-------- ------

SLIME MOLD

Evelyn Fox Keller

.~.\

~...'

'.'

,

' ''e'."X.~.::"~.,

;" ' .. '(~....:;:'. ' "'-' --=-;;:""".:;' ""; _."

..

...,~, "l> '>j r

'''.' ~ ..

Perc('l\lit1g i~ nul '\ 1Uillt,'r (If JJ'j:J·;iv<:,ly nlluw;f'.t!;;;~ Qrgdl1 - <;;1~

Of si~hl or hC:lfl;1g--!() It·(ci\·(~~, r('u:!y TI]c.ei( in,pn',-Sl(l1) from\qthOHt !jj;e ,( r8.krtx r,."d\J11~ '1"'1 fJ \1"1. Re,:'}gni.~ing

r,lnd lTffi(-mbering a.fL' 1)01 Ill, t:tT:, o! ,tl'Tinf lIJ_' old 'llagtb uf

paf'",t Impressiuns. I:. is ge:lcrally r'gre,:l! that ull our lrapressiolls,uc SChC:l\lllicnlly determil1t-d from the start. As p"[ceivers wesl'!t>ct from all tllt' slil11u]J frd!inq; <>n our ~efJ,;,:'" only l!Io_cwbich jJltcTe~l us, and our ime:TM" arc go'.trncd by;, ;JaUcrll­Jn8king tC];(lf'Jlcy, o:ometllll\\', called $rhenw. h~ a chaos orshifling im[)l'cs<.,Jon;;, c8vh of Lli'; uJ11fc>lnlCls <1 staLle world III

which objects h,l\'~ ITcognizalJ1e shnpC',>, nn; localed in deptnano hnvc p'.. rm8nCn(T. In p{'rceiving Vi" 'J.P.' bll.iJd{),~, laking:'\onw co l~~, nnd r~.icctir:golher~ Thr rnO"l ,lC(''-'i1<Il.I>j,> eu,,"!! are

Ihos,.. whi< h fit mOtil \~,lsily into lh~ ])Ht,-crn thid i" b"inp buillt,p. t\ll1biV~wu":lor:es \end 10 be T[t'ilWd w, if they J1iJfln()J)il~d

v.l1h th" ';;1 oj 'he pHHe,'i1 lJi',;cordanl on,.,; tod [0 h ... re­

jeC::'?d. Ii th~,y ',le .\' "r>:' J tr,l F:nh'IU,'- ,f. ;11ll'jl,wm hll.s.to hr; jTI'Ahhed. As l\:arning prQ"ceu,' ,)bj, ;]1 11,1111(,,1 ThelrJlWf!(',," rilt'l1 ,Jlf~cl the way '.I .', ;JfrLo:'i.,-,;cd ,": lil:W' oncel.'1beled liwv iH<' JJJOre "lk:'diL lulted into plgeoll-Iwko, in theJ"utun,

!\, ;;rp.... gut's on ilJl{!':-'!Jnirl'Lc\,'> r'!~ up, we m;,!:r <.I

gTr8ter ,1nd g1'o::0.1<'1. i.nv"~',llrwn1 in Oi...f Sy:;'Tf'~1 0:' j(lb< i>:;,

- ?'-lary Dou)i;hr., Purity (J/1(!l)wlyl'r

I take an organism as my object, the lowly amoeba-likeprotist, Dictyostelium. In times of plenty, it lives as anindividual single-celled organism but, when food sup­plies are exhausted, it regroups. Then, this one-celledprotist becomes part of a complex multicellular motileslug capable of producing fruiting bodies and spores,and of migrating in search of greener pastures wherethe new spores can germinate .• Over the years, it has at­tracted a great deal of scientific interest, partly becauseit so elegantly exemplifies a primitive form of biologicaldevelopment, and partly for the paradoxes it embodies.On the one hand, here is a single-celled organism, exist­ing in a population of apparently identical organisms,and on the other hand, it is a part of a differentiatedorganism assuming a particular role and structure inthe larger entity, the multicellular organism. Here is anobject that traffics back and forth both between the oneand the many and between sameness and difference.For me, this simple being, in its rich ambivalence, hasserved as an intellectual touchstone, a sustaining objectthroughout my academic career. Over and over, mywork would confront me with a dilemma and this objectwould resurface to help, offering itself as a model for anentirely new way to think about it. For me, slime moldhas unarguably served as an object-to-think-with.

