Among biological entities, cells are regarded as of special importancesince they are widely viewed as the simplest organized systems thatare unambiguously alive. Although one can debate about entities suchas viruses, there is little debate that cells are living. Cellsperform all the activities critical to life, from metabolism toreproduction. All cells alive today maintain themselves far fromthermodynamic equilibrium with their environment and are part of acontinuous lineage of cell division that goes back approximately 4billion years. The study of cells has required developing means ofmaterially manipulating them and our contemporary understanding ofcellular phenomena integrates results from a wide range of materialinterventions. In introducing a special issue ofStudies inHistory and Philosophy of Biological and Biomedical Sciences onthe cell, O’Malley and Müller-Wille assert
The cell, we suggest, is a nexus: a connection point betweendisciplines, methods, technologies, concepts, structures andprocesses. Its importance to life, and to the life sciences andbeyond, is because of this remarkable position as a nexus, and becauseof the cell’s apparently inexhaustible potential to be found insuch connective relationships. (2010: 169)
The examination of cell biology, in turn, is a potent nexus forproductive interactions between philosophers, historians, and socialscientists, each of whom raises questions about the study of cellsrelevant to the others.
During the period when philosophy of science focused primarily onlaws, cell biology was largely neglected. Although it is important tocell biologists to discover useful generalizations, these are seldomcharacterized as laws. Once philosophers of science turned theirattention from laws to mechanisms, from representational realism tomodels and their functionality, and from concepts and theories toexperiments and instrumentation, more philosophers of science haveturned their attention to the practices of cell biology. Thesepractices are epistemically challenging. Due to the fact that cellsare mostly microscopic, they raise traditional philosophical issuesconcerning the status of unobservables and the trustworthiness ofevidence. Since a major concern of cell biology is to relate functions(e.g., biochemical reactions) to cell structures, cell biologists neednot just to observe but also to physically manipulate cells and theirconstituent structures and to develop informative representations.Their activities thus provide rich examples for philosophers. Althougha great deal of cell biology has yet to be tapped for philosophicalanalysis, we provide examples in which philosophers (as well ashistorians) have investigated and analyzed material practices in cellbiology. We start, however, closer to theoretical interests, examininghow the discipline of cell biology developed and how metaphors haveplayed crucial roles in shaping how scientists have conceptualizedcells. Subsequently we turn to controversies over whether cells can beunderstood mechanistically or require a vitalist or holistperspective. After that, we turn to questions of epistemology,including strategies for representing cells visually and fordeveloping mechanistic accounts of cell functioning.
It is sometimes assumed that scientific disciplines aresharply-delineated and can be defined by their subject matter. Infact, however, they are dynamically changing, shaped in large part byscientists, their research tools, and academic and researchinstitutions. This is clearly true in the case of cell biology.Scientists began investigating cells in the seventeenth century afterHooke and van Leeuwenhoek reported what they saw with newly inventedmicroscopes. Hooke (1665) gave cells their name. But the cell did notbecome a focal unit for studying biological processes until Schleiden(1838) and Schwann (1839) advanced the cell theory according to whichcells are the basic living units. The termcytology waswidely used for the studies that ensued, which focused mainly ondescribing cells as seen under the light microscope. Although Wilsonused the phrasecellular biology in his introduction toCowdry’sGeneral Cytology (Cowdry 1924), and many ofits contributors aspired to integrate disciplines studying cells, thetermcell biology was only introduced in the yearsimmediately after World War II. Those adopting the termcellbiology advocated complementing new higher-resolution imagescreated with the electron microscope with results from biochemical,biophysical, and molecular approaches to cells. This transition isexemplified in the naming of what would become the flagship journal inthe field. The working title used in discussions about founding a newjournal wasJournal of Cytology. By the time of the firstissue in 1955, it was called theJournal of Biophysical andBiochemical Cytology, a title chosen to emphasize the integrationof multiple approaches to the study of cells. In 1962 it wasre-christenedThe Journal of Cell Biology, by which time theInternational Society for Cell Biology and the American Society forCell Biology had also been established and the termcellbiology had acquired general currency for an integrated approachto cells. (Bechtel 2006: chapter 7, examines the establishment andrenaming of the journal as well as the creation of the AmericanSociety for Cell Biology.)
The introduction of new research techniques, especially cellfractionation and electron microscopy, figured centrally in theefforts to establish cell biology as a distinct discipline in thepost-World War II period as they enabled constructing what Palade(1987) termed a bridge between morphological observations andbiochemistry. Cell fractionation employs the ultracentrifuge toseparate cell contents into fractions that contain the enzymesresponsible for distinct cell activities while electron microscopyenables these fractions to be linked to organelles identifiable incells. Matlin (2018) describes how Claude employed centrifugation toidentify microsomes which exhibited high concentrations of RNA.Porter’s (1953) complementary micrographs of these particles infractions and tissue slices (Figure 1) linked microsomes to theintracellular structure he had named theendoplasmicreticulum. As Matlin describes, this cycling between cellfractionation to establish function and electron microscopy tolocalize it to a structure was pursued repeatedly in cell biology.


Figure 1: Micrographs of a fractioncontaining microsome and of a thin-sliced preparation from a cellshowing the endoplasmic reticulum (marked aser). From Porter(1953), plate 48.
In terms of content, however, the break between cytology and cellbiology was nowhere near as sharp as the founders of cell biologyportrayed. Cowdry’sGeneral Cytology (1924) clearly hadsimilar aspirations of linking studies of morphology and studies ofcellular chemistry (see discussions by historians, philosophers, andbiologists of Cowdry’s project in Matlin, Maienschein, &Laubichler 2018). But in terms of the institutions in which scientistsconducted their research, the break was quite significant. With thecreation of professional societies using the name “cellbiology”, biologists began to self-identify as cell biologists.Laboratories and academic departments adopted the namecellbiology and funding agencies recognized cell biology as asupported field of research.
Although cell biologists actively courted biochemists and to somedegree biophysicists to join their new discipline, these effortslargely failed. The most enduring connections were with molecularbiology, which was established as a new discipline in the same period(see the entry onmolecular biology). While molecular biology began with a focus on bacterial phage(viruses that infect bacteria) and cell biology with eukaryotic cellsfrom mammals and plants, by the 1970s each discipline had broadenedits focus and soon many cell biologists were adopting a molecularapproach to cell structure and function. (A continuing point ofdifference between cell biology and molecular biology is that cellbiology placed high value on linking chemical and molecular processesto cell structure; see Matlin 2016). Academic departments began toinclude bothcell andmolecular in their names. InNovember 1989 the American Society for Cell Biology created a newjournal,Cell Regulation, but with its third volume inJanuary 1992, the name was changed toMolecular Biology of theCell. Although there are textbooks that reference justcellbiology in their title, several of the most prominent textbooksfeature bothcell andmolecular. These includeAlberts’Molecular Biology of the Cell (Alberts,Johnson, Lewis et al. 2015) , Cooper and Hausman’sThe Cell:A Molecular Approach (2007), and Iwasa and Marshall’sKarp’s Cell and Molecular Biology (2016). The firstedition of Alberts’Molecular Biology of the Cell(Alberts, Bray, Lewis et al. 1983) was modeled in part onWatson’sMolecular Biology of the Gene (1976) andincluded Watson as a co-editor. Alberts attributes the vision ofintegrating cell and molecular biology to Watson (see interview withAlberts in the Oral History Collection at Cold Springs HarborLaboratory linked inOther Internet Resources below)
Not only have the disciplines studying cells undergone historicaltransformations, so has the understanding of what cells are. As Hesse(1966) argued, in many fields the objects of study are characterizedthrough metaphors to more familiar objects. In examining the differentmetaphors that have been invoked for cells, Reynolds argues that thechoice of metaphor has consequences for how cells are understood:
Metaphorical language … has been essential not only to theactivity ofdescribing cells but also toseeing andunderstanding them, and has played no less a fundamental rolethan the literal and material lenses of the microscope. (2018: 4)
Reynolds distinguishes two fundamental classes of metaphors:cellsas human artifacts andcells as organisms. Each hastaken on a variety of more specific forms. Among artifacts, cells havebeen characterized as spaces enclosed by solid walls, building blocks,factories, various types of machines, and electronic computers. Theorganismal metaphors have suggested a conceptualization of cells aselementary organisms (like unicellular amoebae) or citizens in a stateor society in which there is a division of labor and in which cellsmake decisions that determine their own developmental‘fates’, including the ultimate decision to initiateprogrammed cell death (Reynolds 2014).[1] These metaphors have given rise to distinctive research agendas thatfocus on specific aspects of cells and their relationship to othercells and their environment.
