A common strategy for simplifying complex systems involves partitioning them into subsystems whose behaviors are roughly independent of one another at shorter timescales. Dynamic causal models clarify how doing so reveals a system’s nonequilibrium causal relationships. Here I use these models to elucidate the idealizations and abstractions involved in representing a system at a timescale. The models reveal that key features of causal representations—such as which variables are exogenous—may vary with the timescale at which a system is considered. This has (...) implications for debates regarding which systems can be represented causally. (shrink)

Causal representations are distinguished from non-causal ones by their ability to predict the results of interventions. This widely-accepted view suggests the following adequacy condition for causal models: a causal model is adequate only if it does not contain variables regarding which it makes systematically false predictions about the results of interventions. Here I argue that this condition should be rejected. For a class of equilibrium systems, there will be two incompatible causal models depending on whether one intervenes upon a certain (...) variable to fix its value, or ‘lets go’ of the variable and allows it to vary. The latter model will fail to predict the result of interventions on the let-go-of variable. I argue that there is no basis for preferring one of these models to the other, and thus that models failing to predict interventions on particular variables can be just as adequate as those making no such false predictions. This undermines a key argument against relying upon causal models inferred from equilibrium data. (shrink)

Causal representations are distinguished from non-causal ones by their ability to predict the results of interventions. This widely-accepted view suggests the following adequacy condition for causal models: a causal model is adequate only if it does not contain variables regarding which it makes systematically false predictions about the results of interventions. Here I argue that this condition should be rejected. For a class of equilibrium systems, there will be two incompatible causal models depending on whether one intervenes upon a certain (...) variable to fix its value, or ‘lets go’ of the variable and allows it to vary. The latter model will fail to predict the result of interventions on the let-go-of variable. I argue that there is no basis for preferring one of these models to the other, and thus that models failing to predict interventions on particular variables can be just as adequate as those making no such false predictions. This undermines a key argument (Dash in Caveats for causal reasoning with equilibrium models. University of Pittsburgh. PhD thesis, 2003) against relying upon causal models inferred from equilibrium data. (shrink)

I examine to what extent accounts of mechanisms based on formal interventionist theories of causality can adequately represent biological mechanisms with complex dynamics. Using a differential equation model for a circadian clock mechanism as an example, I first show that there exists an iterative solution that can be interpreted as a structural causal model. Thus, in principle, it is possible to integrate causal difference-making information with dynamical information. However, the differential equation model itself lacks the right modularity properties for a (...) full integration. A formal mechanistic model will therefore have to leave out either noncausal or causal explanatory relations. (shrink)

Why do people get sick? I argue that a disease explanation is best thought of as causal network instantiation, where a causal network describes the interrelations among multiple factors, and instantiation consists of observational or hypothetical assignment of factors to the patient whose disease is being explained. This paper first discusses inference from correlation to causation, integrating recent psychological discussions of causal reasoning with epidemiological approaches to understanding disease causation, particularly concerning ulcers and lung cancer. It then shows how causal (...) mechanisms represented by causal networks can contribute to reasoning involving correlation and causation. The understanding of causation and causal mechanisms provides the basis for a presentation of the causal network instantiation model of medical explanation. (shrink)

Human and machine discovery are gradual problem-solving processes of searching large problem spaces for incompletely defined goal objects. Research on problem solving has usually focused on search of an instance space (empirical exploration) and a hypothesis space (generation of theories). In scientific discovery, search must often extend to other spaces as well: spaces of possible problems, of new or improved scientific instruments, of new problem representations, of new concepts, and others. This paper focuses especially on the processes for finding new (...) problem representations and new concepts, which are relatively new domains for research on discovery.Scientific discovery has usually been studied as an activity of individual investigators, but these individuals are positioned in a larger social structure of science, being linked by the blackboard of open publication (as well as by direct collaboration). Even while an investigator is working alone, the process is strongly influenced by knowledge and skills stored in memory as a result of previous social interaction. In this sense, all research on discovery, including the investigations on individual processes discussed in this paper, is social psychology, or even sociology. (shrink)