My first encounter with Dietyostelium came in1968. I was working at Cornel! Medical College in NewYork City, looking for ways to fruitfully apply mathe­matical methods to biological problems. Lee Segel, anapplied mathematician from Rensselaer Polytechnic In­stitute was visiting Cornell for the year, and we teamedup to tackle a couple of problems that looked as thoughthey might be tractable, among these the problem ofslime mold aggregation. The onset of aggregation is the

298 Evelyn fOlt K.ll.r

first visible step in the process that eventually leads tothe cellular differentiation observed in the multicellularorganism. Prior to aggregation, there is no apparent dif­ference among cells. But once it occurs, aggregation cre­ates a differential environment for the cells, and thereforeit could presumably account for subsequent cell differ­entiation. The problem is, what sets off the aggregation?Is there some hidden-from-view prior difference, a dif­ference that then serves as the trigger for the develop­ment of more elaborate, structured, and clearly visibleheterogeneity? Does the onset of organization in fact re­quire the existence of such a preexisting "cause"? Mostbiologists seemed to think so, and they hypothesizedsuch prior structures under the name of "founder cells. It

Or was it possible that organization might emerge spon­taneously, out of the dynamics of the population as awhole? The first possibility held little appeal for Segeland me: first, it only pushed the question of the originof heterogeneity further back (where did the foundercells come from?); second, we could find no evidence forthe existence of such specialized cells. We set out todemonstrate the feasibility of the second possibility­the notion that organization could emerge from the dy­namics of the population as a whole.

The model we developed was a highly simplified­in fact, clearly oversimplified-representation of the ac­tual biological case. Like other mathematical modelstraditionally employed in the natural sciences, it in­cluded just enough of the known biological factors togive rise to the essential phenomenon. At its heart, themodel demonstrated that no designated initiator, foundercell, Dr organizer was required for understanding the ad­vent of aggregation in a uniform distribution of previ­ously undifferentiated cells. We showed that clusters of

~I;m.. Mn1<l 'J.QQ

amoeba would result from the collective dynamics of apopulation in which a change in external conditions (inthis esse, depletion of the bacteria that served as theirfood source) induced a change of slate, nnd indeed, thesame change in state, in each individual amoeba. Ouraccount of the onsel of differentiation in at least onekind of biological development offered a way to resolvethe paradox (how does highly structured differencearise from similarity?) that so sharply divided geneticsfrom embryology. We assumed that it would be of inter­est to biologists.

But we were wrong-not so much in our model as in ourexpectations. Biologists, for the most part, showed littleinterest in our ideas, and despite the absence of evi­dence, continued to adhere to the belief that foundercells (or pacemakers) were responsible for aggregation.At the time, Segel and I were disappointed and per­plexed, but only after ten years had passed did I see howthis fact was of interest in itself. What made my newrecognition possible was a shift in intellectual mindsetand the focus provided by a sharp question posed on theother end of a telephone.

By the early 1980s, I had found a new calling. rhadmoved my intellectual center of gravity from theorelicalphysics, molecular biology, and mathematical biology toissues of gender and science. To me, its questions werecompelling: How to liberate science from its history ofattachment to masculinist ideologies? How to under­stand the implications of the very different approachto science manifested in work such as that done by thegeneticist Barbara McClintock? McClintock had nottried to separate herself from her objects of study-corncells-to stay more ·objective." She imagined herselfmore ~among them," herself reduced to their size, per­haps as a way of becoming one of them.'