When Hooke introduced the termcell inMicrographiain 1665 to describe the microscopic structure of cork that he observedusing a compound microscope (Figure 2A), he was comparing cells to the polygonal cells of beeswax and thesmall rooms occupied by monks in a monastery. (Hooke 1665: 116, alsocommented on the fluid content of cells, labeling them“succus nutritus, or appropriate juices ofvegetables”. As discussed below, the fluid content of cells tookon new importance with the advent of the protoplasm theories in themid-nineteenth century.) Although researchers who focused on plantstended to find Hooke’s metaphor appropriate, since plant cellshave clearly observable walls, others, beginning with van Leeuwenhoek,who examined spermatozoa (Figure 2B), animal tissues, and bacteria, tended to adopt other terms such ascorpuscle orglobule.

A. Hooke’s (1665: Schem: XI, fig.1 facing page 115) drawing of his observations of cells in cork.

B.Van Leeuwenhoek’s drawing ofspermatozoa from his Letter to Nehemiah Grew, 18 March 1678.
Figure 2: Hooke’s drawing clearlyshows cells as areas enclosed by walls whereas van Leeuwenhoek’spresents what we callcells as simple organisms.
Most of these early investigators focused on describing what they sawin the microscope. But some, such as Buffon (1749), began to viewcells as basic living units. Nicholson characterizes this as thebeginning of an atomist tradition in biology that identified
a basic indivisible unit of life and [sought] to explain themorphological constitution and physiological operation of all livingbeings in terms of these fundamental units. (2010: 203)
This more theoretically committed conception of cells is exemplifiedin the publications of Schleiden (1838) and Schwann (1839), who aregenerally credited with establishing the cell theory, the fundamentaltenet of which is that cells are the basic units of living things.
Focusing exclusively on plants, Schleiden asserted that“… every plant … is an aggregate of fullyindividualized, independent, separate beings, the cellsthemselves” (1838 [1847: 231–2]). In addition to theirwalls, Schleiden appealed to the nucleus, which had been identifiedwith improved microscopes by Brown (1833), to identify plantcells. The nucleus figured centrally in Schleiden’s account ofcell formation according to which a new cell formed through a processlike crystal formation—one type of material is deposited aroundthe nucleolus to form the nucleus and then other material wasdeposited to form the cell body. The metaphor comparing cell formationto crystal formation played a yet larger role when Schwann confrontedthe challenge that animal cells differ vastly in theirappearance. Schwann argued that they were nonetheless all cells sincethey all formed through a process analogous to crystal formation:
The elementary parts of all tissues are formed of cells in ananalogous, though very diversified manner, so that it may be assertedthat there is one universal principle of development for theelementary parts of organisms, however different, and that thisprinciple is the formation of cells. (Schwann 1839 [1847: 165])
(For discussion of this and other uses of the crystal metaphor inbiology, see Haraway 1976.)
Given that Schwann and Schleiden were masters of microscopy (aswitnessed by their drawings based on their microscopic observations),and that other equally competent microscopists reported cellsdividing, their claim that cells form like crystals seems anomalous.Bechtel (1984) argues that Schwann’s strong commitment todeveloping mechanistic accounts of vital processes may have played amajor role in how he interpreted what he saw—the crystalformation metaphor provided a mechanistic model whereas at the timethere were no mechanical models for cell division. In any case,Schleiden’s and Schwann’s accounts of cell formation weresoon set aside as more researchers reported on cell division andSchwann’s concerns with mechanism were supplanted by otherconcerns. Virchow (1855), a pathologist, offered a theoretic argumentthat only a process like cell division could explain the transmissionof disease and coined the oft-cited dictum “omnis cellula ecellula” (all cells come from cells).
Schwann’s “theory of the cell”, however, was muchricher than his soon rejected view of cell formation. He argued thatcells were the basic units in which the processes of life (he coinedthe termmetabolism) occurred. He viewed metabolic processesas catalyzed by the specific materials that were deposited in cells asthey formed. Schwann (1836) had himself recently discovered pepsin, acatalyst that breaks down egg albumin. (Pepsin operates, however, notwithin cells but after being secreted into digestive fluid.) When ayear later he argued that fermentation could not occur without theinvolvement of a whole living yeast cell (Schwann 1837), some criticstook Schwann to be claiming that living systems exhibited mysteriousvital properties. But for Schwann this was consistent with hismechanist approach—the metabolic processes only occurred whenthe responsible catalysts were brought together in cells.
In the mid- to late-nineteenth century theorizing about cells went ina number of directions. One direction focused even more than Schwannon the material stuff constituting the cell. In the same period asSchwann was advancing his theory of the cell, von Mohl (1835), a plantresearcher, introduced the nameprotoplasm for the fluidmaterial in cells. Animal researchers such as Dujardin (1835) calledthe fluid seen in animal cellssarcode. Remak (1852) proposedthe protoplasm theory according to which the basic material of plantsand animal was identical and T. H. Huxley (1869) took this a stepfurther, arguing that protoplasm was the “physical basis oflife”. As discussed by Reynolds, protoplasm theory offered adifferent fundamental biological theory—whereas Schleiden andSchwann
sought to unify all the various forms of life through a commonmorphological type and developmental principle, the protoplasm theoryattempted to achieve this through the identification of a commonsubstance or material. (2018: 32)
Accordingly, some supporters of protoplasm sought to dislodge the cellas a basic living unit. The botanist Julius Sachs (1892), for example,asserted “to call the protoplasm unit a cell was about asappropriate as calling a live bee in a honeycomb a cell” (astranslated by Welch & Clegg 2010: C1281). Since protoplasm appearsas a viscous substance that is only artificially constrained by cellboundaries, some protoplasm theorists viewed it as supplanting thecell as an organizing unit. Where Schwann understood the cell to bethe fundamental unit of biological development, protoplasm theoristsunderstood protoplasm to drive ontogeny and cells to be merelysecondary structures deposited in the course of this primarysubstance’s development. Recently a few biologists (Welch &Clegg 2010, 2012) and philosophers (Nicholson, 2010) have argued forreviving protoplasm theory. Their argument is that protoplasm theorypresents a more holistic systems or organismal perspective. (Wediscuss holism and organicism inSection 3.)
The characterization of protoplasm as just a viscous substanceexisting on its own was complicated by the identification of the cellmembrane as a structure distinct from the cell wall and, as discussedbelow, of membrane-bound organelles. In the end, most biologistsacquiesced in some version of Schultze’s (1861: 11; astranslated by Hall, 1951: 451) compromise position that raisedprotoplasm to be one defining feature of cells when he defined a cellas “a lump of protoplasm inside of which lies anucleus”. Those who adopted such a compromise position ofteninvoked a metaphor, originating with Raspail, according to which thecell is “a kind of laboratory within which all tissues organizeand grow” (1843: 28; as translated by Harris, 1999: 32). Thelaboratory metaphor encouraged attempts to develop chemicalexplanations for all the reactions occurring in cells. There were anumber of variants on this theme. Unger analogized the plant cell witha “mächtige chemische Werkstätte” (powerfulchemical workshop) (1851: 23) and Virchow stated that “starch istransformed into sugar in the plant and animal just as it is in afactory” (1858: 107).