The import of Hodgkin and Huxley’s classic model of the action potential has been hotly debated in recent years, with particular controversy surrounding claims by prominent proponents of mechanistic explanation. For these authors, the Hodgkin-Huxley model is an excellent predictive tool but ultimately lacks causal/explanatory import. What is more, they claim that this is how Hodgkin and Huxley themselves saw the model. I argue that these claims rest on a problematic reading of the work. Hodgkin and Huxley’s model is both (...) causal and, in an important sense, explanatory. (shrink)

In this paper, I review two related lines of computational research: discovery of scientific knowledge and causal models of scientific phenomena. I also report research on quantitative process models that falls at the intersection of these two themes. This framework represents models as a set of interacting processes, each with associated differential equations that express influences among variables. Simulating such a quantitative process model produces trajectories for variables over time that one can compare to observations. Background knowledge about candidate processes (...) enables search through the space of model structures and associated parameters to find explanations of time-series data. I discuss the representation of such process models, their use for prediction and explanation, and their discovery through heuristic search, along with their interpretation as causal accounts of dynamic behavior. (shrink)

Economics prefers complete explanations: general over partial equilibrium, microfoundational over aggregate. Similarly, probabilistic accounts of causation frequently prefer greater detail to less as in typical resolutions of Simpson’s paradox. Strategies of causal refinement equally aim to distinguish direct from indirect causes. Yet, there are countervailing practices in economics. Representative-agent models aim to capture economic motivation but not to reduce the level of aggregation. Small structural vector-autoregression and dynamic stochastic general-equilibrium models are practically preferred to larger ones. The distinction between exogenous (...) and endogenous variables suggests partitioning the world into distinct subsystems. The tension in these practices is addressed within a structural account of causation inspired by the work of Herbert Simon’s, which defines cause with reference to complete systems adapted to deal with incomplete systems and piecemeal evidence. The focus is on understanding the constraints that a structural account of causation places on the freedom to model complex or lower-order systems as simpler or higher-order systems and on to what degree piecemeal evidence can be incorporated into a structural account. (shrink)

What are the pillars on which the success of modern science rest? Although philosophers have much discussed what is behind science’s success, this paper argues that much of the discussion is misdirected. The extant literature rightly regards the semantic and inferential tools of formal logic and probability theory as pillars of scientific rationality, in the sense that they reveal the justificatory structure of important aspects of scientific practice. As key elements of our rational reconstruction toolbox, they make a fundamental contribution (...) to our understanding of the success of science.At the same time, any science, however exact, is dominated by approximation, error, and uncertainty, a fact that makes one wonder how science can be so successful. This paper articulates and illustrates general themes—e.g., that truth-preserving arguments often fail to preserve approximate truth—that highlight the need for additional semantic resources. Thus, our proposal is that persistent failures to unravel the reasons behind the success of science in the face of pervasive error and uncertainty should be attributed to an insufficiently rich way of rationally reconstructing scientific and mathematical knowledge. What is missing? This paper claims that there is a third formal method of reasoning that constitutes a distinct pillar on which rests the success of science, namely, perturbation theory. The paper outlines how the representational and inferential tools of perturbation theory differ from those of logic and probability theory, and how they enable us to understand the apparently elusive aspects of the success of science.However, compared to its peers, perturbative reasoning has not received the attention it deserves. As the paper explains, this partly results from the circumstances in which perturbation theory is taught, and partly from the fact that perturbation theory first appears to be a vaguely related collection of methods offering no systematic semantic insight. In an attempt to show that this first impression is wrong, this paper presents its contribution to the semantic dimension of scientific representation and inference in terms of what I call “semantic layering.”. (shrink)

