Developmental biology is the science that investigates how a variety of interacting processes

generate an organism’s heterogeneous shapes, size, and structural features that arise on the
trajectory from embryo to adult, or more generally throughout a life cycle. It represents an
exemplary area of contemporary experimental biology that focuses on phenomena that have
puzzled natural philosophers and scientists for more than two millennia. Philosophers of
biology have shown interest in developmental biology due to the potential relevance of
development for understanding evolution, the theme of reductionism in genetic explanations,
and via increased attention to the details of particular research programs, such as stem cell
biology. Developmental biology displays a rich array of material and conceptual practices
that can be analyzed to better understand the scientific reasoning exhibited in experimental
life science. This entry briefly reviews some central phenomena of ontogeny and then
explores four domains that represent some of the import and promise of conceptual reflection
on the epistemology of developmental biology.

 1. Overview
o 1.1 Historical Considerations
o 1.2 Developmental Phenomena
o 1.3 Developmental Mechanisms
 2. The Epistemological Organization of Developmental Biology
o 2.1 No Theory of Development?
o 2.2 Erotetic Organization
 3. Explanatory Approaches to Development
o 3.1 Genetics
o 3.2 Physics
o 3.3 Integrating Approaches: Genetics and Physics
 4. Model Organisms for the Study of Development
 5. Development and Evolution
o 5.1 Functional Homology in Developmental Genetics
o 5.2 Normal Stages and Phenotypic Plasticity
 6. Conclusion
 Bibliography
o References
o Figure Credits
 Academic Tools
 Other Internet Resources
  Related Entries

1. Overview
1.1 Historical Considerations
Developmental biology is the science that investigates how a variety of interacting processes
generate an organism’s heterogeneous shapes, size, and structural features that arise on the
trajectory from embryo to adult, or more generally throughout a life cycle (Love 2008;
Minelli 2011a). It represents an exemplary area of contemporary experimental biology that
focuses on phenomena that have puzzled natural philosophers and scientists for more than
two millennia. How do the dynamic relations among seemingly homogeneous components in
the early stages of an embryo produce a unified whole organism containing heterogeneous
parts in the appropriate arrangement and with correct interconnections? More succinctly,
how do we explain ontogeny (or, more archaically, generation)? In Generation of Animals,
Aristotle provided the first systematic investigation of developmental phenomena and
recognized key issues about the emergence of and relationships between hierarchically
organized parts (e.g., bone and anatomical features containing bone), as well as the
explanatory difficulty of determining how a morphological form is achieved reliably in
offspring (e.g., the typical shape and structure of appendages in a particular species).
Generation remained a poignant question throughout the early modern period and was
explored by many key figures writing at the time, including William Harvey, René
Descartes, Robert Boyle, Pierre Gassendi, Nicolas Malebranche, Gottfried Wilhelm Leibniz,
Anne Conway, Immanuel Kant, and others (Smith 2006). Observations of life cycle
transitions, such as metamorphosis, fed into these endeavors and led to striking conclusions,
such as Leibniz’s denial of generation sensu stricto.
Animals and all other organized substances have no beginning … their apparent generation is
only a development, a kind of augmentation … a transformation like any other, for instance
like that of a caterpillar into a butterfly. (Smith 2011: 186–187)
A major theme that crystallized in this history of investigation is the distinction between
epigenesis and preformation (see the entry on theories of biological development).
Proponents of epigenesis claimed that heterogeneous, complex features of form emerge from
homogeneous, less complex embryonic structures through interactive processes. Thus, an
explanation of the ontogeny of these form features requires accounting for how the
interactions occur. Proponents of preformation claimed that complex form preexists in the
embryo and “unfolds” via ordinary growth processes. An adequate explanation involves
detailing how growth occurs. Although preformation has a lighter explanatory burden in
accounting for how form emerges during ontogeny (on the assumption that growth is easier
to explain than process interactions), it also must address how the starting point of the next
generation is formed with the requisite heterogeneous complex features. This was sometimes
accomplished by embedding smaller and smaller miniatures ad infinitum inside the organism
(Figure 1). Epigenetic perspectives were often dependent on forms of teleological reasoning
(see the entry on teleological notions in biology) to account for why interactions among
homogeneous components eventually result in a complex, integrated whole organism.
Though nothing prevents mixing features of these two outlooks in explaining different
aspects of development, polarization into dichotomous positions has occurred frequently
(Rose 1981; Smith 2006).
In the late 19th and early 20th century, the topic of development was salient in controversies
surrounding vitalism, such as the disagreement between Wilhelm Roux and Hans Driesch
over how to explain ontogeny (Maienschein 1991). Roux thought that a fertilized egg
contains inherited elements that represent different organismal characteristics. During the
process of cellular division, these elements become unequally distributed among daughter
cells leading to distinct cell fates. Driesch, in contrast, held that each cell retained its full
potential through division even though differentiation occurred. Although this issue is often
understood in terms of the metaphysics of life (vitalism versus materialism), Driesch’s
interpretation of development and the autonomy of an organism had epistemological
dimensions (Maienschein 2000). The explanatory disagreement involved different
experimental approaches and divergent views on the nature of differentiation in early
ontogeny (e.g., to what degree cells are pre-specified). A familiar philosophical theme
running through these discussions, both epistemological and metaphysical, is the status
of reductionism in biology. Through the middle of the 20th century, embryology—the
scientific discipline studying development—slowly transformed into developmental biology
with a variety of reworked and recalcitrant elements (Berrill 1961). In conjunction with the
issue of reductionism, a key aspect of this history is the molecularization of experimental (as
opposed to comparative) embryology (Fraser and Harland 2000), with a concomitant
emphasis on the explanatory power of genes (see the entry on gene and Section 3.1). This
complex and fascinating history, including interrelations with medicine and reproductive
technology, has been detailed elsewhere (see, e.g., Oppenheimer 1967; Horder et al. 1986;
Hamburger 1988; Hopwood 2019; Maienschein 2014; Maienschein et al. 2005; Gilbert 1991;
Embryo Project in Other Internet Resources).
Developmental biology has increasingly become an area of exploration for philosophy of
biology due to the potential relevance of development for understanding evolution (Love
2015; Section 5), the theme of reductionism in biology and explanations from molecular
genetics (Robert 2004; Rosenberg 2006; Section 3), and via increased attention to the details
of particular research programs, such as stem cell biology (Fagan 2013; Laplane 2016).
However, it should not be forgotten that ontogeny was on the radar of philosophical scholars
in the 20th century, as seen in Ernest Nagel’s treatment of hierarchical organization and
reduction in the development of living systems (Nagel 1961: 432ff). For contemporary
philosophy of science, developmental biology displays a rich array of material and
conceptual practices that can be analyzed to better understand the scientific reasoning
exhibited in experimental life science (see the entry on experiment in biology). After a brief
review of some central phenomena of ontogeny, this entry explores four domains that
represent some of the import and promise of conceptual reflection on the epistemology of
developmental biology.

1.2 Developmental Phenomena

Developmental biology is the science that seeks to explain how the structure of organisms
changes with time. Structure, which may also be called morphology or anatomy,
encompasses the arrangement of parts, the number of parts, and the different types of parts.
(Slack 2006: 6)
Most of the properties that developmental biologists attempt to explain are structural rather
than functional. For example, a developmental biologist concentrates more on how tissue
layers fold or how shape is generated than on what the folded tissue layers do or how the
shape functions. The ontogeny of function, at all levels of organization, is an element of
developmental biology, but it is often bracketed because of the predominance (both past and
present) of questions surrounding the ontogeny of form or structure (Love 2008).
Textbooks (e.g., Gilbert 2010; Slack 2013; Wolpert et al. 2010) typically describe a
canonical set of events surrounding the changing structures displayed during animal
development.[1] The first of these is fertilization (in sexually reproducing species), where an
already semi-organized egg merges with a sperm cell, followed by the fusion of their nuclei
to achieve the appropriate complement of genetic material. Second, the fertilized egg
undergoes several rounds of cleavage, which are mitotic divisions without cell growth that
subdivide the zygote into many distinct cells (Figure 2). After many rounds of cleavage, this
spherical conglomerate of cells (now called a blastula) begins to exhibit some specification
of germ layers (endoderm, mesoderm, and ectoderm) and then proceeds to invaginate at one
end, a complex process referred to as gastrulation that eventually yields a through-gut. All
three germ layers, from which specific types of cells are derived (e.g., neural cells from
ectoderm), become established during gastrulation or shortly after it completes.
[2]
 Organogenesis refers to the production of tissues and organs through the interaction and
rearrangement of cell groups. Events confined to distinct taxonomic groups
include neurulation in chordates, whereas other events correlate with mode of development
(metamorphosis from a larval to adult stage) or individual trauma (regeneration of a limb).
Several key processes underlie these distinct developmental events and the resulting features
of form that emerge (e.g., the through-gut formed subsequent to gastrulation or the heart
formed during organogenesis). These are critical to the ontogeny of form and link directly to
major research questions in developmental biology (Section 2). First, cellular properties,
such as shape, change during ontogeny. This is a function of differentiation whereby cells
adopt specific fates that include shape transformations (Figure 3). Second, regions of cells in
the embryo are designated through arrangement and composition alterations that correspond
to different axes in different parts of the embryo (e.g., dorsal-ventral, anterior-posterior, left-
right, and proximal-distal). The successive establishment of these regions is referred to
as pattern formation. Third, cells translocate and aggregate into layers (e.g., endoderm and
ectoderm, followed by the mesoderm in many lineages) and later tissues (aggregations of
differentiated cell types). Fourth, cells and tissues migrate and interact to produce new
arrangements and shapes composed of multiple tissue layers with novel functions (i.e.,
organs). These last two sets of processes are usually termed morphogenesis (Davies 2005)
and occur via many distinct mechanisms (Section 1.3). Fifth, there is growth in the size of
different form features in the individual, remarkably obvious when comparing zygote to
adult, although proportional change between different features (allometry) is also striking.
None of these processes occur in isolation and explanations of particular form features
usually draw on several of them simultaneously, presuming other features that originated
earlier in ontogeny by different instantiations and combinations of the processes. This sets a
broad agenda for investigation: how do various iterations and combinations of these
processes generate form features during ontogeny? Consider the concrete example of
vertebrate cardiogenesis. How does the vertebrate heart, with its internal and external shape
and structure originate during ontogeny (Harvey 2002)? How does the heart come to exhibit
left/right asymmetry in the body cavity? What causes cells to adopt a muscle cell fate or
certain tissues to interact in the prospective region of the heart? How do muscle cells migrate
to, aggregate in, and differentiate at the correct location? How does the interior of the heart
adopt a particular tubular structure with various chambers (that differs among vertebrate
species)? How does the heart grow at a particular rate and achieve a specific size? Solutions
relevant to explaining the ontogeny of form characterize causal factors that account for how
different processes occur and yield various outcomes (Section 3).