Embracing this chemical perspective, several chemists sought toidentify the chemical catalysts needed for the reactions to occur.Fermentation was a common focus and Kühne (1877) coined the termenzyme (Greek for “in yeast”) for the putativecatalyst operative in yeast. Enthusiasm for this effort wastemporarily dampened by Pasteur (1860). A noted chemist, henonetheless maintained that fermentation could only be carried out inwhole living cells. This reflects a conception of the cell as anexplanatorily irreducible unit of life—a conception associatedwith vitalism (discussed in more detail below). Enthusiasm for thefactory conception, though, was rekindled by Buchner (1897). Guided byPasteur’s contention that fermentation could not occur in theabsence of cells, he added sugar to an extract he prepared bydestroying all whole cells, thinking it would serve as a preservative.When he observed the emission of bubbles, indicating fermentation inthe absence of living cells, he changed course and galvanized thepursuit of chemical investigations of metabolism that resulted in theestablishment of modern biochemistry in the first decades of thetwentieth century (Kohler 1971; Cornish-Bowden 1997). One of its earlysuccesses was the characterization in the 1930s of a pathway of enzymecatalyzed reactions responsible for fermentation (Bechtel 1986).
At first the factory metaphor simply identified the cell as the placein which chemical reactions occur, but over time researchers began tofocus on the different machines within the factory. The focus onseparate machines in the factory was promoted in the late nineteenthand early twentieth century when, using improved microscopes andapplying stains to enhance contrast, investigators began to identifystructures within cells and theorize about their functionalsignificance. Researchers were particularly successful incharacterizing the nuclear events in cell division and fertilization.Flemming (1879, 1882) described in detail how the threads that hecalledchromatin (due to their absorption of dye), laternamedchromosomes, divided longitudinally, with the twohalves moving apart so that one of each would end up in each daughtercell (Figure 3). Soon after researchers such as Weismann and Correns pointed to linksbetween chromosome transmission and heredity, but it was Boveri (1902)and Sutton (1903) who provided the compelling evidence thatMendel’s factors (what would soon be calledgenes) arein or on chromosomes. Darden and Maull (1977) analyze the linking ofgenes to chromosomes as a major example of what they termedinterfield theories, theories that do not try to reduce oneaccount to another but integrate the findings of different fields in atheory that bridges them.

Figure 3: Flemming’s (1882)drawings of the stages of mitosis that highlights the formation ofspindles and their role in segregating chromosomes. Images 1–3 are from Tafel IIIa; 4–7 from Tafel IIIb. (Figure fromwiki commons.)
Late in the nineteenth century cytologists also succeeded inidentifying membrane-enclosed structures in the cytoplasm that came tobe known asorganelles. Altmann (1890), using new stains hedeveloped, observed filaments within cells that he took to beelementary organisms (a view he explicitly set in opposition to theprotoplasm theory). Although many researchers challengedAltmann’s observations, Benda (1899) confirmed the existence offilaments using a different stain and gave them the namemitochondria (Greek for “thread” and“granule”). Because of their reactivity with oxidativestains, Michaelis (1899) proposed that they figured in oxidativereactions in cells. Yet other researchers identified other organellessuch as the Golgi apparatus (Golgi 1898) and ergastoplasm (Garnier1897), which was ultimately identified as the endoplasmicreticulum.
In the early decades in the twentieth century biochemists andcytologists developed their own research techniques and pursued theirinvestigations independently of the other. Most biochemists implicitlyembraced the assumption that the cell was a bag of chemicals thatcould be studied in the extracts remaining after cell structure wasdestroyed whereas those pursuing cytological inquiries tended toembrace the factory metaphor in which organelles were distinctmachines. As witnessed by Cowdry’s (1924)GeneralCytology and Bourne’s (1942)Cytology and CellPhysiology, there were researchers who desired to build bridgesbetween biochemistry and cytology. There were some techniques, albeitlimited, for determining the chemical composition of organelles(essentially, those of cytochemistry and histochemistry, which reliedon determining to which chemicals various stains bound). The nearlysimultaneous introduction of cell fractionation and electronmicroscopy in the late 1940s provided the needed research techniquesand, as noted inSection 1, helped establish modern cell biology (Bechtel 2006; Matlin 2018). Inaddition to the linkage of microsomes to the endoplasmic reticulum,described above, researchers linked oxidative metabolism to the innermembrane of the mitochondrion. New structures were also discovered andconnected to functions, such as the lysosome which was linked to thebreakdown and recycling of disrupted cell components (de Duve1958).
The process of advancing new conceptions of cells continues. We notejust one example here. The pioneers of cell biology in the 1940s and1950s embraced the machines in a factory metaphor, treating cellorganelles as compartments in which different chemical reactions werecatalyzed by the enzymes housed there. Once formed, the compartments,on this view, did not change—the crucial activities occurredwithin them. Over time, however, some researchers within cell biologybegan to identify cell structures that executed mechanical movements,as anticipated by Flemming’s (1882) characterization of the cellspindle as pulling chromosomes apart in mitosis. Of particularimportance was research on the cytoskeleton (consisting of actinfibres and microtubules). The termcytoskeleton was invokedin the 1930s and 1940s to denote a rigid structure, and indeed it doeshelp give cells their shape. But it was soon found to be the locus ofmovement. H. E. Huxley (1969) advanced an account of musclecontraction as resulting from myosin molecules forming bridges thatpull on actin filaments. Both actin filaments and microtubules werefound to continually extend themselves at one end by incorporating newproteins while removing them from the other, a process characterizedastreadmilling (Cleveland 1982). Video observations oforganelles moving along microtubules (R. Allen et al. 1982) led to thediscovery of kinesins (Vale, Reese, & Sheetz 1985), molecularmotors thatwalk along microtubules carrying cargo. It alsofocused new attention on dyneins (Paschal & Vallee 1987),previously only known for their roles in cilia and flagella, ascarrying cargo in the opposite direction. Recognition of thiscontinual movement of material within cells gave rise to a newmetaphor of the cell as a city with bustling traffic (Vale &Milligan 2000).
Almost all of those pursuing the various metaphors discussed in theprevious section embraced a view that had its origins in Descartes,who maintained both non-living and living systems (with the exceptionof the human mind) operated like machines. But a significant number ofresearchers rejected this perspective and maintained that livingorganisms, including cells, are fundamentally different from ordinaryphysical, mechanical systems. They argued that in one way or anothercomposition from material components is insufficient to account forthe phenomena associated with cells. We begin with these opponents ofmechanistic conceptions of cells and then examine how those advocatingmechanistic approaches responded.
In the eighteenth and nineteenth century the opponents of mechanismwere typically referred to asvitalists. The variousvitalists all rejected mechanism, taken as the view that organisms arecomposed of physical parts that operate in accord with the sameprinciples as processes in the non-living world. Their positive viewsvaried. Some vitalists adopted a position much like that of substancedualists with respect to the mind, arguing that some non-materialcomponent—a vital force (vis vitalis)—operates inliving beings and accounts for their distinctive activities. Othersavoided positing an extra component but maintained that different lawsapply in living organisms than in non-living systems (for discussionsof these different versions of vitalism, see contributions inNormandin & Wolfe 2013). Regardless of how they expressed theirpositive views, vitalists commonly pointed to activities of livingorganisms that they claimed could not be accounted for in the samemanner as physical processes. This is well illustrated in Bichat(1805), who focused first on the apparent lack of determinism in thebehavior of biological organisms and second on the fact that organismsseemed to oppose physical processes that threatened to destroy them(in his words, theyresist death). As noted above, Schwannand Pasteur both claimed that living yeast are needed to producefermentation. While for Schwann this only entailed that fermentationdepended on the specific combination of materials found in cells, forvitalists it entailed that living cells perform activities that couldnot be performed by the collection of component molecules. Bichat andPasteur, as well as other prominent vitalists such as Müller(1837–1840), embraced empirical and experimental research, butdrew limits with respect to what could be explained by appeal to thematerial components of an organism alone.
In many cases, mechanists did not explicitly respond to vitalists butsimply moved forward with their research. Bernard (1865), was anexception. He addressed Bichat’s challenges by introducing adistinction between an organism’s internal and externalenvironment. Mechanistic operations within organisms are carried outin the internal environment and jointly serve to maintain thatenvironment in a constant state. Because mechanistic operationsrespond to conditions in the internal environment, they appearindeterminate when considered only in relation to external stimuli.Moreover, because these mechanisms work to maintain a constantinternal environment, one could explain the ability of organisms toresist physical processes that might otherwise destroy them (fordiscussions of Bernard see Holmes 1974 and LaFollette & Shanks1994). Bernard’s approach was the foundation for Cannon’slater well-known work on homeostasis (1929).
