Genetic regulatory networks are complex, involving tens or hundreds of genes and scores of proteins with varying dependencies and organizations. This invites the application of artificial techniques in coming to understand natural complexity. I describe two attempts to deploy artificial models in understanding natural complexity. The first abstracts from empirically established patterns, favoring random architectures and very general constraints, in an attempt to model developmental phenomena. The second incorporates detailed information concerning the genetic structure, organization, and dependencies in actual systems in an attempt to explain developmental differences. The results offered by these models, pitched at these different levels of abstraction, are different. The more detailed models are more continuous with classical developmental approaches.

In this paper, I examine an experimental technique, gene targeting, used for establishing genotype/phenotype relationships. Through analyzing a case study, I identify many pitfalls that may lead to false conclusions about these relationships. I argue that some of these pitfalls may seriously affect gene targeting's usefulness for associating phenotypes with genes cataloged by the Human Genome Project. This case also shows the use of gene targeted mice as model systems for studying genotype/phenotype relationships in humans. Moreover, I argue that it reveals the weakness of one attempt to draw conclusions about the biological determination of sexual and aggressive behaviors in humans.

Biology deals, notoriously, with complex systems. In discussing biological methodology, all three papers in this symposium honor the complexity of biological subject matter by preferring models and theories built to reflect the details of complex systems to models based on broad general principles or laws. Rheinberger's paper, the most programmatic of the three, provides a framework for the epistemology of discovery in complex systems. A fundamental problem is raised for Rheinberger's epistemology, namely, how to understand the referential continuity of the theoretical terms and concepts employed in typical case studies involving complex systems.

In this paper the claim that laws of nature are to be understood as claims about what necessarily or reliably happens is disputed. Laws can characterize what happens in a reliable way, but they do not do this easily. We do not have laws for everything occurring in the world, but only for those situations where what happens in nature is represented by a model: models are blueprints for nomological machines, which in turn give rise to laws. An example from economics shows, in particular, how we use—and how we need to use—models to get probabilistic laws.

Mary Hesse's well-known work on models and analogies gives models a creative role to play in science, which rests on developing certain analogical properties considered neutral between the two fields. Case study material from Irving Fisher's work (The Purchasing Power of Money, 1911), in which he used analogies to construct models of monetary relations and the monetary system, highlights certain omissions in Hesse's account. The analysis points to the importance of taking account of the negative properties in the analogies and to certain differences between “ready-made” analogies (models of systems based on existing analogical structures) and “designed” analogies (models built up from separate analogical features).

In addition to its obvious successes within the kinetic theory the ideal gas law and the modeling assumptions associated with it have been used to treat phenomena in domains as diverse as economics and biology. One reason for this is that it is useful to model these systems using aggregates and statistical relationships. The issue I deal with here is the way R. A. Fisher used the model of an ideal gas as a methodological device for examining the causal role of selection in producing variation in Mendelian populations. The model enabled him to create the kind of population where one could measure the effects of selection in a way that could not be done empirically. Consequently we are able to see how the model of an ideal gas was transformed into a biological model that functioned as an instrument for both investigating nature and developing a new theory of genetics.

A general account of modeling in physics is proposed. Modeling is shown to involve three components: denotation, demonstration, and interpretation. Elements of the physical world are denoted by elements of the model; the model possesses an internal dynamic that allows us to demonstrate theoretical conclusions; these in turn need to be interpreted if we are to make predictions. The DDI account can be readily extended in ways that correspond to different aspects of scientific practice.

I argue that supervenience is an inadequate device for representing relations between different levels of phenomena. I then provide six criteria that emergent phenomena seem to satisfy. Using examples drawn from macroscopic physics, I suggest that such emergent features may well be quite common in the physical realm.

Examination of attempts at theory reduction (S to T) shows that a process of cognitive emergence is involved in which concepts of S, Cs, emerge from T. This permits the ‘bridge laws’ to be stated. These are not in conflict with incommensurability of the Cs with the CT. Cognitive emergence may occur asymptotically or because of similarities of mathematical expressions; it is not necessarily holistic. Mereologically and nonmereologically related theory pairs are considered. Examples are chosen from physics. An important distinction is made between ‘theory reduction’ and ‘reductive explanation’.

It is widely held that (1) there are autonomous levels of organization above that of the macromolecule and that (2) at least sometimes macromolecular processes are best explained in terms of such autonomous kinds. I argue that molecular developmental biology honors neither of these claims, and I show that the only way they can be rendered consistent with a minimal physicalism is through the adoption of controversial claims about causation and explanation which undercut the force of these two antireductionism claims.