1.3 Developmental Mechanisms

A developmental mechanism is a mechanism or process that operates during ontogeny (see
McManus 2012 for discussion). At least two different types of developmental mechanisms
can be distinguished (Love 2017a): molecular genetic mechanisms (signaling or gene
regulatory networks; Section 3.1) and cellular-physical mechanisms (cell migration or
epithelial invagination; Section 3.2). Philosophical explorations of mechanisms in science
and mechanistic explanation have grown dramatically over the past two decades (Craver and
Darden 2013; Glennan and Illari 2017; Illari and Williamson 2012). Among different
accounts of scientific mechanisms, four shared elements are discernable: (1) what a
mechanism is for, (2) its constituents, (3) its organization, and, (4) the spatiotemporal context
of its operation. Developmental explanations seek to characterize these four elements
through various experimental interventions. Together these elements provide a template for
characterizing the two different types of developmental mechanisms.
A well-established molecular genetic mechanism is the initial formation of segments
in Drosophila due to the segment polarity network of gene expression (Wolpert et al. 2010,
70-81; Damen 2007). By Stage 8 of development (~3 hours post-
fertilization), Drosophila embryos have 14 parasegment units that were defined by pair-rule
gene expression in earlier stages. The transcription factor Engrailed accumulates in the
anterior portion of each parasegment. This initiates a cascade of gene activity that defines the
boundaries of each compartment of cells that will eventually become a segment. One element
of this activity is the expression of hedgehog, a secreted signaling protein, in cells anterior to
the band of cells where Engrailed has accumulated, which marks the posterior boundary of
each nascent segment. This, in turn, activates the expression of wingless, another secreted
signaling protein, which maintains the expression of both engrailed and hedgehog in a
feedback loop so that segment boundaries persist (Figure 4). The segment polarity network
exhibits all four of the shared elements of a mechanism. It is constituted by a number of parts
(e.g., Engrailed, Wingless, Hedgehog) and activities or component operations (e.g., signaling
proteins bind receptors, transcription factors bind to DNA and initiate gene expression),
which are organized into patterns of interacting relationships (feedback loops, signaling
pathways) within a spatiotemporal context (in parasegments of the Drosophila embryo, ~3
hours post-fertilization) so as to produce a specific behavior or phenomenon (a set of distinct
segments with well-defined boundaries).

Next, consider the cellular-physical mechanism of branching morphogenesis, which refers to

combinations of cellular proliferation and movement that yield branch-like structures in
kidneys, lungs, glands, or blood vessels. There are many types of branching morphogenesis,
but one primary mechanism is epithelial folding, which involves cells invaginating at
different locations on a structure to yield branches (Davies 2013, ch. 20). Different cellular-
physical mechanisms can produce invaginations that lead to branching structures (Varner and
Nelson 2014): the constriction of one end of a subset of columnar cells in an epithelium
(“apical constriction”); increased cell proliferation of one epithelial sheet in relation to
another (“differential growth”); and compression of an epithelium leading to periodic
invaginations (“mechanical buckling”). That different mechanisms can lead to the same
morphological outcome means it can be difficult to discern which mechanism is operating in
an embryonic context. Branching morphogenesis also exhibits all four of the shared elements
of a mechanism. The parts are cells and tissues with activities or component operations (e.g.,
apical constriction, differential growth, mechanical buckling) being organized into patterns
of interacting relationships (apical constriction leading to epithelial invagination) within a
spatiotemporal context (in tracheal precursors within the Drosophila embryo around Stage 7
and 8). This organization produces a specific behavior or phenomenon (a set of branching
structures—the trachea).
Once these types of developmental mechanisms have been distinguished, several conceptual
issues become salient. The first pertains to how the two types of mechanisms are interrelated
during ontogeny, and how different investigative approaches do or do not successfully
provide integrated accounts of them (Section 3.3). A second is their distinct patterns of
generality. Molecular genetic mechanisms are widely conserved across phylogenetically
disparate taxa as a consequence of evolutionary descent, whereas cellular-physical
mechanisms are widely instantiated as a consequence of shared physical organization but not
due to evolutionary descent (Love 2017a). The divergence of these patterns has prompted
explicit epistemological reflection by developmental biologists. [3]

2. The Epistemological Organization of

Developmental Biology
One recurring theme in the long history of investigations into development is that explaining
the ontogeny of form consists of many interrelated questions about diverse phenomena
(Section 1.2). Sometimes philosophers have attempted to compress these questions into one
broad problem.
The real question concerning metazoan ontogeny is just how a single cell gives rise to the
requisite number of differentiated cell lineages with all the right inductive developmental
interactions required to reproduce the form of the mature organism (Moss 2002: 97).
The central problem of developmental biology is to understand how a relatively simple and
homogeneous cellular mass can differentiate into a relatively complex and heterogeneous
organism closely resembling its progenitor(s) in relevant aspects (Robert 2004: 1).
This language is not necessarily incorrect but can lead to skewed interpretations. For
example, Philip Kitcher has argued that:
In contemporary developmental biology, there is … uncertainty about how to focus the big,
vague question, How do organisms develop? (Kitcher 1993: 115)
This is simply false. While it is true that these questions have been manifested with differing
frequency and vigor through history, and the ability to answer them (as well as the nature of
the questions themselves) has been contingent on different research strategies and methods,
the issue has not been an unwieldy central problem. But scrutinizing the structure of
developmental biology’s questions is not merely an exercise in clarification. It is crucial for
understanding how the science of developmental biology is organized.