By the beginning of the twentieth century few biologists investigatingcells (Driesch, 1914,being a notable exception) espoused vital forces and vitalismperse ceased to be regarded as a tenable position. But manybiologists were still concerned to account for differences betweenliving cells and ordinary material systems. A prominent positionadopted by many investigators (Haldane 1929, 1931; Lillie 1934;Needham 1936; Russell 1945, 1930; Von Bertalanffy 1952; Weiss 1963;Woodger 1929) washolism, sometimes referred to asorganicism. (See Nicholson and Gawne 2015, for a discussionof how the organicists differentiated their position from bothvitalism and mechanism.) Holists accepted that living systems werebuilt out of material parts. They insisted, however, that theactivities performed by components of living systems depended not juston those components and their composition but on their organization.Accordingly, one cannot just add together the activities of thecomponents to account for the whole (“the whole is not just thesum of its parts”). The organized whole in part determines howthe parts behave. This attitude is manifest, for example, in J. S.Haldane’s opposition to investigating the origins of life fromnon-living matter: “There is and can be no origin of life out ofmechanical conditions. Such an origin is inconceivable” (Haldane1930: 12).
The organization of cellular systems plays a central role in thedialectic between mechanistic biologists and holists. Holists oftenconstrue mechanistic biologists as downplaying the importance oforganization, but many mechanistic biologists deny this. Bernard hademphasized the role of organization in allowing components to maintainthe constancy of the internal environment and those seeking to explainhomeostasis appealed to feedback loops. Recognizing that mechanisticaccounts are not limited to a simple, additive view of organization,J. S. Haldane’s son, J. B. S. Haldane, abandoned hisfather’s commitment to holism and embraced a mechanisticframework that emphasized how biological components are affected bybeing incorporated within organized systems. In making this break withhis father he was heavily influenced by working with the biochemistHopkins (1913), who likewise made organization central to his accountsof biochemical processes. Accordingly, J. B. S. Haldane became one ofthe pioneers in formulating biological inquiry into the origins oflife while still insisting that complex organization figured centrallyin activities of living organisms (Martin 2010).
Philosophers and scientists often invoke the wordemergentfor phenomena that are different from the phenomena generated by theircomponents (see the entry onemergent properties). Sometimes emergent phenomena are viewed as incapable of beingexplained in terms of their constituents. In the context ofcharacterizing systems biological accounts of cell phenomena, Boogerdet al. (2005) develop an account according to which emergent phenomenaare ones that are fully explicable in terms of how their constituentsbehave in the organized system, but not in terms of how they behave insimpler (less complexly organized) systems. This recognizes that manycellular constituents behave differently when incorporated intoparticular systems in which they receive distinctive inputs. For otherrecent treatments of emergence and its applications to cell biology,see (Hooker 2011a; S.Mitchell 2012; Mossio, Bich, & Moreno 2013; Winning & Bechtel2019).
The desire to understand how the organization in cells andmulticellular organisms differs from that found in most naturallyoccurring systems or human-made artifacts motivated a body of researchintheoretical biology in the 1970s and 1980s. Among the mostprominent contributors to theoretical biology were Pattee (many of hismost important papers have been collected in Pattee 2012), Rosen(1985, 1991), Polanyi (1968), and Waddington (1961). Philosophy ofbiology, as it developed as a specialty in philosophy of science inthe 1970s and 1980s, largely ignored this tradition. Today however anumber of philosophers concerned with cell biology are drawing uponits insights. Here we focus on one key conceptual tool theoreticalbiologists provide for understanding the distinctive activities ofliving systems such as cells: constraints. The notion of constraint isdrawn from classical mechanics, where constraints serve to constitutemacro-scale objects from their micro-scale particles. Constraints thusaccount for why macro-scale objects exhibit different properties thantheir constituents. Constraints are not explained by laws but ratherserve as boundary conditions that must be ascertained empirically.Accordingly, to the degree constraints explain biological activitiesof cells, these activities cannot be reduced, in the sense of beingderived from the principles of chemistry or physics. Instead,researchers must, on the basis of empirical inquiry, identify theconstraints actually realized in living cells.
The importance of constraints for understanding cellular and otherbiological phenomena has been developed recently by Hooker (2011b, 2013) and Moreno and Mossio(2015). Hooker makes clear that, although the termconstraintsuggests limitations, constraints also extend possibilities—tooffer a cellular example, microtubules that run from the cell centerto the periphery restrict the movement of the kinesin and dyneinmotors that move on them but also provide a possibility for transportof organelles to distant locations. Moreno and Mossio in particulardevelop a perspective that links the focus on constraints to the ideathat living cells are autopoietic. Drawing upon Maturana andVarela’s (1980) conception of living systems as autopoieticmachines—machines that provide a network of production thatenables the construction of the living system—Moreno and Mossioadd a focus on the thermodynamic requirements of cells. Cells, ashighly organized systems, are far from equilibrium and require acontinual source of free energy to carry out the operations requiredto synthesize new components and resist the tendency towardsequilibrium. What constraints do in organisms, on their account, isdirect flows of free energy to perform the work of building,repairing, and reproducing the organism.
Some constraints are flexible, and these make possible an importantfeature of living systems—the ability of organisms to controlproduction mechanisms (such as those involve in fermentation or musclecontraction) through the actions of control mechanisms. On Pattee’s account, control mechanismschange the flexible constraints inproduction mechanisms inlight of information that is procured by making measurements. Negativefeedback control mechanisms, such as thermostats, are simple examples:the thermostat makes a measurement of a variable (temperature in theroom) that is affected by the operation of the productionmechanisms—the furnace—and based on the measurementexecutes action on the constraints in the furnace mechanism. Examplesof feedback control mechanisms are widespread in cells. Some of thebest-known examples are thelac operon (Jacob & Monod1961) and feedback control of glycolysis by ATP (Ghosh & Chance1964). The measurements used by control systems need not be restrictedto states affected by the activity of the production mechanism; theycan also measure states in the organism or states in its environment.Using such measurements, production mechanisms can be controlled so asto operate only in particular circumstances, enabling the organism,for example, to navigate to a food source or avoid a predator asneeded.
Moreno and Mossio (2015) offer an account in which appropriatelyorganized productive and control mechanisms enable cells, and byextension multi-cellular organisms, to achieve what they refer to asbiological autonomy:
a distinctive regime of causation, able not only of producing andmaintaining the parts that contribute to the functioning of the systemas an integrated, operational, and topologically distinct whole butalso able to promote the conditions of its own existence through itsinteraction with the environment. (2015: xvi–xvii)
Note that mechanisms, on this view, are contained within cells andthat it is cells, not mechanisms that are autonomous. What becomescrucial for understanding the autonomy of cells is the organization ofcontrol mechanisms that orchestrate the activities of variousproductive mechanisms so as to maintain the cell (or the multicellularorganism). One notable feature of this focus on the organizationneeded to maintain autonomy is that it is compatible with the moretraditional philosophical accounts of mechanism (discussed insection 5) but emphasizes a feature not prominent in them—that productionmechanisms are subject to control mechanisms. To understand thebehavior of cellular mechanisms, researchers must not only look insidemechanisms to their organized parts and operations but outside to howthey are situated in cells and organisms in such a way that they canbe controlled by other mechanisms (Winning & Bechtel 2018; Bechtelin press). Such an ontological framework for understanding cells (aswell as multicellular organisms) integrates insights from traditionalmechanists in biology and their vitalist/organicist/holist critics,capturing what is distinctive of living organisms within a frameworkthat accepts them as consisting of mechanisms, but mechanismsorganized in appropriate ways.
We turn now from questions about the ontological status of cells toepistemic questions about how scientists study them. We begin with howscientists represent cells and information about them. Traditionally,philosophy of science has focused on linguistic representations ofscientific knowledge. But in many fields of biology and especially incell biology, information is often presented in images.Scientists’ very familiarity with cells results from visualrepresentations generated using microscopes. Developing microscopesand techniques for using them to produce interpretable imagespresented a number of challenges. We begin in section 4.1 with thechallenges in generating images (micrographs) at all and in section4.2 consider challenges in evaluating the reliability of the resultingimages. Beyond these replete (highly detailed) representations, cellbiologists rely on a variety of less replete diagrams. (Thecharacterization of images as “replete” is due to Perini,2013. Drawing on Goodman, she uses relative repleteness todifferentiate diagrams from pictures.) In particular, as discussed insection 4.3, when they are developing mechanistic hypotheses about cells, cellbiologists rely on cell diagrams that represent types of cellcomponents and mechanism diagrams that represent select componentswithin cells that are hypothesized to constitute parts of themechanism responsible for a given phenomenon (Downes 1992).