Complex organisms thwart the simple rectilinear causality paradigm of “necessary and sufficient,” with its experimental strategy of “knock down and overexpress.” This Essay organizes the eccentricities of biology into four categories that call for new mathematical approaches; recaps for the biologist the philosopher's recent refinements to the causation concept and the mathematician's computational tools that handle some but not all of the biological eccentricities; and describes overlooked insights that make causal properties of physical hierarchies such as emergence and downward causation (...) straightforward. Reviewing and extrapolating from similar situations in physics, it is suggested that new mathematical tools for causation analysis incorporating feedback, signal cancellation, nonlinear dependencies, physical hierarchies, and fixed constraints rather than instigative changes will reveal unconventional biological behaviors. These include “eigenisms,” organisms that are limited to quantized states; trajectories that steer a system such as an evolving species toward optimal states; and medical control via distributed “sheets” rather than single control points. (shrink)

Complex organisms thwart the simple rectilinear causality paradigm of “necessary and sufficient,” with its experimental strategy of “knock down and overexpress.” This Essay organizes the eccentricities of biology into four categories that call for new mathematical approaches; recaps for the biologist the philosopher's recent refinements to the causation concept and the mathematician's computational tools that handle some but not all of the biological eccentricities; and describes overlooked insights that make causal properties of physical hierarchies such as emergence and downward causation (...) straightforward. Reviewing and extrapolating from similar situations in physics, it is suggested that new mathematical tools for causation analysis incorporating feedback, signal cancellation, nonlinear dependencies, physical hierarchies, and fixed constraints rather than instigative changes will reveal unconventional biological behaviors. These include “eigenisms,” organisms that are limited to quantized states; trajectories that steer a system such as an evolving species toward optimal states; and medical control via distributed “sheets” rather than single control points. (shrink)

Weber argues that causal modelers face a dilemma when they attempt to model systems in which the underlying mechanism operates according to some set of differential equations. The first horn is that causal models of these systems leave out certain causal effects. The second horn is that causal models of these systems leave out time-dependent derivatives, and doing so distorts reality. Either way causal models of these systems leave something important out. I argue that Weber’s reasons for thinking causal modeling (...) is limited in this domain are lacking. (shrink)

Part I Introduction The specific problem adressed in this thesis is: what is the rational use of theory and experiment in the process of scientific discovery, in theory and in the practice of drug research for Parkinson’s disease? The thesis aims to answer the following specific questions: what is: 1) the structure of a theory?; 2) the process of scientific reasoning?; 3) the route between theory and experiment? In the first part I further discuss issues about rationality in science as (...) introduction to part II, and I present an overview of my case-study of neuropharmacology, for which I interviewed researchers from the Groningen Pharmacy Department, as an introduction to part III. Part II Discovery In this part I discuss three theoretical models of scientific discovery according to studies in the fields of Logic, Cognition, and Computation. In those fields the structure of a theory is respectively explicated as: a set of sentences; a set of associated memory chunks; and as a computer program that can generate the observed data. Rationality in discovery is characterized by: finding axioms that imply observation sentences; heuristic search for a hypothesis, as part of problem solving, by applying memory chunks and production rules that represent skill; and finding the shortest program that generates the data, respectively. I further argue that reasoning in discovery includes logical fallacies, which are neccesary to introduce new hypotheses. I also argue that, while human subjects often make errors in hypothesis evaluation tasks from a logical perspective, these evaluations are rational given a probabilistic interpretation. Part III Neuropharmacology In this last part I discusses my case-study and a model of discovery in a practice of drug research for Parkinson’s disease. I discuss the dopamine theory of Parkinson’s disease and model its structure as a qualitative differential equation. Then I discuss the use and reasons for particular experiments to both test a drug and explore the function of the brain. I describe different kinds of problems in drug research leading to a discovery. Based on that description I distinguish three kinds of reasoning tasks in discovery, inference to: the best explanation, the best prediction and the best intervention. I further demonstrate how a part of reasoning in neuropharmacology can be computationally modeled as qualitative reasoning, and aided by a computer supported discovery system. (shrink)