2.1 No Theory of Development?

Although it is common in philosophy to associate sciences with theories, such that the
individuation of a science is dependent on a constitutive theory or group of models, it is
uncommon to find presentations of developmental biology that make reference to a theory of
development (see discussion in Minelli and Pradeu 2014). Instead, we find references to
families of approaches (developmental genetics, experimental embryology, cell biology,
and molecular biology) or catalogues of “key molecular components” (transcription factor
families, inducing factor families, cytoskeleton or cell adhesion molecules, and extracellular
matrix components). No standard theory or group of models provides theoretical scaffolding
in the major textbooks (e.g., Slack 2013; Wolpert et al. 2010; Gilbert 2010). The absence of
any reference to a constitutive theory of development or some set of core explanatory models
is prima facie puzzling. Three interpretations of this situation are possible: (a) despite the
lack of reference to theories, one can reconstruct a theory (or theories) of developmental
biology out of the relevant discourse (e.g., multiple allied molecular models); (b) the lack of
reference to theories indicates an immaturity in developmental biology because mature
sciences always have systematic theories; and, (c) the lack of reference to theories should be
taken at face value.
Developmental biology is not an immature science, groping about for some way to explain
its phenomena: “some of the basic processes and mechanisms of embryonic development are
now quite well understood” (Slack 2013: 7). The impetus for this type of interpretation arises
out of commitments to a conception of mature science that presumes theories are abstract
systems with a small set of laws or core principles (see the entry on the structure of scientific
theories). On the other hand, holding that developmental biology already has a theory
costumed in different guise—not referred to as such by developmental biologists—is a
possible interpretation. It arises out of a view that sciences must have theories, which has
been expanded to allow for different understandings of theory structure, such as
constellations of models without laws, even though the assumption is that theory still plays a
similar organizing role in guiding research. However, this assumption should be challenged
and rejected on methodological grounds in the case of developmental biology. An analysis of
the reasoning in a science should exhibit epistemic transparency and not postulate “hidden”
reasoning structure (Love 2012). This criterion is based on the premise that the basis of
successes in scientific inquiry must be available to those engaged in its practice (i.e.,
scientists). If we postulate hidden structure not present in scientific discourse to account for
inductive inference, explanation, or other forms of reasoning, then we risk obscuring how
scientists themselves access this structure to evaluate it (Woodward 2003: ch. 4). The
successes of developmental biology would become mysterious when viewed from the
vantage point of its participants.
Epistemic transparency demands a descriptive correspondence between philosophical
accounts of science and scientific practice. This does not mean that every claim made by any
scientist should be taken with the same credence. A ruling concern is pervasive features of
practice. The problem with assuming laws are required for explanation is their relative
absence from a variety of successful sciences routinely offering explanations, not that no
scientist ever appeals to laws as explanatory. Pervasive features of scientific practice should
be prominent in philosophical accounts of sciences. Thus, it is not surprising that the desire
for a theory can be found among some developmental biologists: “Developing a theory is of
utmost importance for any discipline” (Sommer 2009: 417). But the fact that these calls are
rare means we should not assume theories are actually needed to govern and organize inquiry
within the domain.[4]
It was once thought that each science must have laws in order to offer explanations (see the
entry on scientific explanations), but now this is seen as unnecessary (Giere 1999;
Woodward 2003). The expectation that a science have a theory to accomplish the task of
organizing and guiding inquiry is of similar vintage. It derives from an intuitive expectation
of what counts as a mature science in the first place. Even if we find empirically successful
and coherent traditions of research without a systematic theoretical framework providing
guidance, then the science cannot be mature. One might shrug off these quasi-positivist
appeals to maturity by invoking more flexible conceptions of theory and theory structure.
But why retain the expectation that theories should accomplish the same epistemic tasks? It
is a preconception about knowledge structure that is not plausible in light of the diversity of
research practices found across the sciences. The few scientists who favor this philosophical
response have different motivations. Instead of maturity, other reasons are salient, such as
guidance in the face of a welter of biochemical detail or the need to forge a synthesis
between evolution and development.[5]
Developmental biologists recognize that the “curse of detail” is one of the costs of
developmental biology’s meteoric success over the past three decades: “The principal
challenge today is that of exponentially increasing detail” (Slack 2013: ix). While something
must provide organization and guidance to developmental biology, it need not be theories
that accomplish the task. Regarding calls for a synthesis of evolution and development, these
often assume that having a developmental theory is a precondition for synthesis (Sommer
2009): “Our troubles … derive from our standing lack of an explicit theory of development”
(Minelli 2011a: 4). However, this line of argument relies on the degree to which evolutionary
theory exhibits the supposed structure to which developmental biologists should aspire. The
actual practice associated with evolutionary theory indicates a more flexible framework with
chameleon qualities that is responsively adjusted to the diverse investigative aims of
evolutionary researchers (Love 2013). Therefore, it is not clear that evolutionary theory
supplies the preferred template. A productive way forward is to relinquish the prior
expectation that sciences must have theories of a certain kind to govern and guide their
activity. Instead, sciences that display empirical success and fecundity should be studied to
discover what features are responsible, without assuming that those features will be the same
for all sciences: “Science need not be understood in these terms and, indeed, may be better
understood in other terms” (Giere 1999: 4).

2.2 Erotetic Organization

The criterion of epistemic transparency (Section 2.1) encourages an exploration of our third
interpretive option—the lack of reference to theories should be taken at face value.
Developmental biology is organized primarily by stable, broad domains of problems that
correspond to abstract representations of major ontogenetic processes (differentiation, pattern
formation, growth, and morphogenesis; Section 1.2). Yet how do we interpret the
“theoretical” aspects of developmental biology (e.g., positional information models of
pattern formation) and the utilization of theories from other domains (e.g., biochemistry)?
One way is to distinguish between theory-informed science—using theoretical knowledge—
and theory-directed science—having a theory that directs inquiry and organizes knowledge
(Waters 2007b); developmental biology is theory-informed but not theory-directed. Theories
need not be wholly absent from developmental biology but—when present—they play roles
very different from standard philosophical expectations. Developmental biology uses
theoretical knowledge from biochemistry when appealing to morphogen gradients to explain
how segments are established or chemical thermodynamics when invoking reaction–
diffusion mechanisms to explain pigmentation patterns. It also uses theoretical knowledge
derived from within developmental biology, such as positional information models. Different
kinds of theory inform developmental biology, but these do not organize research—they are
not necessary to structure the knowledge and direct investigative activities. Developmental
biologists are not focused on confirming and extending the theory of reaction–diffusion
mechanisms, nor are they typically organizing their research around positional information.
[6]
 This theoretical knowledge is used in building explanations but does not provide rails of
guidance for how to proceed in a research program. All sciences may use theoretical
knowledge, but this is not the same as all sciences having a theory providing direction and
organization.
Why think that problems provide organizational architecture for the epistemology of
developmental biology? They are a pervasive feature of its reasoning practices, illustrated in
textbooks that capture substantial community consensus about standards of explanation,
experimental methods, essential concepts, and empirical content. Unlike evolutionary
biology textbooks that discuss the theory of natural selection or economics textbooks that
talk about microeconomic theory, an examination of several editions of major textbooks
indicate that developmental biology does not have similar kinds of theories.
Jonathan Slack’s Essential Developmental Biology (Slack 2006, 2013) is organized around
four main types of processes, also described as clustered groups of problems, which occur
during embryonic development: regional specification (pattern formation), cell
differentiation, morphogenesis, and growth. These broad clusters are then fleshed out along a
standard timeline of early development, highlighting gametogenesis, fertilization, cleavage,
gastrulation, and axis specification (see Section 1.2). Different experimental approaches (cell
and molecular biology, developmental genetics, and experimental embryology) are utilized
in a specific set of model organisms (see below, Section 4) to dissect the workings of these
developmental phenomena. Subsequent chapters cover later aspects of development (e.g.,
organogenesis), with different systems treated in depth by tissue layer, differentiation and
growth, or in relation to evolutionary questions (see below, Section 5). Throughout this
presentation, no specific theory, set of hypotheses, or dominant model is invoked to organize
these different domains of investigation. Instead, broad clusters of questions that reflect
generally delineated processes (differentiation, specification, morphogenesis, and growth) set
the agenda of research.
Scott Gilbert’s Developmental Biology exhibits a similar pattern (Gilbert 2000 [2003, 2006,
2010]). Developmental biology is constituted by two broad questions (“How does the
fertilized egg give rise to the adult body? And how does that adult body produce yet another
body?”), which can then be subdivided into further categories, such as differentiation,
morphogenesis, growth, reproduction, regeneration, evolution, and environmental regulation.
These questions can be parsed more analytically in terms of five variables: abstraction,
variety, connectivity, temporality, and spatial composition. The values given to these
variables structure the constellation of research questions within the broad problem agendas
corresponding to generally delineated processes. For example, research questions oriented
around events in zebrafish gastrulation are structured in a way that differs from the research
questions oriented around vertebrate neural crest cell migration because they involve
different values for the five variables: abstraction (zebrafish vs. vertebrates), temporality
(earlier vs. later), spatial composition (tissue layer interactions vs. a distinctive population of
cells), variety (epiboly vs. epithelium to mesenchyme transition), and connectivity (gut
formation and endoderm vs. organogenesis and ectoderm/mesoderm). These configurations
can be adjusted readily in response to shifts in the values for different variables (Love 2014).
[7]

This anatomy of problems, with explicit epistemological structure derived from different
values for these variables, operates to organize the science of development. Investigators
from different disciplines can be working on the same problem but asking different questions
that require distinct but complementary methodological resources. Knowledge and inquiry in
developmental biology are intricately organized, just not by a central theory or group of
models, and this erotetic organization is epistemologically accessible to the participating
scientists. While theoretical knowledge, especially that drawn from molecular biological
mechanisms (see the entry on molecular biology) and mathematical models (e.g., reaction–
diffusion models) is ubiquitous (theory-informed), the clusters of problems that reappear
across the textbooks and correspond to different types of processes provide the governing
architecture (not theory-directed), which can be characterized explicitly according to the
variables described. Further analysis of this problem anatomy is possible, including how it is
displayed in regular research articles and not just textbooks, as well as other areas of biology
(see, e.g., Brigandt and Love 2012).