As discussed above, in the seventeenth century both Hooke and vanLeeuwenhoek pioneered the use of light microscopes to observe cells.Subsequent investigators often designed their own microscopes. Thevariations in these designs contributed to variability in theresulting images. (Before photography, microscopists drew what theysaw using the microscope, introducing another source of variation.)The variability of the images different researchers produced was onefactor that led biologists in the eighteenth and nineteenth centuriesto examine more carefully the processes through which microscopesgenerate images (Schickore 2001, 2007). In his theoretical studies oflenses, Newton (1704) characterized two types of aberrations createdby lenses: spherical aberrations, resulting from light rays cominginto focus at different points, and chromatic aberrations, resultingfrom light of different wavelengths being refracted at differentangles. Schickore (2007) describes the many efforts by microscopedevelopers to correct for these aberrations and the creation of testobjects for evaluating the reliability of particular microscopes.During the same time enthusiasts were advancing many claims about whatthey saw, some of which were shown later to be artifacts. For example,both Milne-Edwards (1823) and Dutrochet (1824) reported roundstructures of a constant size similar to that reported for cells,which they termedglobules. (See Schickore 2009 for adetailed examination of the reports of globules and an argument thatit was the variability in these reports that contributed to thegrowing sense that something was amiss in the practices of themicroscopists.) Globules were, however, soon shown to be the productsof spherical aberrations. In the early nineteenth century several lensmakers developed strategies for eliminating spherical aberrations andgreatly limiting chromatic aberration (chromatic aberrations were notfully eliminated until the introduction of the apochromatic lenses inmid-century). As a result, the observations by Schleiden, Schwann, andothers discussed above were largely free of these distortions.
Light microscopes faced another limit, that on magnification (alimitation imposed by the wave length of light). For cell researchersto obtain higher-resolution images that could reveal constituents ofcells, a microscope relying on different physical principles wasrequired. The most important alternative to the light microscope forstudying cells was the (transmission) electron microscope, whichemploys beams of electrons to create images in a manner comparable tophotography: locations on the photographic plate hit by many electronsare black in the negative and white in the positive image. Whenstructures in the cell scatter electrons, the location in the negativeremains white and appears dark in the positive image. Althoughelectron microscopes were available in the early 1930s, only in theyears just before and during World War II did biologists begin toexplore their potential. One difficulty they confronted is that mosteukaryotic cells are too thick to be penetrated by the 50kV electronbeam available in the first electron microscopes. Microtomes had beendeveloped for cutting slices of cells for light microscopy, but newapproaches to microtome design were required to slice cellssufficiently thinly without creating distortions. This problem was notsolved until the early 1950s (Porter & Blum 1953). Accordingly,some of the first electron micrograph studies focused on fibrousmaterial such as collagen (Schmitt, Hall, & Jakus 1942) or onbacteria (Stanley & Anderson 1941). Porter, Claude, and Fullam(1945) created the first electron micrograph of a eukaryotic cell byculturing it under conditions where the periphery spread very thinly,allowing electrons to penetrate. They generated an image (Figure 4) that showed at the periphery
filamentous mitochondria of various lengths and fairly constant width;scattered, small elements of high density especially abundant aroundthe nucleus and presumably representing Golgi bodies; and a delicatelace-work extending throughout the cytoplasm. (1945: 246)

Figure 4: Comparison of images offibroblast from a tissue cultured chick embryo as seen with electronmicroscope (left) and light microscope (right). From Porter, Claude,and Fullam (1945: plate 10).
Besides the challenge of creating preparations sufficiently thin to bepenetrated by the electron beam, researchers confronted a number ofother challenges in preparing biological material for electronmicroscopy. For example, specimens must be placed in vacuum, and thisrequires first removing all water, the primary constituent of cells,without inducing to many distortions. Several historians andphilosophers (Rheinberger 1995, Rasmussen 1997, and Bechtel 2006) haveexamined how biologists confronted these challenges and evaluated thereliability of the resulting micrographs. How they addressed onechallenge—that of creating sufficient contrast in theimages—greatly affected the images that were produced. This wasalready a challenge with light microscopy: cell material is mostlytranslucent so that the light transmitted is mostly of the samewavelength, making it hard to differentiate the various structures inthe image. To address the challenge, light microscopists in themid-nineteenth century began experimenting with dyes used for fabrics.As noted inSection 2, Flemming named the nuclear structures he observed chromatin sincethey bound the aniline dye he was using. The problem for electronmicroscopy was similar—cell components differ little in theirability to block electron transmission. Electron microscopists foundthat several of the fixatives used in light microscopy, especiallythose involving heavy metals, enhanced the ability to block electrons,and accordingly the contrast in the resulting images. Given that therewas little knowledge about what given chemicals would bind to withinthe cell, the investigation of these stains was mostly pursued byresearchers trying out different compounds and procedures for applyingthem (exemplified in Palade’s 1952, study of osmium tetroxide)to see what images they could generate. Indeed, different stains(osmium tetroxide, glutaraldehyde, etc.) did yield differentimages.
Since there was little understanding about what stains bound to,skeptics often raised doubts that they were revealing actualstructures in cells. Bechtel (2000, 2006) highlights threeconsiderations that often figured in scientists’ evaluation ofwhether micrographs were informative about cells or only artifacts ofthe methods of preparation: (1) the quality of the micrographsthemselves—do they exhibit distinct patterns? (2) the robustnessof the results—can comparable results be generated withdifferent techniques (e.g., with light and electron microscopy or withmultiple stains)? and (3) the theoretical plausibility of theresults—do they fit into a coherent theoretical account? Whilethe first is seldom commented on in philosophical accounts, it isnotable that scientists are inclined to assume that if an imagereveals a distinct, replicable pattern, it reflects something in thesource (although this assessment may be retracted if, for example, aresearcher shows how the pattern could be generated by other means).There has been extensive philosophical discussion of thesecond—the inference that when the same result is generated byindependent means, it reflects a preexisting entity in the world(Hacking 1983; Culp 1995; Stegenga & Menon 2017). However, thiscriterion proves insufficient at just the point at which results withnew techniques, such as the electron microscope, are mostcontroversial—when the images contain structures beyond thosethat can be detected with other existing techniques. In these cases,the distinctness of the patterns together with considerations as tothe theoretical plausibility of the findings are the criteriaresearchers can use. Appeal to theoretical plausibility, however,would seem to be circular since in traditional philosophical accounts,theories are tested by the evidence generated by the technique. Butdeveloping a plausible theory that fits with an experimental findingand other evidence is not easy and when they are able to do so,scientists view it as buttressing their judgments that the imagesreflect real structures. To illustrate how these considerations havebeen invoked in in cell biology, we present two examples philosophershave examined of conflicts over whether the structure shown inmicroscopic images was real or an artifact. The cases endeddifferently, one with the acceptance of the structure, one with itsrejection.
The first case involves the Golgi apparatus, first described by Golgi(1898) in light microscope studies using the silver nitrate stain hehad introduced. Palade and Claude (1949a,b), two of the pioneers ofmodern cell biology, who would ultimately share the Nobel Prize,argued that it was an artifact of staining, including with osmiumtetroxide (which Palade helped establish as a primary stain forelectron microscopy). They appealed not only to the variableappearance of the Golgi bodies, as had some earlier skeptics(Strangeways & Canti 1927; Parat 1928; Baker 1944), but also totheir own ability to create myelin figures similar in appearance tothe Golgi bodies by adding osmium to egg white. As discussed byBechtel (2006), this is compelling evidence for the claim that theGolgi apparatus is an artifact, but in this case the evidence waseventually set aside without being refuted. Although Palade remained askeptic about the Golgi apparatus for 15 years (researchers in his labreported that they were not allowed to mention it during thatinterval), he finally accepted its reality when researchers in hislaboratory demonstrated that many newly synthesized proteins passthrough the region of the cell where the Golgi apparatus appeared ontheir way to being secreted. He did not, however, explain why hechanged his mind. This later research on the Golgi apparatus, but nothis earlier skepticism about the existence of the Golgi apparatus, wasnoted in Palade’s Nobel Prize citation in 1974. Palade himselflater contributed to two reviews (Farquhar & Palade 1981, 1998)that discuss earlier researchers who cast doubts on the reality of theGolgi apparatus, but he never mentions his own claims that it was anartifact. What seems particularly salient is that, as a result ofresearch in his laboratory, the Golgi apparatus was associated with afunction in the cell in a manner that it had not been previously. Itnow fit into a plausible theory in which it figured in packagingproteins for export from the cell. Bechtel argues that this is often amajor factor in researchers’ acceptance of the reality ofentities identified through new techniques.
