A classic analytic approach to biological phenomena seeks to refine definitions until classes are sufficiently homogenous to support prediction and explanation, but this approach founders on cases where a single process produces objects with similar forms but heterogeneous behaviors. I introduce object spaces as a tool to tackle this challenging diversity of biological objects in terms of causal processes with well-defined formal properties. Object spaces have three primary components: (1) a combinatorial biological process such as protein synthesis that generates objects (...) with parts that are modular, independent, and organized according to an invariant syntax; (2) a notion of “distance” that relates the objects according to rules of change over time as found in nature or useful for algorithms; (3) mapping functions defined on the space that map its objects to other spaces or apply an evaluative criterion to measure an important quality, such as parsimony or biochemical function. Once defined, an object space can be used to represent and simulate the dynamics of phenomena on multiple scales; it can also be used as a tool for predicting higher-order properties of the objects, including stitching together series of causal processes. Object spaces are the basis for a strategy of theorizing, discovery, and analysis in biology: as heuristic idealizations of biology, they help us transform inchoate, intractable problems into articulated, well-structured ones. Developing an object space is a research strategy with a long, successful history under many other names, and it offers a unifying but not overreaching approach to biological theory. (shrink)

While the notion of chance has been central in discussions over the probabilistic nature of natural selection and genetic drift, its role in the production of variants on which populational sampling takes place has received much less philosophical attention. This article discusses the concept of chance in evolution in the light of contemporary work in evo-devo. We distinguish different levels at which randomness and chance can be defined in this context, and argue that recent research on variability and evolvability demands (...) a causal understanding of variational probabilities under which development acquires a creative, rather than a constraining role in evolution. We then provide a propensity interpretation of variational probabilities that solves a conceptual confusion between causal properties, variational probabilities and extant variation present in the literature, and explore some metaphysical consequences that follow from our interpretation, specifically with regards to the nature of developmental types. (shrink)

Formal spaces have become commonplace conceptual and computational tools in a large array of scientific disciplines, including both the natural and the social sciences. Morphological spaces are spaces describing and relating organismal phenotypes. They play a central role in morphometrics, the statistical description of biological forms, but also underlie the notion of adaptive landscapes that drives many theoretical considerations in evolutionary biology. We briefly review the topological and geometrical properties of the most common morphospaces in the biological literature. In contemporary (...) geometric morphometrics, the notion of a morphospace is based on the Euclidean tangent space to Kendall’s shape space, which is a Riemannian manifold. Many more classical morphospaces, such as Raup’s space of coiled shells, lack these metric properties, e.g., due to incommensurably scaled variables, so that these morphospaces typically are affine vector spaces. Other notions of a morphospace, like Thomas and Reif’s skeleton space, may not give rise to a quantitative measure of similarity at all. Such spaces can often be characterized in terms of topological or pretopological spaces. (shrink)

The current implementation of the Neo-Darwinian model of evolution typically assumes that the set of possible phenotypes is organized into a highly symmetric and regular space. Most conveniently, a Euclidean vector space is used, representing phenotypic properties by real-valued variables. Computational work on the biophysical genotype-phenotype model of RNA folding, however, suggests a rather different picture. If phenotypes are organized according to genetic accessibility, the resulting space lacks a metric and can be formalized only in terms of a relatively unfamiliar (...) structure. Patterns of phenotypic evolution—such as punctuation, irreversibility, and modularity—result naturally from the properties of the genotype-phenotype map, which, given the genetic accessibility structure, define accessibility in the phenotype space. The classical framework, however, addresses these patterns exclusively in terms of natural selection on suitably constructed fitness landscapes. Recent work has extended the explanatory level for phenotypic evolution from fitness considerations alone to include the topological structure of phenotype space as induced by the genotype-phenotype map. Lewontin’s notion of “quasi-independence” of characters can also be formalized in topological terms: it corresponds to the assumption that a region of the phenotype space is represented by a product space of orthogonal factors. In this picture, each character corresponds to a factor of a region of the phenotype space. We consider any region of the phenotype space that has a given factorization as a “type”, i.e., as a set of phenotypes that share the same set of phenotypic characters. Thus, a theory of character identity can be developed that is based on the correspondence of local factors in different regions of the phenotype space. (shrink)

A framework is presented in which the role ofdevelopmental rules in phenotypic evolution canbe studied for some simple situations. Usingtwo different implicit models of development,characterized by different developmental mapsfrom genotypes to phenotypes, it is shown bysimulation that developmental rules and driftcan result in directional phenotypic evolutionwithout selection. For both models thesimulations show that the critical parameterthat drives the final phenotypic distributionis the cardinality of the set of genotypes thatmap to each phenotype. Details of thedevelopmental map do not matter. If phenotypesare (...) randomly assigned to genotypes, the lastresult can also be proved analytically. (shrink)

The complexities of modern science are not adequately reflected in many bioethical discussions. This is especially problematic in highly contested cases where there is significant pressure to generate clinical applications fast, as in stem cell research. In those cases a more integrated approach to bioethics, which we call systems bioethics, can provide a useful framework to address ethical and policy issues. Much as systems biology brings together different experimental and methodological approaches in an integrative way, systems bioethics integrates aspects of (...) the history and philosophy of science, social and political theory, and normative analysis with the science in question. In this paper we outline how a careful analysis of the science of stem cell research can help to refocus the discussions related to the clinical applications of stem cells. We show how inaccurate or inadequate scientific assumptions help to create a set of unrealistic expectations and badly inform ethical deliberations and policy development. Systems bioethics offers resources for moving beyond the current impasse. (shrink)