3. Explanatory Approaches to Development

Explanations in developmental biology are usually causal, though unlike standard
mechanistic explanation there is a constant acquisition of new causal capacities (in terms of
constituent entities, activities, and their organization) through development (McManus 2012;
Parkkinen 2014). Although much work remains in characterizing different aspects of
explanation in developmental biology, there is no doubt that a difference making or
manipulability conception of causation (see the entry on causation and manipulability)
provides a core element of the reasoning (Woodward 2003; Strevens 2009; Waters 2007a).
Genetic explanations of development (Section 3.1), similar to what is seen in molecular
genetics, work by identifying changes in the expression of genes and interactions among
their RNA and protein products that lead to changes in the properties of morphological
features during ontogeny (e.g., shape or size), while holding a variety of contextual variables
fixed. More recently, there has been growing interest in physical explanations of
development (Section 3.2) that involve appeals to mechanical forces due to geometrical
arrangements of mesoscale materials, such as fluid flow (Forgacs and Newman 2005).
Researchers agree on the phenomena that need to be explained (Section 1.2 and Section 2.2),
but differ on whether physical rules or genetic factors are more or less explanatory (Keller
2002).[8] The existence of two different types of causal explanations for developmental
phenomena poses an additional question about how they might be combined into a more
integrated explanatory framework (Section 3.3).

3.1 Genetics
Many philosophers have turned to explanations of development over the past two decades in
an effort to esteem or deflate claims about the causal power of genes (Keller 2002;
Neumann-Held and Rehmann-Sutter 2006; Rosenberg 2006; Robert 2004; Waters 2007a).
[9]
 Genetic explanations touch the philosophical theme of reductionism and appear to
constitute the bulk of empirical success accruing to developmental biology over the past
several decades.[10] Statements from developmental biologists reinforce this perspective:
Developmental biology … deals with the process by which the genes in the fertilized egg
control cell behavior in the embryo and so determine its pattern, its form, and much of its
behavior … differential gene activity controls development. (Wolpert et al. 1998: v, 15)
These types of statements are sometimes amplified in appeals to a genetic program for
development.
[Elements of the genome] contain the sequence-specific code for development; and they
determine the particular outcome of developmental processes, and thus the form of the
animal produced by every embryo. … Development is the execution of the genetic program
for construction of a given species of organism (Davidson 2006: 2, 16). [11]
At other times, statements concentrate on genetics as the primary locus of causation in
ontogeny: “Developmental complexity is the direct output of the spatially specific expression
of particular gene sets and it is at this level that we can address causality in development”
(Davidson and Peter 2015: 2). Whether or not these statements can be substantiated has been
the subject of intense debate.[12] The strongest claims about genetic programs or the genetic
control of development have empirical and conceptual drawbacks that include an inattention
to plasticity and the role of the environment, an ambiguity about the locus of causal agency,
and a reliance on metaphors drawn from computer science (Gilbert and Epel 2009; Keller
2002; Moss 2002; Robert 2004). However, this leaves intact the difference-making principle
of genetic explanation exhibited in molecular genetics (Waters 2007a), which yields more
narrow and precise causal claims under controlled experimental conditions, and is applicable
to diverse molecular entities that play causal roles during development, such as regulatory
RNAs, proteins, and environmental signals. We can observe this briefly by reconsidering the
example of vertebrate cardiogenesis (Section 1.2).
Are there problems with claiming that genes contain all of the information (see the entry
on biological information) to form vertebrate hearts? Is there a genetic program in the DNA
controlling heart development? Are genes the primary supplier and organizer of material
resources for heart development, largely determining the phenotypic outcome? Existing
studies of heart development have identified a role for fluid forces in specifying the internal
form of the heart (Hove et al. 2003) and its left/right asymmetry (Nonaka et al. 2002).
Biochemical gradients of extracellular calcium are responsible for activating the asymmetric
expression of the regulatory gene Nodal (Raya et al. 2004) and inhibition of voltage
gradients scrambles normal asymmetry establishment (Levin et al. 2002). Mechanical cues
such as microenvironmental stiffness are crucial for key transitions from migratory cells into
organized sheets during heart formation (Jackson et al. 2017). A number of genes are clearly
difference makers in these processes (Asp et al. 2019; Srivastava 2006; Brand 2003; Olson
2006), but the conclusion that genes carry all the information needed to generate form
features of the heart seems unwarranted. While it may be warranted empirically in some
cases to privilege DNA sequence differences as causal factors in specific processes of
ontogeny (Waters 2007a), such as hierarchically organized networks of genetic difference
makers explaining tissue specification (Peter and Davidson 2011), the diversity of entities
appealed to in molecular genetics and the extent of their individual and joint roles in
specifying developmental outcomes implies that debates about the meaning, scope, and
power of genetic explanations will continue (Griffiths and Stotz 2013). However, a shift
away from genetic programs and genetic determinism to DNA, RNA, and proteins as
difference makers that operate conjointly suggests that we conceptualize other causal factors
in a similar way.

3.2 Physics
Fluid flow, as a physical force, is also a difference maker during the development of the
heart, and ontogeny more generally, and developmental biologists appeal to physical
difference makers, which are understood as factors in producing the morphological
properties of developmental phenomena (Forgacs and Newman 2005). A physical causation
approach was on display in the late 19th century work of Wilhelm His (Hopwood 1999,
2000; Pearson 2018) and especially visible in the early 20th century work of D’Arcy
Thompson and others (Thompson 1992 [1942]; Keller 2002: ch. 2; Olby 1986). This
occurred in the milieu of increasing attention to the chromosomal theory of inheritance and
attempts to explore developmental phenomena via classical genetic methods (Morgan 1923,
1926, 1934). Thompson appealed to differential rates of growth and the constraints of
geometrical relationships to explain how organismal morphology originates. Visual
representations of abiotic, mechanical analogues provided the plausibility, such as the shape
of liquid splashes or hanging drops for the cup and bell configurations of the free-swimming
sexual stage of jellyfish. If physical forces generated specific morphologies in viscoelastic
materials, then analogous morphologies in living species should be explained in terms of
physical forces operating on the viscoelastic materials of the developing embryo. Yet
morphogenetic processes that produce the shape and structure of morphology have been seen
primarily, if not exclusively, in terms of genetics for the last half-century. Physical
approaches moved into the background as molecular genetics approaches went from strength
to strength (Fraser and Harland 2000).
The molecularization of experimental embryology is one of the most striking success stories
of contemporary biology as genes and genetic interactions (e.g., in transcriptional networks
and signaling pathways; see Section 1.3) were discovered to underlie specific details of
differentiation, morphogenesis, pattern formation, and growth when structure originates
during development. Genetic approaches predominate in contemporary developmental
biology and physical modes of causation are often neglected. The frustration among
researchers interested in physical causation during embryogenesis has been palpable.
To the molecular types, a cause is a molecule or a gene. To explain a phenomenon is to
identify genes and characterize proteins without which the phenomenon will fail or be
abnormal. A molecule is an explanation: a force is a description; to argue otherwise brings
pity, at best (Albert Harris to John Trinkaus, 12 March 1996; Source: Marine Biological
Laboratory Library Archives).
Despite this predominance of genetic explanatory approaches and the frustration among
researchers utilizing other approaches, a groundswell of interest has been building around
physical explanations of development, especially in terms of their integration with genetic
explanations (Miller and Davidson 2013; Newman 2015). Some philosophers have argued
that the biomechanical modeling of physical causal factors constitutes a rejection of certain
forms of reductive explanation in biology (Green and Batterman 2017).