The second example stems from early electron microscopy of bacterialcells. Chapman and Hillier (1953) observed invaginations of the plasmamembrane in gram-negative bacteria, which they calledperipheralbodies. Robertson (1958) renamed themmesosomes and theywere implicated by many researchers in a variety of cell functionsbefore being rejected as artifacts in the 1970s (Silva et al. 1976).Analyzing this example, Rasmussen (1993) argues that philosophicalcriteria for distinguishing real entities from artifacts, such asrobustness, are insufficient to explain scientists’ changingjudgments about the mesosome. When Chapman and Hillier first madetheir case for the existence of mesosomes, they had to explain awaydifferences between their micrographs and the observations of lightmicroscopists by arguing that a membrane that only appeared in imageswith the light microscope was in fact due to the mesosomes beingimaged under low resolution. Rasmussen contends they offered thisconvoluted argument rather than treating the mesosomes as an artifactbecause they were promoting the new electron microscopes. He furtherdescribes how their claims motivated research programs aimed atpurifying and biochemically characterizing the mesosome so as toevaluate proposals regarding their function. This initially supportedclaims to the mesosome’s reality, but other biochemists offeredconflicting evidence. In addition, another new technique for preparingspecimens for electron microscopy by freezing them, which had its ownpassionate advocates, generated micrographs that did not showmesosomes. According to Rasmussen, these competing findings, notrobustness, determined scientists’ judgments aboutmesosomes.
The case of the mesosome has attracted substantial interest from otherphilosophers of science. Culp (1994) challenged Rasmussen’sinterpretation of the history of mesosomes, contending that therejection of mesosomes as an artifact is in fact best explained ongrounds of robustness. In particular, she points to the combined datafrom biochemists that revealed few differences between cytoplasmicmembranes and specimens supposedly from mesosomes and from a newgeneration of electron microscopists that suggested that mesosomesresulted from the fixative, glutaraldehyde, used to generate themicrographs purporting to show them. These robustness considerations,she claims, sufficed to lead the community to reject mesosomes. In alater paper, Rasmussen responds to Culp’s analysis, continuingto maintain that local details, not principles like robustness,determine scientists’ judgments about artifacts:
general principles like robustness are too vague to warrant anythingwhatsoever, because when described in sufficient detail it emergesthat the way such principles are instantiated is in flux—and thedevil is in the details. (2001: 643)
Several other philosophers have also taken up the case of themesosome. Allchin (2000) characterizes the initial evidence for themesosome as robust and describes how subsequent research showing howparticles were generated by degeneration of the membrane inpreparation for electron microscopy led to a reassessment, culminatingin mesosomes being recognized as artifacts. Weber (2005) argues thatthe process of evaluating claims about artifacts employs causalreasoning of the same sort as used in testing theoretical hypotheses.According to him, mesosomes were judged to be artifacts when evidencesupported the claim that they were produced by chemical fixation.Hudson (2014: chapter 2) advances yet another alternative: accordingto which what mattered most to researchers was whether they regardedthe process for generating evidence for or against mesosomes as areliable process.
Microscopic images, whether hand-drawn, as they were in the nineteenthcentury, or captured in photographs, are highly detailed. However, theknowledge cytologists and cell biologists seek to develop is moreabstract and general—they seek to identify the types ofstructures found in cells, not all their instances. Accordingly, cellbiologists frequently generate diagrams that leave out details.Maienschein (1991) examines the origin of this practice inWilson’s (1896) classic text,The Cell in Development andInheritance. Early in the book Wilson provided a diagram (Figure 5A) which abstracts from the much more replete photographs he hadpresented just a year earlier (Wilson 1895). Instead of showing allthe instances of different organelles, he presents just a fewinstances of each type. Organelles are shown using icons that aresuggestive of their shapes. Such diagrams serve to convey the types oforganelles found in cells and their typical location, but falselysuggest that most of the space in the cell is empty. Maienscheincontends that this transition from photographs to abstract diagramsreflects Wilson’s growing confidence in the correctness of hisinterpretation of what he was seeing through the microscope and atransition “from presenting data to representingtheory” (Maienschein 1991: 252). Asign of their theoretical status is that such figures, unlikemicrographs or data plots, often undergo numerous revisions asscientists develop their account. Revising these diagrams is a means,in fact, of developing theory. With the development of biochemical andmolecular accounts of cell phenomena, researchers often make whatSerpente (2011) characterizes as a transition from the iconic to thesymbolic. He presents protein-protein interaction maps and generegulatory diagrams as examples of symbolic representations.

A. Wilson’s (1896: 14, figure 5)cell diagram that leaves out much of the detail that would be seenthrough the microscope to focus on

B. A mechanism diagram of the processesof heterophagy and autophagy from de Duve and Wattiaux (1966: 468,figure 6).
Figure 5
Similar points also apply to another type of diagram that appearsfrequently in cell biology—diagrams of hypothesized mechanismsthat are taken to be responsible for a particular phenomenon (Sheredoset al. 2013; Abrahamsen, Sheredos, & Bechtel 2018). Such diagramsdo not try to show all the organelle types but, as inFigure 5B, only those thought to be involved in generating a particularphenomenon—in this case, the breakdown of materials taken intothe cell (heterophagy) or of cell components themselves (autophagy).One challenge with mechanism diagrams is that they are static whereasmechanisms are engaged in change (the digestion of organelles and theexpulsion of the remaining material). One common strategy is to usearrows to represent activities, although often within the same diagramarrows may have multiple different meanings. Ultimately, however, itis up to the viewer to animate a mechanism diagram (Hegarty1992)—to rehearse mentally the different activities that arerepresented and to imagine the changes that are being produced bydifferent parts.
One might think that diagrams are only important as a means ofillustrating results that are presented textually. But examination ofthe practices of scientists reveals that they are far more central totheir reasoning. Early in the development of a mechanistic hypothesisscientists sketch their ideas. Often mechanism diagrams (as well asthe other figures that show data or the workflow of the research) arecrafted long before text is drafted. Diagrams are commonly presentedin lab meetings and revised multiple times as scientists are refiningtheir claims. Researchers often generate the text of the paper only atthe end of this process. Taking advantage of access to the draftfigures and text for two research projects, Sheredos and Bechtel(2017, in press) examined the process of interactive engagement inwhich the investigators modified diagrams, changed text, and thenfurther modified the diagram. What this process suggests is thatsketching and resketching mechanisms plays a central role asresearchers seek to identify what they can conclude from theirexperimental studies. In one of the cases they examined, an earlyversion of a diagram serves to pose a question that was addressedthrough the experimental studies, resulting in a final diagram thatoffers an answer to the question posed. Beyond supporting theempirical claims of a research project, Jones and Wolkenhauer (2012)provide an illuminating discussion of how diagrams serve to locateinformation required for a computational model in a representation ofthe cellular mechanism that is being modeled,
Historically prominent philosophers of science such as Popper andReichenbach rejected inquiries into how scientific theories arediscovered as not philosophical (see the entry onscientific discovery). Beginning with Hanson (1958), though, discovery has attracted theinterest of many philosophers of science. Investigations intodiscovery in cell biology (and related fields such as biochemistry,molecular biology, and neuroscience) inspired Bechtel and Richardson(1993 [2010]) to argue that the goal in these fields was not toconstruct nomological explanation (Hempel 1965) but rather to identifythe mechanism responsible for a phenomenon and determine how itworked. Examples from cell biology have also figured prominently inaccounts of mechanistic explanation by Machamer, Darden, and Craver(2000), Craver and Darden (2013), Bechtel (2006), and others (seeGlennan & Illari, 2018, for a compendium of contemporary accountsof mechanisms and mechanistic explanation, and the entrymechanisms in science). The conceptmechanism figures both in discussion ofontological issues and epistemic issues, but the two uses can bedistinguished in the manner proposed by G. Allen (2005). In thissection we are concerned with what he called “operative orexplanatory mechanism” (2005: 261; the thesis that forexplanatory purposes components of cells should be conceived tofunctionas if they were machines);Section 3 concerned “philosophical mechanism” (2005: 261; theontological thesis that organismsare or are constituted bymachines).