Most philosophical accounts of emergence are incompatible with reduction. Most scientists regard a system property as emergent relative to properties of the system's parts if it depends upon their mode of organization—a view consistent with reduction. Emergence can be analyzed as a failure of aggregativity—a state in which “the whole is nothing more than the sum of its parts.” Aggregativity requires four conditions, giving tools for analyzing modes of organization. Differently met for different decompositions of the system, and in different degrees, these conditions provide powerful evaluation criteria for choosing decompositions, and heuristics for detecting biases of vulgar reductionisms. This analysis of emergence is compatible with reduction.

Feyerabend's “Classical Empiricism” (1970) draws on a 17th century Jesuit argument against Protestant fundamentalism. The argument is very general, and applies to any simple foundationalist epistemology. Feyerabend uses it against Classical Empiricism—roughly, the view that what is to be believed is exactly what experience establishes, and no more—which he identifies as among other things Newton's “dogmatic ideology.”

I suggest following Paul Feyerabend's own advice, and interpreting Feyerabend's work in light of the principles laid out by John Stuart Mill. A review of Mill's essay, On Liberty, emphasizes the importance Mill placed on open and critical discussion for the vitality and progress of various aspects of human life, including the pursuit of scientific knowledge. Many of Feyerabend's more unusual stances, I suggest, are best interpreted as attempts to play certain roles—especially the role of “defender of unpopular minority opinion”—that are necessary to fulfilling Mill's conditions for rational exchange and optimal human development.

Paul Feyerabend recommended the methodological policy of proliferating competing theories as a means to uncovering new empirical data, and thus as a means to increase the empirical constraints that all theories must confront. Feyerabend's policy is here defended as a clear consequence of connectionist models of explanatory understanding and learning. An earlier connectionist “vindication” is criticized, and a more realistic and penetrating account is offered in terms of the computationally plastic cognitive profile displayed by neural networks with a recurrent architecture.

In attempting to assess the legacy of Paul Feyerabend's philosophical work, matters are complicated by the fact that there was a change in his basic orientation towards the philosophy of science around the end of the 1960s. Here I shall indicate one aspect of Feyerabend's divided legacy. My main aims are to sketch the principal themes in his (fairly extensive but little-known) 1990s output, to situate that later output insofar as it bears on the realism/antirealism debate, and (rather precipitously, perhaps) to identify what I take to be the single common premise of his entire philosophical work.

“Theoretical pluralism” obtains when there are good evidential reasons for accommodating multiple theories of the same domain. Issues of “relative significance” often arise in connection with the investigation of such domains. In this paper, I describe and give examples of theoretical pluralism and relative significance issues. Then I explain why theoretical pluralism so often obtains in biology—and why issues of relative significance arise—in terms of evolutionary contingencies and the paucity or lack of laws of biology. Finally, I turn from explanation to justification, and raise questions about the purpose and value of concerns and disagreements about relative significance.

In this paper I argue that we can best make sense of the practice of experimental evolutionary biology if we see it as investigating contingent, rather than lawlike, regularities. This understanding is contrasted with the experimental practice of certain areas of physics. However, this presents a problem for those who accept the Logical Positivist conception of law and its essential role in scientific explanation. I address this problem by arguing that the contingent regularities of evolutionary biology have a limited range of nomic necessity and a limited range of explanatory power even though they lack the unlimited projectibility that has been seen by some as a hallmark of scientific laws.

John Beatty (1995) and Alexander Rosenberg (1994) have argued against the claim that there are laws in biology. Beatty's main reason is that evolution is a process full of contingency, but he also takes the existence of relative significance controversies in biology and the popularity of pluralistic approaches to a variety of evolutionary questions to be evidence for biology's lawlessness. Rosenberg's main argument appeals to the idea that biological properties supervene on large numbers of physical properties, but he also develops case studies of biological controversies to defend his thesis that biology is best understood as an instrumental discipline. The present paper assesses their arguments.

Beatty, Brandon, and Sober agree that biological generalizations, when contingent, do not qualify as laws. Their conclusion follows from a normative definition of law inherited from the Logical Empiricists. I suggest two additional approaches: paradigmatic and pragmatic. Only the pragmatic represents varying kinds and degrees of contingency and exposes the multiple relationships found among scientific generalizations. It emphasizes the function of laws in grounding expectation and promotes the evaluation of generalizations along continua of ontological and representational parameters. Stability of conditions and strength of determination in nature govern projectibility. Accuracy, ontological level, simplicity, and manageability provide additional measures of usefulness.