3.3 Integrating Approaches: Genetics and Physics

Thompson held that physical forces were explanatory but inadequate in isolation to account
for the developmental origin of morphology; heredity (genetics) was also a necessary causal
factor.[13] Yet Thompson was quick to highlight that mechanical modes of causation might be
neglected in the midst of growing attention to heredity (genetics):
it is no less of an exaggeration if we tend to neglect these direct physical and mechanical
modes of causation altogether, and to see in the characters of a bone merely the results of
variation and of heredity. (Thompson 1992 [1942]: 1023)
Despite this latter form of exaggeration manifesting itself through much of the 20 th century,
an agenda to combine or integrate the two approaches is now explicit. [14]
There is no controversy about whether genetic and physical modes of causation are at work
simultaneously:
both the physics and biochemical signaling pathways of the embryo contribute to the form of
the organism. (Von Dassow et al. 2010: 1)
They are not competing causal explanations of the same phenomenon. Explanations should
capture how their productive interactions yield developmental outcomes:
an increasing number of examples point to the existence of a reciprocal interplay between
expression of some developmental genes and the mechanical forces that are associated with
morphogenetic movements. (Brouzés and Farge 2004: 372)
Genetic causes can lead to physical causation and vice versa. Physical causation brings about
genetic causation through mechanotransduction. Stretching, contraction, compression, fluid
shear stress, and other physical dynamics are sensed by different molecular components
inside and outside of cells that translate these environmental changes into biochemical
signals (Hoffman et al. 2011; Wozniak and Chen 2009). Genetic causation brings about
physical causation by creating different physical properties of cells and tissues through the
presence, absence, or change in frequency of particular proteins. For example, different
patterns of expression for cell adhesion molecules (e.g., cadherins) can lead to differential
adhesion across epithelial sheets of tissue and thereby generate phase separations or
compartments via surface tension variation (Newman and Bhat 2008). If these modes of
causation are not competing, then how might one combine genetic and physical difference
makers into an integrated causal explanation? How much explanatory unity can be achieved
for this “reciprocal interplay”?
Finding philosophical models for the explanatory integration of genetics and physics remains
an open question (Love 2017b). Apportioning causal responsibility in the sense of
determining relative contributions (e.g., the composition of causal magnitudes among
different physical forces in Newtonian mechanics) is problematic because this requires
commensurability with respect to how causes produce their effects (Sober 1988). In the
context of causation understood in terms of difference makers, the difficulty of integration is
a variation on a problem in causal reasoning identified by John Stuart Mill and labeled the
“intermixture of effects,” which involves multiple causes contributing in a blended fashion to
yield an outcome.
This difficulty is most of all conspicuous in the case of physiological phenomena; it being
seldom possible to separate the different agencies which collectively compose an organized
body, without destroying the very phenomena which it is our object to investigate. (Mill
1843 [1974]: 456 [book 3, chapter 11, section 1, paragraph 7])
Careful statistical methodology in experiments can answer whether one type of difference
maker accounts for more of the variation in the effect variable for a particular population.
But a ranking of causal factors with respect to how much of a difference they made is not the
same as combining two modes of causation into an integrated account. Another response is
to dissolve the integration problem by reducing all of the causal interactions to one of the
two distinct modes, thereby achieving a kind of explanatory unity (Rosenberg 2006).
However, this approach is eschewed by working biologists who take both genetic and
physical modes of causation as significant and not reducible one to the other.
A different strategy is integrative pluralism (Mitchell 2002). This involves a two-step
procedure for explaining complex phenomena whose features are the result of multiple
causes: (a) formulate idealized models where particular causal factors operate in isolation
(“theoretical modeling”); and, (b) integrate idealized models to explain how particular,
concrete phenomena originate from these causes in combination. This model is suggestive
but also has key drawbacks that include the fact that genetic causal reasoning in
developmental biology does not typically involve theoretical modeling and the precise nature
of the integration is underspecified. Integration of genetic and physical difference makers in
a single mechanism offers a further possibility (Darden 2006; Craver 2007). Although this
valuably highlights the productive continuity between difference makers through stages in a
sequence (i.e., their reciprocal interplay), it also has handicaps. These include:
1. Divergent approaches to measuring time. Instead of time “in the mechanism,” time is
measured with external standardized stages (see below, Section 5.2). Stages facilitate
the study of different kinds of developmental mechanisms, with different
characteristic rates and durations for their stages, within a common framework for a
model organism (e.g., Drosophila), while also permitting conserved molecular
mechanisms to be studied in different species because the corresponding mechanism
description is not anchored to the temporal sequence of the model organism.
2. An expectation that mechanism descriptions “bottom-out” in lowest level activities of
molecular entities (Darden 2006). In the case of combining genetic and physical
difference makers, the reciprocal interplay means that there is a studious avoidance of
bottoming out in one or the other mode of causation.
3. The requirement of stable, compositional organization for mechanisms:
Mechanistic explanations are constitutive or componential explanations: they
explain the behavior of the mechanism as a whole in terms of the organized
activities and interaction of its components. (Craver 2007: 128)
But these mechanism descriptions are often embedded in different developmental
contexts (at different times in ontogeny) with distinct compositional relations (within
and between species). The reciprocal interplay between genetic and physical
difference makers is not maintained precisely because these compositional differences
 alter relationships of physical causation (fluid flow, tension, etc.; see Section 1.3).
Developmental biologists have been able to generalize relationships of genetic
causation (in terms of genetic mechanisms; see Section 1.3) across species quite
widely, but the attempt to combine these with physical causation has necessitated
narrowing the scope of the causal claims.
Adequate philosophical models of the systematic dependence between genetic and physical
difference makers in ontogeny need to account for how the temporal relations necessary for
making causal claims are anchored in an external periodization used by developmental
biologists. The imposition of different temporal scales can lead to distinct factors being
significant or salient, which matters for ascertaining how different kinds of causes can be
combined into integrated explanations. One possibility is to juxtapose these difference
makers at distinct stages via experimental verification such that they exhibit productive
continuity within the constraints of the external periodization (Love 2017b). This facilitates
representing symmetry between causal factors because genetic difference makers can be
placed before or after physical difference makers (and vice versa). Although this does not
provide a way to combine causal magnitudes (as in vector addition from Newtonian
mechanics), it offers an explicit strategy for assigning responsibility among different kinds of
causes through the vehicle of temporal organization that goes beyond ranking difference
makers. The periodization serves as a template from the practices of developmental
biologists for providing wholeness or unity to the different modes of causation to yield a kind
of integrated explanation of the morphology that results from a sequence of developmental
processes.
Not all types of causal explanation involve an external periodization and there are other ways
to combine causes in order to produce more integrated explanatory frameworks. One area
where combined explanations for developmental phenomena are being analyzed pertains to
mechanism descriptions and mathematical modeling in systems biology (Brigandt 2013;
Fagan 2013). For example, Fagan (2013: ch. 9) shows how an integrated explanation
emerges from a step-wise procedure that starts with a detailed description of a molecular
mechanism followed by the formulation of an abstracted wiring diagram of component
interactions, which is then translated into a system of equations that can account for changes
in component interactions over time. Solutions to these systems of equations and a mapping
of solutions for the interactions among components of the system onto the behavior of the
overall system within a shared landscape representation more systematically explains cellular
differentiation.

4. Model Organisms for the Study of Development

Model organisms are central to contemporary biology and studies of embryogenesis (Ankeny
and Leonelli 2011; Steel 2008; Bier and McGinnis 2003; Davis 2004). Biologists utilize only
a small number of species to experimentally elucidate various properties of ontogeny
(e.g., C. elegans, Drosophila, and Brachydanio [zebrafish]; see Figure 5). These
experimental models permit researchers to investigate development in great depth and
facilitate a precise dissection of causal relationships. Critics have questioned whether these
models are good representatives of other species because of inherent biases involved in their
selection, such as rapid development and short generation time (Bolker 1995), and
problematic presumptions about the conservation of gene functions and regulatory networks
(Lynch 2009). For example, C. elegans embryogenesis is not representative of nematodes in
terms of pattern formation and cell specification (Schulze and Schierenberg 2011) and
zebrafish appendage formation is a poor proxy for the development of appendages in
tetrapods (Metscher and Ahlberg 1999).
One response to this criticism is to emphasize the conserved genetic mechanisms shared by
all animals despite differences in developmental phenomena (Gerhart and Kirschner 2007;
Ankeny and Leonelli 2011; Weber 2005). Fruit flies may be unrepresentative in exhibiting
syncytial development, but they use the collinear expression of Hox genes to specify their
anterior-posterior body axis. This response indicates that whether an entire model organism
is representative per se is too coarse-grained a criterion to capture the rationale behind their
use. We have to ask about representation with respect to what, and some accounts have
moved in this direction. Jessica Bolker has distinguished exemplary and surrogate modes of
representation (Bolker 2009), where the former serve basic research by exemplifying a larger
group and the latter correspond to models designed to provide indirect experimental access to
otherwise inaccessible phenomena, such as mouse models of human psychological disorders
(e.g., depression). Surrogate models are adopted in biomedical contexts where the
phenomena of interest are manifested in humans. Most developmental biologists consider
model organisms as exemplars, not surrogates.[15] Thus, in order to respond to a criticism of
non-representativeness, the criterion of representation must be explored in more detail. [16]
A basic presumption about model organisms is that they bear appropriate similarity
relationships to larger groups of animals. This presumption is an instantiation of what is
discussed generally for models in science meant to represent phenomena. Model organisms
represent developmental phenomena in species that are either studied little or never studied
at all: “we study flies and frogs as examples for the development of animals in general”
(Nüsslein-Volhard 2006: 87). One source of confidence in treating them as exemplars
derives from an inductive inference over discovered patterns of evolutionary conservation
with respect to developmental phenomena (e.g., gastrulation or somite formation). If all or
most model organisms share a developmental feature, then all or most animals will share the
feature. This inference can be circumscribed more or less narrowly (e.g., if all or most
vertebrate model organisms share somite formation, then all or most vertebrates will share
it).
As a consequence of this confidence, the model organism (“source”) can represent these
other unstudied species (“targets”). This basic distinction between the model or source and
the phenomena or target it is supposed to represent is ubiquitous in reasoning with model
organisms (Ankeny and Leonelli 2011). Zebrafish is a model or representation of vertebrate
development, the target phenomena, because we expect to learn about vertebrate
development generally by studying ontogeny in zebrafish specifically. We do not invest time
and resources into zebrafish as a model organism only because we are interested in zebrafish.
Researchers plan to make claims about somite formation from observations in zebrafish that
will apply to somite formation in other vertebrates that we will never have the time or money
to investigate.
Developmental biologists often speak of investigating mechanisms that account for
phenomena in ontogeny (see Section 1.3), and focus on conserved genetic and cellular
mechanisms in model organisms (Gerhart and Kirschner 2007; Ankeny and Leonelli 2011;
Weber 2005). This suggests a distinction between representation with respect to
developmental phenomena and representation with respect to genetic and cellular
mechanisms operating in development. If we are interested in explaining how hearts
(phenomena) develop, then we might investigate the molecular or
cellular mechanisms occurring in the heart field during zebrafish ontogeny. Some of these
mechanisms could be conserved even though the phenomena are not. Drosophila has only
one cardiac cell type, no neural crest cells, and a heart with no atrial or ventricular chamber
morphology (Kirby 1999). However, cardiogenesis in all invertebrates and vertebrates
investigated thus far depends essentially on the expression of the homeobox gene Nkx2-
5/tinman (Gajewski and Schulz 2002). The reverse situation also can hold: similar
phenomena may be manifested but genetic and cellular mechanisms might differ.
Amphibians form a neural tube (neurulation) through a process of invagination (the folding
of an epithelial sheet), whereas teleost fishes form a neural tube via cavitation (the hollowing
out of a block of tissue via cell death). The neural tube is homologous across vertebrates (i.e.,
a conserved phenomenon), but the cellular and genetic mechanisms involved in invagination
versus cavitation are distinct (Davies 2013: ch. 4).
The distinction between phenomena and mechanisms assumes specificity; i.e., there are
specific phenomena (somite formation in vertebrates) or mechanisms (collinear Hox gene
expression) in view when judging the relationship between source (model) and target. But
animal development consists of a multitude of different processes that involve a host of
different mechanisms. Therefore, another distinction operating in the representational
criterion pertains to questions of specificity versus variety when selecting and using model
organisms. A model might represent one type of target phenomena (differentiation or
growth) or mechanism (cell signaling or cell cycling) but not others—specificity—or may do
so better or worse with respect to particular types of phenomena or mechanisms. A model
might represent several types of target phenomena and mechanisms simultaneously—variety
—with variability in how each type is represented. Trade-offs exist with respect to how well
different phenomena or mechanisms are co-instantiated in a model organism. Note that
experimental organisms may be selected with respect to variety and specificity
simultaneously, such as when a biologist working on a specific phenomenon intends to work
on others using the same model in the future. They also may be selected with one or the other
of these two aspects predominant. A model might be desirable if it has representational
variety in both mechanisms and phenomena even if it is not the best representative for every
specific mechanism or phenomenon. Conversely, a model organism might be desirable if it is
the best representative for a specific mechanism despite being a very poor model for other
phenomena or mechanisms. Variety is indicative of the “whole organism” being the model.
[17]
 A further distinction can be introduced between “model organisms” and “experimental
organisms” (Ankeny and Leonelli 2011) or “general model organisms” and “Krogh-principle
model organisms” (Love 2010). General model organisms are selected and used with the
variety aspect of the representational criterion preeminent; experimental or Krogh-principle
model organisms are selected and used with specificity preeminent.
Other issues relevant to the representation criterion include how individual cells or cell types
serve as developmental models (Fagan 2016), how developmental mechanisms in different
model organisms are compared and evaluated (Yoshida forthcoming), how the use of model
organisms constitutes an example of case-based reasoning (Ankeny 2012), and how model
organisms involve idealizations or known departures from features present in the model’s
target as the result of laboratory cultivation (Ankeny 2009; Section 5.2). Additionally, the
question of representation is not the only one germane to understanding model organism use.
Because model organisms are utilized for experimental intervention, questions of
representation must be juxtaposed with questions of manipulation (see the supplement
on Model Organisms and Manipulation).