One of the central objectives of many philosophers of science focusedon mechanistic explanations is to characterize reasoning strategies orheuristics that scientists use to develop mechanistic explanations(Bechtel & Richardson 1993 [2010]; Craver & Darden 2013; Gross2018). We discuss reasoning strategies that figure in different phasesof developing mechanistic explanations in cell biology: delineatingphenomena and situating them in responsible mechanisms, identifyingand characterizing the components of mechanisms (a reductionisticphase), and determining the organization within mechanisms and betweenmechanisms and their contexts (a more holistic phase).
Challenging the common characterization of explanations as explainingdata, Bogen and Woodward advanced the claim that scientificexplanations are targeted at phenomena. Instead of defining whatphenomena are, they give examples: “weak neutral currents, thedecay of the proton, and chunking and recency effects in humanmemory” (1988: 306). From these examples, it is clear that Bogenand Woodward understand phenomena to be repeatable processes that canbe observationally or experimentally detected in multiple ways. Onthis characterization, cell activities such as protein synthesis orcell division count as cellular phenomena. Although phenomena areoften construed as the starting point of research, Craver and Darden(2013: Chapter 4) identify some of the experimental tasks involved incharacterizing phenomena such as identifying precipitating conditions,manifestations, inhibiting conditions, modulating conditions, andnonstandard conditions that alter the manifestation of the phenomenon.They also emphasize the role of multiple experimental techniques inspecifying features of phenomena. A good deal of research in cellbiology is devoted to determining conditions under which cellphenomena such as programmed cell death are initiated or inhibited andcell biologists have been inventive in developing experimentaltechniques needed to produce these phenomena. Bechtel (forthcoming)points out that phenomena that are the target of explanation rangefrom highly specific (protein synthesis in liver cells under lowoxygen conditions) to much more general (protein synthesisgenerically). Individual research projects often address highlyspecific phenomena whereas textbooks or review articles discuss moregeneral phenomena.
While some account of the phenomenon under study is generally adoptedbefore researchers set out to identify the responsible mechanism, thecharacterization often changes radically as research on theresponsible mechanism proceeds (Bechtel & Richardson 1993 [2010],refer to researchersreconstituting the phenomenon). This isillustrated with an example from the study of cells. A starting pointfor early inquiries into how animals store energy for their activitiesfocused on the heat generated by metabolizing foodstuffs. This heatwas assumed to power other activities and approximately a hundredyears of research was devoted to explaining animal heat (seeMendelsohn 1964). However, after Lohmann (1929) and Fiske and Subbarow(1929) had identified adenosine triphosphate (ATP) and demonstratedthat when hydrolyzed it would release considerable energy, heat wasdetermined to be just a waste product of metabolism and researchersinstead focused on the synthesis and hydrolysis of ATP. This becamethe reconstituted phenomenon to which much of the older research wasnow applied.
What is distinctive of mechanistic explanations is that they decomposemechanisms into parts or entities and the operations or activitiesthat contribute to realizing the phenomenon. Although sometimesresearchers can proceed by reasoninga priori about whatcomponent activities would be required to produce a given phenomenon,most often the decomposition is developed on the basis of experiments.While the goal is to identify both parts and operations and to linkthem, a given group of researchers may only have research techniquesthat allow them to identify one or the other. This was the case incell biology. As discussed inSection 1, microscopic research enabled researchers to identify component parts(organelles) of cells, but except for indirect clues such as shape orthe ability of an organelle to take up a stain, these researchers wereunable to procure information about the functions they performed.Biochemists on the other hand were able to identify reactions involvedin many cell activities, but since they studied these in chemicalsoups made by grinding up the cell, they were not able to link theseto cell structures. Cell biology developed with the introduction ofnew techniques such as cell fractionation and electron microscopy thatopened up the possibility of localizing biochemical reactions inorganelles.
It is worth emphasizing that the sense in which decomposition isreductionistic is very different from that involved in moretraditional philosophical accounts of theory reduction, which involvesthe derivation of one theory from another (Nagel 1961; see the entriesonscientific reduction andreductionism in biology). Researchers pursue decomposition without assuming the existence offull theories at either level or that a theory at one level can bederived from one at a lower level. Moreover, in pursuing decompositionresearchers typically do not assume that knowledge of the lower-levelcomponents is sufficient for explaining the phenomena—at aminimum, they recognize that how the components are organized is alsoimportant. The association of a cell process with a specific enzyme isnot the end of the explanatory process—the activity of enzymesis often affected by the context in which the enzyme exists, forexample, in a membrane. For this reason, researchers valueinvitro reconstitution experiments in establishing that they havecorrectly accounted for a phenomenon.
Allchin (1996, 2007) and Weber (2002), analyze an importantreconstitution experiment at the interface of cell biology andbiochemistry that played a major role in resolving a conflict betweentwo opposing accounts of the synthesis of ATP in oxidative metabolism.Biochemists spent a couple decades trying to identify a purelychemical pathway that used the energy released in oxidative metabolismto synthesize ATP (as happens in glycolysis). P. Mitchell (1961)advanced an alternative, chemiosmotic hypothesis according to whichenergy was transferred via the creation of a proton gradient over amembrane. Kagawa and Racker (1966) had already linked ATP synthase tosmallknobs on the inner mitochondrial membrane but it was anin vitro chimeric system combining fragments from bacteria,plants, and animals created by Racker and Stoeckenius (1974) thatdemonstrated that energy could first be captured in a proton gradientand then used in mitochondria to synthesize ATP. Unlike purelybiochemical accounts that discounted cell structure, this researchshowed the importance of not just enzymes but how they are situatedwith respect to cell structures in explaining cellular phenomena.
Even in chemical soups individual molecules are arranged in aparticular pattern. This pattern affects, for example, which moleculeencounters another. When analyzing such soups, however, chemists donot try to decipher the arrangement but instead rely on statisticalmeasures about the likelihood of encounters. But cells aredifferent—different molecules are segregated in differentlocations and how they are organized affects the resulting behavior(this is true even in prokaryotic cells that lack internal membranes,with molecules segregated in different parts of the cell, sometimeschanging location over the course of a day; see Cohen, Erb,Selimkhanov, et al., 2014). Organization is also important inmanufactured products. If the parts of your computer were distributedon your desk, and each part was provided with some input so that itwas performing an operation, the parts would still not carry out thesame activities as they do when they are in their proper arrangement.Organization ensures that the outputs of one component are passed tothe appropriate others as inputs. The importance of organization isfurther recognized when one considers that what human designers do isimpose new organization on existing components to achieve noveldesired effects. Evolution often does the same in biology.
Organization is especially important for attempts to understand theactivities of cells but, until recently, cell biologists have hadlimited tools for determining organization. At a coarse-grained level,the combination of cell fractionation and electron microscopy providedinformation about organization and this provided insight into howcells perform their activities. For example, recognizing that enzymesthat can break apart cell components are segregated from the rest ofthe cell within the lysosome explains why they only carry out theiractivities on structures that are first transported into the lysosome.And knowing that a collection of enzymes is localized in an adjacentmembrane explains how products of one reaction are readily taken up inanother. In these cases the organization realized in cells and itsconsequences can be characterized qualitatively and presented in amechanism diagram, as discussed insection 4.3. In other cases, however, the organization is more complex than can bedescribed in such qualitative terms. Especially when components areorganized into feedback loops, and the individual operations arecharacterized in nonlinear functions, cell biologists turn tocomputational models to understand their behavior. Bechtel andAbrahamsen (2010) and Brigandt (2013) describe a number of examplesfrom recent research on cell phenomena that require computationalmodels and refer to the resulting explanations that apply dynamicalanalysis to mechanistic accounts as “dynamical mechanisticexplanations”. Using computational modeling of spindle formationin cell division as an example, Gross (2018) shows how computationalmodels can go beyond what has been established experimentally andserve as heuristics guiding further research.
