoveries: it is hard to imagine how data could be used as evidence for a claim, or as a reason to set up a research project, in the absence of intuitions about what those data tell scientists about specific entities or processes. However, little philosophical reflection has addressed the problem of what gives scientific meaning to data available online – what makes it possible for scientists to interpret them and understand their evidential value. I shall argue that no such understanding is possible without embodied knowledge of the target system(s) that data are taken to document. In other words, familiarity with in vivo experimentation on actual entities and processes is crucial to assessing the quality and scientific meaning of data available in silico. Notably for the aims of data-intensive science, the experimental context and target systems in which data find meaning can vary: it does not necessarily have to be the same context, or even the same set of objects, in which data were originally produced. These philosophical contentions will be illustrated through the analysis of case studies from experimental and systems biology.



A Practice Driven approach to MEPP in terms of Conceptual Tools and Conceptual Resources

Daniele Molinini


Starting from the framework proposed by Henk De Regt and Dennis Dieks in their paper “A contextual approach to scientific understanding” (2005), I will propose my approach to MEPP (Mathematical Explanations of Physical Phenomena) in terms of Conceptual Tools and Conceptual Resourches. I will claim that Conceptual Tools are vectors of Conceptual Resources in MEPP (where conceptual resources depends on the particular situation under study). Without endorsing any specific model of Explanation (such as Kitcher’s Unificationist model or Van Frassen’s), I will show how a Practice-Driven approach to MEPP supports my claim. In particular, I will offer an example where, when faced with a mathematical explanation of a same phenomenon, we can evaluate the explanatory potential by weighting the conceptual resources which come into play (through the Conceptual Tools of visualization and abstract reasoning). Finally, I will sketch the payoff of adopting such an approach (ontological debate, "Pluralism" in explanation, understanding, asymmetry problem).



Quantum optics as a mechanistic understanding of light

Gerard Nienhuis

Universiteit Leiden


Quantum optics studies the interaction between light and matter in situations where quantum mechanics is needed for its understanding. The simplest component of matter is a single two-state atom, and the simplest component of a light field is a single mode. We shall discuss two idealized experimental situations where the state of one component (light or matter) is manipulated by measurements performed exclusively on the other component (matter or light). A measurement on one component with a null result (for instance, no photon is seen by the detectors) can change the state of the other component (the atom). Or the detection of the position of an atom can modify the number of photons in the mode, although no photons are absorbed or emitted. The theoretical prediction of the measurement outcome is unambiguous. In this sense the measurement is understood. But our understanding of the mechanism by which the effects come about can vary remarkably. A causal-mechanical explanation does not seem possible.



Understanding with and without explanation

Antigone Nounou and Fred Muller


In the decades-long discussions concerning science, its characteristics and its aims, the notion of scientific understanding had either been conspicuously missing or played second fiddle to explanation. Thus, in its rare appearances in the philosophical literature of the past, scientific understanding was thought to duly accompany, in the sense of being conferred by, scientific explanations. In recent philosophical writings the focus has shifted, and scientific understanding is being presented as one of science’s distinctive aims. Nonetheless, the idea that there exists an intimate relation between explanation and understanding seems to have taken root so deeply that understanding is still invariably associated with explanations. In our talk we will argue that in analyzing the notion of scientific understanding, one should distinguishing between two projects: understanding _with_ explanation and understanding _without_explanation. None of these two projects is bulletproof, we will contend, neither is a third project, which can legitimately be taken as encompassing the two.



Compression-driven progress in science

Leo Pape


Explaining and predicting the world, based on regularities underlying observations, are essential ingredients of scientific research. In computer science, these scientific activities can be captured in the formal theory of compression-driven progress. This theory considers scientists as computationally limited observers that try to store and compress observations in an efficient manner. Finding efficient representations entails identifying regularities that allow the observer to compress the original observations and predict future observations. Compression progress is achieved when the observer discovers previously unknown regularities that provide increased compression of observations. The theory further postulates that scientists direct their attention to interesting data, that is, data that is neither impossible to compress (i.e. truly random) nor easily compressible with existing methods, but is expected to hold previously unknown regularities that allow for further compression. I will discuss how the theory of compression progress can be used for automated discovery of regularities (i.e. an artificial scientist), and can assist human scientists in selecting appropriate abstraction levels for their models.



Simulation and Understanding in the Study of Weather and Climate

Wendy S. Parker


For decades, scientists who study weather and climate have emphasized the value of computer simulation models as a resource for advancing understanding. Both individual models as well as “hierarchies” of models are said to have important roles to play.  What sort of understanding do these scientists have in mind? How are simulations – and hierarchies of models – supposed to promote this understanding? I will address these questions and then consider the recent suggestion (Held 2005) that more attention should be devoted to the construction of “appropriate” model hierarchies in order to narrow the gap between our ability to simulate complex weather and climate phenomena and our ability to understand those phenomena.  What makes for an “appropriate” hierarchy? What does Held mean when he refers to an E. coli of climate models?