5. Development and Evolution

The relationships that obtain between development and evolution are complicated and under
ongoing investigation (for a review, see Love 2015). Two main axes dominate within a loose
conglomeration of research programs (Raff 2000; Müller 2007): (a) the evolution of
development, or inquiry into the pattern and processes of how ontogeny varies and changes
over time; and, (b) the developmental basis of evolution, or inquiry into the causal impact of
ontogenetic processes on evolutionary trajectories—both in terms of constraint and
facilitation. Two examples where the concepts and practices of developmental and
evolutionary biology intersect are treated here: the problematic appeal to functional
homology in developmental genetics that is meant to underwrite evolutionary generalizations
about ontogeny (Section 5.1) and the tension between using normal stages for developmental
investigation and determining the evolutionary significance of phenotypic plasticity (Section
5.2). These cases expose some of the philosophical issues inherent in how development and
evolution can be related to one another.

5.1 Functional Homology in Developmental Genetics

The conserved role of Hox genes in axial patterning is referred to as functionally
homologous across animals (Manak and Scott 1994), over and above the relation of
structural homology that obtains between DNA sequences. And yet “functional homology” is
a contradiction in terms (Abouheif et al. 1997) because the definition of a homologue is “the
same organ in different animals under every variety of form and function” (Owen 1843: 379)
—the descendant, evolutionary distinction between homology (structure) and analogy
(function) is founded on this recognition. Therefore, the idea of functional homology appears
theoretically confused and there is a conceptual tension in its use by molecular
developmental biologists.

The reference to “organ” in Owen’s definition is indicative of a structure (an entity) found in
an organism that may vary in its shape and composition (form) or what it is for (function) in
the species where it occurs. Translated into an evolutionary context, sameness is cashed out
by reference to common ancestry. Since structures also can be similar by virtue of natural
selection operating in similar environments, homology is contrasted with analogy.
Homologous structures are the same by virtue of descent from a common ancestor,
regardless of what functions these structures are involved in, whereas analogous structures
are similar by virtue of selection processes favoring comparable functional outcomes,
regardless of common descent (Figure 6).
This is what makes similarity of function an especially problematic criterion of homology
(Abouheif et al. 1997). Because functional similarity is the appropriate relation for analogy,
it is not necessary for analogues to have the same function as a consequence of common
ancestry—similarity despite different origins suffices (Ghiselin 2005). Classic cases of
analogy involve taxa that do not share a recent common ancestor that exhibits the structure,
such as the external body morphology of dolphins and tuna (Pabst 2000). Thus, functional
homology seems to be a category error because what a structure does should not enter into an
evaluation of homologue correspondence and similarity of function is often the result of
adaptation via natural selection to common environmental demands, not common ancestry.
Although we might be inclined to simply prohibit the terminology of functional homology,
its widespread use in molecular and developmental biology should at least make us pause.
[18]
 While it is important to recognize this pervasive practice, some occurrences may be illicit.
Swapping structurally homologous genes between species to rescue mutant or null
phenotypes is not a genuine criterion of functional homology, especially when there is little
or no attention to establishing a phylogenetic context. This makes a number of claims of
functional homology suspect. To not run afoul of the conceptual tension, explicit attention
must be given to the meaning of “function.” Biological practice harbors at least four separate
meanings of function (Wouters 2003, 2005): activity (what something does), causal role
(contribution to a capacity), fitness advantage or viability (value of having something), and
selected effect or etiology (origination and maintenance via natural selection). Debate has
raged about which of them (if any) is most appropriate for different aspects of biological and
psychological reasoning or most general in scope (i.e., what makes them all function
concepts?) (see discussion in Garson 2016). Here the issue is whether we can identify a
legitimate concept of homology of function.
If we are to avoid mixing homology and analogy, then the appropriate notion of function
cannot be based on selection history, which is allied with the concept of analogy and
concerns a particular variety of function. Similarly, viability interpretations concentrate on
features where the variety of function is critical because of conferred survival advantages.
Any interpretation of function that relies on a particular variety of function (because it was
selected or because it confers viability) clashes with the demand that homology concern
something “under every variety of form and function.” A causal role interpretation
emphasizes a systemic capacity to which a function makes a contribution. It too focuses on a
particular variety of function, though in a way different from either selected effect or
viability interpretations. Only an activity interpretation (‘what something does’) accents the
function itself, apart from its specific contribution to a systemic capacity and position in a
larger context. Therefore, the most appropriate meaning to incorporate into homology of
function is “activity-function” because it is at least possible for activity-functions to remain
constant under every variety. An evaluation of sameness due to common ancestry is made
separately from the role the function plays (or its use), whether understood in terms of a
causal role, a fitness advantage, or a history of selection.[19] Activity-functions can be put to
different uses while being shared via common descent (i.e., homologous). More precisely,
homology of function can be defined as the same activity-function in different animals under
every variety of form and use-function (Love 2007). This unambiguously removes the
tension that plagued functional homology.
Careful discussions of regulatory gene function in development and evolution recognize
something akin to the distinction between activity- and use-function (i.e., between what a
gene does and what it is for in some process within the organism).
When studying the molecular evolution of regulatory genes, their biochemical and
developmental function must be considered separately. The biochemical function of PAX-
6 and eyeless are as general transcription factors (which bind and activate downstream
genes), but their developmental function is their specific involvement in eye morphogenesis
(Abouheif 1997: 407).
The biochemical function is the activity-function and the developmental function is the use-
function. This distinction helps to discriminate between divergent evolutionary trajectories.
Biochemical (activity-functions) of genes are often conserved (i.e., homologous), while
simultaneously being available for co-option to make causal role contributions (use-
functions) to distinct developmental processes. The same regulatory genes are evolutionarily
stable in terms of activity-function and evolutionarily labile in terms of use-function. [20] By
implication, claims about use-function homology for genes qua developmental function are
suspect compared to those concerning activity-function homology for genes qua biochemical
function because developmental functions are more likely to have changed as phylogenetic
distance increases.
The distinction between biochemical (activity) function and developmental (use) function is
reinforced by the hierarchical aspects of homology (Hall 1994). A capacity defining the use-
function of a regulatory gene at one level of organization, such as axial patterning, must be
considered as an activity-function itself at another level of organization, such as the
differentiation of serially repeated elements along a body axis. (Note that “level of
organization” need not be compositional and thus the language of “higher” and “lower”
levels may be inappropriate.) The developmental roles of Hox genes in axial patterning may
be conserved by virtue of their biochemical activity-function homologies but Hox genes are
not use-function homologues because of these developmental roles. Instead of focusing on
the activity of a gene component and its causal role in axial patterning, we shift to the
activity of axial patterning and its causal role elsewhere (or elsewhen) in embryonic
development.
Introducing a conceptually legitimate idea of homology of activity-function is not about
keeping the ideas of developmental biology tidy. It assists in the interpretation of evidence
and circumscribes the inferences drawn. For example, NK-2 genes are involved in mesoderm
specification, which underlies muscle morphogenesis. In Drosophila, the expression of a
particular NK-2 gene (tinman) is critical for both cardiac and visceral mesoderm
development. If tinman is knocked out and transgenically replaced with its vertebrate
orthologue, Nkx2-5, only visceral mesoderm specification is rescued; the regulation of
cardiac mesoderm is not (Ranganayakulu et al. 1998). A region of the vertebrate protein near
the 5′ end of the polypeptide differs enough to prevent appropriate regulation in cardiac
morphogenesis. The homeodomains (stretches of sequence that confer DNA binding) for
vertebrate Nkx2-5 and Drosophila tinman are interchangeable. The inability of Nkx2-5 to
rescue cardiac mesoderm specification is not related to the activity-function of differential
DNA binding. One component of the orthologous (homologous) proteins in both species
retains an activity-function homology related to visceral mesoderm specification but another
component (not the homeodomain) has diverged. This homeobox gene does not have a single
use-function (as expected), but it also does not have a single activity-function. Any adequate
evaluation of these cases must recognize a more fine-grained decomposition of genes into
working units to capture genuine activity-function conservation. We can link activity-
function homologues directly to structural motifs within a gene, but there is not necessarily a
single activity-function for an entire open reading frame.
Defusing the conceptual tensions between developmental and evolutionary biology with
respect to homology of function has a direct impact on the causal generalizations and
inferences made from model organisms (Section 4). Activity-function homology directs our
attention to the stability or conservation of activities. This conservation is indicative of when
the study of mechanisms in model organisms will produce robust and stable generalizations
(Section 1.3). The widespread use of functional homology in developmental biology is aimed
at exactly this kind of question, which explains its persistence in experimental biology
despite conceptual ambiguities. Generalizations concerning molecular signaling cascades are
underwritten by the coordinated biochemical activities in view, not the developmental roles
(though sometimes they may coincide). Thus, activity-function details about a signaling
cascade gleaned from a model organism can be generalized via homology to other unstudied
organisms even if the developmental role varies for the activity-function in other species.