In recent years cell biologists have acquired new tools for studyingorganization, many of these advanced in systems biology (Green 2017;see the entry onphilosophy of systems and synthetic biology). One approach in systems biology seeks to develop comprehensivedetailed models of the numerous components identified as involved inspecific phenomena (Gross & Green 2017). High-throughput dataabout, for example, which proteins in a cell can bind to each other orwhich pairs of genes, when mutated together, are lethal, has massivelyincreased the number of cell components associated with any givenphenomenon. To make sense of this cell researchers often seek abstractmodels by constructing networks in which nodes stand for entities andedges for interactions between entities (Green et al. 2018). Thechallenge researchers face is to make sense of thesenetworks—given the large number of entities involved and theirmany interactions, these networks can often appear as hairballs. Webriefly consider two strategies for analyzing networks in cell biologythat philosophers have analyzed.
Alon and his collaborators, focusing on gene transcription networks and metabolicnetworks in bacteria and yeast, identified “recurring,significant patterns of interconnections” involving two, three, or fournodes, which they calledmotifs (Milo et al., 2002: 824). One example, occurring in“hundreds of non-homologous gene systems” (Mangan,Zaslaver, & Alon 2003: 197) in the transcription network ofE.coli, is thefeedforward loop. In a feedforward loop (Figure 6), one unit sends inputs to two other units, the first of which alsosends an input to the other, which serves as the final output unit.Using Boolean models, Alon and his colleagues demonstrated thatdepending on whether the connections are excitatory or inhibitory,such a motif can perform a number of different functions. For example,if all the connections are excitatory and the final output unit isonly active when it receives input from both of the other units (theconnections to it constitute an AND-gate, as shown on the left inFigure 6), then the feedforward loop functions as a persistencedetector—the output unit only becomes active when the input tothe first unit endures at least until the second unit becomes active.This, as Mangan et al. explain, protects the output (which might serveto start the transcription of a gene) from being generated in responseto random noise presented to the input unit. The feedforward loop andAlon’s other motifs are characterized abstractly withoutspecific reference to what entities and interactions correspond to thenodes and edges. Philosophers who have examined this research refer tothese organizational patterns asdesign principles (Green2015; Green, Levy, & Bechtel 2015; Levy & Bechtel 2013).
Figure 6: Three examples of feedforwardloops studied by Alon and his colleagues
The second strategy employs computational tools such as clusteranalysis to find collections of nodes that are especially highlyinterconnected. To interpret these clusters in terms of mechanismsoperative in the cell, researchers often align them with Gene Ontology(GO), a resource that was developed to represent published informationabout cell components, molecular functions, and biological processesin directed acyclic graphs so as to facilitate communication betweenresearchers working on different species (Ashburner et al. 2000).Leonelli (2016) examines many of the epistemic issues the developersof GO confronted in developing such a resource. Beyond its originalfunction, GO is now widely used to interpret network graphsmechanistically. By annotating nodes in cellular networks withinformation from GO when it is available researchers interpret theseclusters as corresponding to known mechanisms or sometimes asconstituting previously unknown mechanisms. Researchers makeinferences about nodes for which such information is lacking usingprinciples such asguilt by association—the entity isinferred to occur in the same location or to contribute to the sameprocess as those with which it clusters (Bechtel 2017 analyzes several examples).
As a result of computational models and analyses of large networks,cell biology in the twenty-first century is much more focused onorganization and appears much less reductionistic than it did in themid twentieth century. For some philosophers, this reliance on networkanalyses represents a move away from mechanistic explanations towardsa process view (Théry 2015; Nicholson 2018; Dupré 2012).Huneman (2010, 2018) construes the explanations resulting from networkanalysis as a distinct type of explanation he labelstopologicalexplanation. Others, however, argue that since these analysesstill draw upon the components constituting the system and thebiologists developing them continue to think of them as mechanistic,we should extend the conception of mechanistic explanation to includethe holistic focus on organization developed in computational andnetwork analyses (Baetu 2015; Bechtel 2015; Levy 2014).
We have focused on core domains of cell biology, concerned withexplaining the basic activities of cells. There are, though, manyspecialized fields concerned with cells. Some of these have becomeactive areas for philosophical inquiry in their own right. Here wemerely identify some of these and point readers to relevant work byphilosophers, including in several cases entries in the SEP. One suchdomain is microbiology, concerned with single-cell organisms, whetherprokaryotic (lacking a nucleus and other organelles) or eukaryotic.O’Malley (2014) has identified distinctive features ofmicroorganisms (e.g., lateral gene transfer) and explores how featuresof microorganismal life challenge major assumptions about livingorganisms that have resulted from focusing predominantly onmulti-cellular organisms. In other work (O’Malley 2010), she hasexamined from a philosophical perspective the competing hypothesesabout the origin of eukaryotic cells from prokaryotic cells. Much ofcell biology has focused on mature cells of specific types and hasattended less to the processes by which cells in multi-cellularorganisms differentiate from a common cell, known as astemcell. Fagan (2013) has pioneered the philosophical examination ofstem-cell research, including the advent of techniques to revertmature cells to stem cells. Cells are not only transformed indevelopmental processes in multi-cell organisms, but also in diseasessuch as cancer in which cells not only proliferate in uncontrolledways but also defeat many normally operating cellular mechanisms thatnormally prevent proliferation. Plutynski (2018) has identified manyof the philosophical challenges arising in the efforts to explaincancer (see the entry oncancer). An importantcapability of many multicell organisms is the ability to detectpathogens, viruses, and other threats and defend against them. Suchimmune responses require, among other things, the capacity todistinguish cells that belong to the organism itself from others (seethe entry onphilosophy of immunology). A fundamental issue that arises in many of these contexts as well astopics raised above is whether and how one should conceptualize cellsas individuals or agents (see the entry onthe biological notion of individual).
In addition to approaching cells as objects of scientific study,contemporary researchers often adopt an engineering approach to cells.One context is in the domain of synthetic biology in which researchersengineer cells, sometimes for research ends but other times togenerate products society finds useful (see the entry onphilosophy of systems and synthetic biology). The recent development of gene editing tools such as CRISPR opens upboth epistemic and ethical issues (for the ethical issues, see theentry onneuroethics). One context in which attempts to synthesize cells has played acentral role is in the attempt to understand the origins of life. Muchof the research on origins of life involves the development ofprotocells—self-organized, spherical systems composed of lipids(Rasmussen et al. 2009). This has become a prominent topic of inquiryfor both theoretically minded biologists and philosophers of biology(see the entry onlife; Bedau 2012; Moreno 2016; Dunér, Malaterre, and Geppert2016).
How to cite this entry. Preview the PDF version of this entry at theFriends of the SEP Society. Look up topics and thinkers related to this entry at the Internet Philosophy Ontology Project (InPhO). Enhanced bibliography for this entryatPhilPapers, with links to its database.
biological individuals |cancer |emergent properties |immunology, philosophy of |life |molecular biology |neuroethics |reduction, scientific |reduction, scientific: in biology |scientific discovery |systems and synthetic biology, philosophy of
We are most grateful for the helpful comments provided on earlierdrafts of this manuscript by Douglas Allchin, Sara Green, Arnon Levy,Alan Love, Jane Maienschein, Karl Matlin, Maureen O’Malley, AnyaPlutynski, Andrew Reynolds, Hans-Jörg Rheinberger, James Tabery,and Marcel Weber.
Copyright © 2019 by
William Bechtel<bechtel@ucsd.edu>
Andrew Bollhagen<abollhag@ucsd.edu>
View this site from another server:
The Stanford Encyclopedia of Philosophy iscopyright © 2023 byThe Metaphysics Research Lab, Department of Philosophy, Stanford University
Library of Congress Catalog Data: ISSN 1095-5054