Held, Isaac (2005) “The Gap between Simulation and Understanding in Climate Modeling”, Bulletin of the American Meteorological Society 86: 1609-1614.



Visualization as a tool for scientific understanding

Henk W. de Regt

VU University Amsterdam, NIAS Wassenaar


Visualizability and intelligibility have often been associated and many scientists have explicitly stated that only visualizable theories can provide understanding. I will examine the relation between visualizability and intelligibility, via a case study of the transition from classical to quantum physics, which witnessed a gradual loss of visualizability. In the mid-1920s, the issue of the relation between intelligibility and visualizability (Anschaulichkeit) became a central topic in the debate between Schrödinger and Heisenberg, who defended rival versions of quantum theory. I analyze this debate, and the subsequent reinterpretation of the notion of Anschaulichkeit by Heisenberg. I employ this reinterpretation, which implies that visualizability is not a necessary condition for intelligibility, to develop a theory of scientific understanding. According to my theory, visualizability is but one out of many possible tools for understanding, albeit one that has proved to be very effective in science.



Causal inference in the social science. Why causal mechanisms matter and how to find out about them

Attilia Ruzzene

Phd student at Erasmus Institute for Philosophy and Economics (Rotterdam)


Causal knowledge in the social science -in economics in particular- serves a variety of purposes. It is used for explaining social phenomena; furthermore, it constitutes ground for devising social policies. In my phd thesis I study two cases of scientific practice (a case-study research on regional development and a model for causal effects) in which causal knowledge is used for explanatory and intervention purposes respectively. I argue that in both cases causal mechanisms are relevant to the purposes at hand. Provided that causal mechanisms would help us attaining our epistemic and non epistemic goals, valid evidence of them has to be found in the first place. Arguably, this task is extremely challenging for the social science. Causal mechanisms in the social realm are plagued by non-observability, context-dependence and uncertainty (what mechanisms trigger when?). This fact suggests that contextual information cannot be done away with when looking for mechanisms. Moreover, it might (and maybe should) be regarded as clue to what type of mechanisms are at work in given circumstances. I suggest that social scientists should pay (more) careful attention to contexts and devise methods that enable them to collect contextual information in a reliable and efficient way.



Explanatory Autonomy and Explanatory Irreducibility

Michael Strevens


A powerful argument for anti-reductionism turns on the premise that the biological, behavioral, and social sciences are, in the way that they explain their characteristic subject matters, in some sense autonomous from physics. The argument is formulated and strengthened in this paper, and then undermined by showing that a reductionist account of explanation is not only consistent with, but provides a compelling account of, explanatory autonomy. Two kinds of explanatory abstraction, objective and contextual, play key roles in the story.



Where Explanation Ends: Understanding as the Place the Spade Turns in the Social Sciences

Stephen Turner

Dept of Philosophy, University of South Florida


Mechanisms need to be made of something. In the physical case, they are models made of regularities realized in the model. In the case of causal models in the social sciences, they are made of correlations whose “causal” character is assured by “assumptions” whose epistemically problematic content is minimized. Much of this minimized content, however, takes the form of highly plausible possible interpretations of actions and decisions or abstractions of sequences involving actions and decisions.   But action explanations are themselves subject to understanding. These “understandings” are a further step, and they raise complex problems: what is it to understand a concept? What does the fact that we can understand a choice as rational mean? Does  this kind of understanding add any epistemic weight to the explanation, as an empirical theory would? Is either kind of understanding the explanatory dead-end, and if so, what does this imply for the distinctiveness of social science knowledge? A cognitive science approach can clarify some of these issues by accounting for understanding of others directly.



Horizons of Understanding of Space and Time 

Herman Verlinde


Event horizons represent a theoretical  frontier of our observable world and of our understanding of our current laws of physics. Recent progress towards uncovering the true nature of event horizons is used to illustrate the tools and methods by which theoretical physicists aim to advance scientific understanding.  I will describe how, starting from basic paradoxes and analogies, and by requiring unity of existing physical laws, theorists are uncovering deep new physical principles, known as `dualities',  that are reshaping the classical notion of locality and causality. 



Thinking about Scientific Understanding and Explanation as a Structural Realist

Ioannis Votsis


Structural Realism is a viewpoint in the scientific realism debate. In its epistemological guise it holds that our knowledge of the physical world is at best structural. More precisely, we can only know the physical world up to isomorphism. In its ontological guise it explains this structural limitation to our knowledge by appeal to an ontology which is itself in some sense or other wholly structural. Although research into structural realism is booming, little has been said about what its implications are for scientific understanding and explanation. In this talk I explore these implications and argue

that at least when it comes to the natural sciences what counts as understanding and explanation has taken a highly abstract and mathematical turn that is very much in line with the structural realist pronouncements.