5.2 Normal Stages and Phenotypic Plasticity

All reasoning strategies combine distinctive strengths alongside of latent weaknesses. For
example, decomposing a system into its constituents to understand the features manifested by
the system promotes a dissection of the causal interactions of the localized constituents,
while downplaying interactions with elements external to the system (Wimsatt 1980; Bechtel
and Richardson 1993). Sometimes the descriptive and explanatory practices of the sciences
are successful precisely because they intentionally ignore aspects of natural phenomena or
use a variety of approximation techniques. Idealization is one type of reasoning strategy that
scientists use to describe, model, and explain that purposefully departs from features known
to be present in nature. For example, the interior space of a cell is often depicted as relatively
empty even though intracellular space is known to be crowded (Ellis 2001); the variable of
cellular volume takes on a value that is known to be false (i.e., relatively empty).
Idealizations involve knowingly ignoring variations in properties or excluding particular
values for variables, in a variety of different ways, for descriptive and explanatory purposes
(Jones 2005; Weisberg 2007).
“Normal development” is conceptualized through strategies of abstraction that manage
variation inherent within and across developing organisms (Lowe 2015, 2016). The study of
ontogeny in model organisms (Section 4) is usually executed by establishing a set of normal
stages for embryonic development (see Other Internet Resources). A developmental
trajectory from fertilized zygote to fully-formed adult is broken down into distinct temporal
periods by reference to the occurrence of major events, such as fertilization, gastrulation, or
metamorphosis (Minelli 2003: ch. 4; see Section 1.2). This enables researchers in different
laboratory contexts to have standardized comparisons of experimental results (Hopwood
2005, 2007). They are critical to large communities of developmental biologists working on
well-established models, such as chick (Hamburger and Hamilton 1951) or zebrafish
(Kimmel et al. 1995): “Embryological research is now unimaginable without such standard
series” (Hopwood 2005: 239). These normal stages are a form of idealization because they
intentionally ignore kinds of variation in development, including variation associated with
environmental variables. While facilitating the study of particular causal relationships, this
means that specific kinds of variation in developmental features that might be relevant to
evolution are minimized in the process of rendering ontogeny experimentally tractable (Love
2010).
Phenotypic plasticity is a ubiquitous biological phenomenon. It involves the capacity of a
particular genotype to generate phenotypic variation, often in the guise of qualitatively
distinct phenotypes, in response to differential environmental cues (Pigliucci 2001; DeWitt
and Scheiner 2004; Kaplan 2008; Gilbert and Epel 2009). One familiar example is seasonal
caterpillar morphs that depend on different nutritional sources (Greene 1989). Some of the
relevant environmental variables include temperature, nutrition, pressure/gravity, light,
predators or stressful conditions, and population density (Gilbert and Epel 2009).
The reaction norm is a summary of the range of phenotypes, whether quantitatively or
qualitatively varying, exhibited by organisms of a given genotype for different
environmental conditions. When the reaction norm exhibits discontinuous variation or
bivalent phenotypes (rather than quantitative, continuous variation), it is often labeled
a polyphenism (Figure 7).
Phenotypic plasticity has been of recurring interest to biological researchers and
controversial in evolutionary theory. Extensive study of phenotypic plasticity has occurred in
the context of quantitative genetic methods and phenotypic selection analyses, where the
extent of plasticity in natural populations has been demonstrated and operational measures
delineated for its detection (Scheiner 1993; Pigliucci 2001). Other aspects of plasticity
require different investigative methods to ascertain the sources of plasticity during ontogeny,
the molecular genetic mechanisms that encourage plasticity, and the kinds of mapping
functions that exist between the genotype and phenotype (Pigliucci 2001; Kirschner and
Gerhart 2005: ch. 5). These latter aspects, the origin of phenotypic variation during and after
ontogeny, are in view at the intersection of development and evolution: How do molecular
genetic mechanisms produce (or reduce) plasticity? What genotype-phenotype mapping
functions are prevalent or rare? Does plasticity contribute to the origination of evolutionary
novelties (Moczek et al. 2011; West-Eberhard 2003)?
In order to evaluate these questions experimentally, researchers need to alter development
through the manipulation of environmental variables and observe how a novel phenotype can
be established within the existing plasticity of an organism (Kirschner and Gerhart 2005: ch.
5). This manipulation could allow for the identification of patterns of variation through the
reliable replication of particular experimental alterations within different environmental
regimes. However, without measuring variation across different environmental regimes, you
cannot observe phenotypic plasticity. These measurements are required to document the
degree of plasticity and its patterns for a particular trait, such as qualitatively distinct morphs.
An evaluation of the significance of phenotypic plasticity for evolution requires answers to
questions about where plasticity emerges, how molecular genetic mechanisms are involved
in the plasticity, and what genotype-phenotype relations obtain.
Developmental stages intentionally ignore variation associated with phenotypic plasticity.
Animals and plants are raised under stable environmental conditions so that stages can be
reproduced in different laboratory settings and variation is often viewed as noise that must be
reduced or eliminated if one is to understand how development works (Frankino and Raff
2004). This practice also encourages the selection of model organisms that exhibit less
plasticity (Bolker 1995). The laboratory domestication of a model organism may also reduce
the amount or type of observable phenotypic variation (Gu et al. 2005), though laboratory
domestication also can increase variation (e.g., via inbreeding). Despite attempts to reduce
variation by controlling environmental factors, some of it always remains (Lowe 2015) and
is displayed by the fact that absolute chronology is not a reliable measure of time in
ontogeny, and neither is the initiation or completion of its different parts (Mabee et al. 2000;
Sheil and Greenbaum 2005). Developmental stages allow this recalcitrant variation to be
effectively ignored by judgments of embryonic typicality. Normal stages also involve
assumptions about the causal connections between different processes across sequences of
stages (Minelli 2003: ch. 4). Once these stages have been constructed, it is possible to use
them as a visual standard against which to recognize and describe variation as a deviation
from the norm (DiTeresi 2010; Lowe 2016). But, more typically, variation ignored in the
construction of these stages is also ignored in the routine consultation of the stages in day-to-
day research contexts (Frankino and Raff 2004).
Normal stages fulfill a number of goals related to descriptive and explanatory endeavors that
developmental biologists engage in (Kimmel et al. 1995). They yield a way to measure
experimental replication, enable consistent and unambiguous communication among
researchers, especially if stages are founded on commonly observable morphological
features, facilitate accurate predictions of developmental phenomena, and aid in making
comparisons or generalizations across species. As idealizations of ontogeny, normal stages
allow for a classification of developmental events that is comprehensive with suitably sized
and relatively homogeneous stages, reasonably sharp boundaries between stages, and
stability under different investigative conditions (Dupré 2001), which encourages more
precise explanations within particular disciplinary approaches (Griesemer 1996).
Idealizations also can facilitate abstraction and generalization, both of which are a part of
extrapolating findings from the investigative context of a model organism to other domains
(Steel 2008; see Section 4 and 5.1).
There are various weaknesses associated with normal stages that accompany the fulfillment
of these investigative and explanatory goals. Key morphological indicators sometimes
overlap stages, terminology that is useful for one purpose may be misleading for another,
particular terms can be misleading in cross-species comparisons, and manipulation of the
embryo for continued observation can have a causal impact on ontogeny. Avoiding
variability in stage indicators can encourage overlooking the significance of this variation, or
at least provide a reason to favor its minimization.
Thus, there are good reasons for adopting normal stages to periodize model organism
ontogeny, and these reasons help to explain why their continued use yields empirical success.
However, similar to other standard (successful) practices in science, normal stages are often
taken for granted, which means their biasing effects are neglected (Wimsatt 1980), some of
which are relevant to evolutionary questions (e.g., systematically underestimating the extent
of variation in a population). This is critical to recognize because the success of a
periodization is not a function of the eventual ability to relax the idealizations; periodizations
are not slowly corrected so that they become less idealized. Instead, new periodizations are
constructed and used alongside the existing ones because different idealizations involve
different judgments of typicality that serve diverse descriptive and explanatory aims. In
addition to the systematic biases involved in developmental staging, most model organisms
are poorly suited to inform us about how environmental effects modulate or combine with
genetic or other factors in development—they make it difficult to discover details about
mechanisms underlying reaction norms. Short generation times and rapid development are
tightly correlated with insensitivity to environmental conditions through various mechanisms
such as prepatterning (Bolker 1995).
The tension between the specific practice of developmental staging in model organisms and
uncovering the relevance of variation due to phenotypic plasticity for evolution can be
reconstructed as an argument.

1. Variation due to phenotypic plasticity is a normal feature of ontogeny.

2. The developmental staging of model organisms intentionally downplays variation in
ontogeny associated with the effects of environmental variables (e.g., phenotypic
plasticity) by strictly limiting the range of values for environmental variables and by
removing variation in characters utilized to establish the comprehensive periodization.
3. Therefore, using model organisms with specified developmental stages will make it
difficult, if not impossible, to observe patterns of variation due to phenotypic
plasticity.
Although this tension obtains even if the focus is not on evolutionary questions, sometimes
encouraging developmental biologists to interpret absence of evidence as evidence of the
developmental insignificance of phenotypic plasticity, it is exacerbated for evolutionary
researchers. The documentation of patterns of variation is precisely what is required to gauge
the evolutionary significance of phenotypic plasticity. Practices of developmental staging in
model organisms can retard our ability to make either a positive or negative assessment.
Developmental staging, in conjunction with the properties of model organisms, tends to
encourage a negative assessment of the evolutionary importance of phenotypic plasticity
because the variation is not manifested and documented, and therefore is unlikely to be
reckoned as substantive. Idealizations involving normal stages discourage a robust
experimental probing of phenotypic plasticity, which is an obstacle to determining its
evolutionary significance.
The consequences of this tension for the intersection of development and evolution are two-
fold. First, the most powerful experimental systems for studying development are set up to
minimize variation that may be critical to comprehending how evolutionary processes occur
in nature. Second, if evolutionary investigations revolve around a character that was assessed
for typicality to underwrite the temporal partitions that we call stages, then much of the
variation in this character was conceptually removed as a part of rendering the model
organism experimentally tractable.[21]
The identification of drawbacks that accompany strategies of idealization used to study
development invites consideration of ways to address the liabilities identified (Love 2006).
We can construct a principled perspective on how to address these liabilities by adding three
further premises:

4. Reasoning strategies involving idealization, such as (2), are necessary to the

successful prosecution of biological investigations of ontogeny.
5. Therefore, compensatory tactics should be chosen in such a way as to specifically
redress the blind spots arising from the kind of idealizations utilized.
6. Given (1)–(3), compensatory tactics must be related to the effects of ignoring
variation due to phenotypic plasticity that result from the developmental staging of
model organisms.
At least two compensatory tactics can promote observations of variation due to phenotypic
plasticity that is ignored when developmental stages are constructed for model organisms:
the employment of diverse model organisms and the adoption of alternate periodizations.
Variation often will be observable in non-standard model organisms because experimental
organisms that do not have large communities built around them are less likely to have had
their embryonic development formally staged, and thus the effects of idealization on
phenotypic plasticity are not operative. In turn, researchers are sensitized to the ways in
which these kinds of variation are being muted in the study of standard models. Stages can
be used then as visual standards to identify variation as deviations from a norm and thereby
characterize patterns of variability.[22]
A second compensatory tactic is the adoption of alternative periodizations. This involves
choosing different characters to construct new temporal partitions, thereby facilitating the
observation of variation with respect to characteristics previously stabilized in the normal
stage periodization. These alternative periodizations often divide a subset of developmental
events according to processes or landmarks that differ from those used to construct the
normal stages, and they may not map one-one onto the existing normal stages, especially if
they encompass events beyond the trajectory from fertilization to a sexually mature adult.
This lack of isomorphism between periodizations also will be manifested if different
measures of time are utilized, whether sequence (event ordering) or duration (succession of
defined intervals), and whether sequences or durations are measured relative to one another
or against an external standard, such as absolute chronology (Reiss 2003; Colbert and Rowe
2008). These incompatibilities prevent assimilating the alternative periodizations into a
single, overarching staging scheme. In all of these cases, idealization is involved and
therefore each new periodization is subject to the liabilities of ignoring kinds of variation.
However, alternative periodizations require choosing different characters to stabilize and
typify when defining its temporal partitions, which means different kinds of variation will be
exposed than were previously observable.[23]
The compensatory tactics of employing a diversity of model organisms and adopting
alternative periodizations may be conceptually appropriate for addressing how the practice of
developmental staging has an impact on the detection of phenotypic plasticity, but this does
not remove associated costs (human, financial, and otherwise) or controversy. The
advantages of a single, comprehensive periodization for a general model organism (e.g.,
zebrafish normal stages) must be weighed in light of the advantages of alternative, process-
specific periodizations. However, by openly scrutinizing these practices in relation to the
phenomenon of interest and recognizing both advantages and drawbacks involved in the
idealizations utilized, developmental and evolutionary biologists are better positioned to
offer systematic descriptions and comprehensive explanations of biological phenomena.

6. Conclusion
This entry has only sampled a small portion of work relevant to the import and promise of
conceptual reflection on the epistemology of developmental biology. Much more could be
said about each of the above domains, such as a more fine-grained analysis of how normal
stages operate as types in developmental biology (DiTeresi 2010; Lowe 2016). Additionally,
little has been said about how evidence works in developmental biological experimentation
or differences between confirmatory and exploratory experimentation (Hall 2005; O’Malley
2007; Waters 2007b), nor have I treated the role of metaphors and models that characterize
key practices in developmental biology (Fagan 2013; Keller 2002). The latter have been
perspicuously analyzed via increased attention to the details of particular research programs.
Finally, nothing has been said about the metaphysical implications of developmental
phenomena (a key input for Aristotle’s metaphysics). Concepts of potentiality are very
natural in descriptions of embryological phenomena (e.g., the pluripotency of stem cells or
the potential of a germ layer to yield different kinds of tissue lineages) and some have argued
that empirical advances in developmental biology support a new form of essentialism about
biological natural kinds (Austin 2019). This bears on how we understand dispositions (see
the entry on dispositions) because the triggering conditions are often complex and multiply
realized (including manifestations without a trigger), as well as the fact that cells exhibit
dispositions with multiple possible manifestations (cell types) in specific sequential
orderings (Hüttemann and Kaiser 2018; Laplane 2016). Metaphysical issues also arise in the
context of human developmental biology, such as how to understand the ontology of
pregnancy (Kingma 2018; Sidzinska 2017). Thus, developmental biology displays not only a
rich array of material and conceptual practices that can be analyzed to better understand the
scientific reasoning exhibited in experimental life science, but also points in the direction of
new ideas for metaphysics, especially when that endeavor explicitly considers the input of
empirically successful sciences.