So, in 1981, slime mold aggregation was far frommy mind when, one evening, 1 received a call from Alan

300 Evelyn Fox Keller

Garfinkel, a recent convert to mathematical biology. Hehad recently come across the paper Segel and I had writ­ten on the subject, but he had not been able to find anyfollow-up, either in the form of critique of the work orex­pansion OJ' the ideas. He asked what was going on. Heasked if there was a conspiracy. The call took me aback,not because [ thought there had or had not been anyconspiracy, bUl because it immediately brought into fo­cus a problem I had been struggling with around thedisparity between McClintock's perceptions and thoseof her colleagues. I had been wondering why accountsof biological processes that<r'brought such explana­tory satisfaction to her colleagues had failed to satisfyMcClintock. And vice versa? The sharply worded tele­phone call reminded me that when I had tried to talk tobiologists about our model of slime mold aggregation Ihad experienced that same wall blocking both interestand understanding.

Suddenly, I saw my own experience as an exampleof a general phenomenon-a widespread disposition tokinds of explanation that posit a single central causallo­cus (governor, founder cell, pacemaker)-and that thisdisposition was crucial in understanding the gap be­tween conventional understandings of biol~gical devel­opment as DNA driven and McClintock's own moredynamic proposals. Following David Nanney, J referred tosuch explanations as ~master-moleculetheories· and be­gan to wonder why it should be that people tend to findsuch accounts more natural and conceptually simplerthan global, interactive accounts in which causal force isdistributed.3 One possible hypothesis that seemed plau­sible to me was that we tend to project onto nature ourfirst and earliest social experiences, ones in which we feelpassive and acted upon. But in any case, I wrote,

As scientists, our mission is to understand andexplain natural phenomena, but the words under­stand and explain have many different meanings.

In our zealous desire for familiar models of expla­nation, we risk not noticing the discrepancies be·tween our own predispositions and the range ofpossibilities inherent in natural phenomena. Inshort, we risk imposing on nature the very storieswe like to hear.·

Another twenty years have passed since my workon scientists and their preferred narratives of nature.My intellectual preoccupations have shifted again. Formore than a decade I wrote about genetics and devel­opmental biology, and today I find myself turning todevelopmental psychology. Insofar as my focus hasremained on the nature of developmental processes perse, the shift has been but a small step, and a new com­munity of intellectual allies was easy to identify. For ex­ample. in my work on developmental biology, I hadmounted a strong critique of the concept of the "geneticprogram" (understood as a program for development en­coded in the DNA), and 1 found an entire school of psy­chology engaging in similar arguments. Its scholarswere making strong claims about the value of dynami­cal systems theory (or understanding developmentalprocesses. What drew these psychologists to dynamicalsystems theory was the language that theory providedfor describing the emergence of novel patterns of or­ganization in complex, nonlinear systems, patterns thatcould not have been predicted from studying the behav­ior of individual components in isolation.

To me, this had the ring of deja vu, but much hadchanged over the years since my early foray into dynam­ical systems. In particular, familiarity with examples ofself-organization-in physics, in computer science, andeven in biology (where slime mold aggregation has be­come a canonical example)-now extends beyond theseacademic communities. One effect of a more commonknowledge of examples of self-organization is that, overthe last fifteen years, a series of proposals from different

..n.., 5'....1.." F.... I(.lI...

disciplinary quarters urge the reframing of all psycho­logical and biologicalargumenls in t.erms of dynamicalsystems. 5

The word reframing here is crucial. Almost all ofthese proposals suggest alternatives to conventionalframings of development in terms of programs (eithergenetic or developmental), where the very term programis seen as implying the unfolding or elaboration of in­nate capacities. The authors who have taken up thecause of dynamical systems see in this approach an an­tidote to the prevailing innatism of so much of contem­porary writing in biology and cogrfltive science.

All of these authors seek to redress what they seeas the disproportionate emphasis currently placed oninternal factors of development; I have enormous sym­pathy with their concerns. But in this particular in­tellectual dispute, I found myself jolted out of anycomforting alliance by a sharp recollection of slimemold, a call from my past. Slime mold aggregation isillustrative of self-organization in dynamical systems,but does it not equally well illustrate the power of a de­velopmental program embodied in an individual cell?And if it does, then how can it be said that dynamicalsystems offer an alternative to notions of program? In·deed, I am prompted to ask, can a viable distinction be­tween the two even be made? For me, it cannot. Theimplications of this heresy are large; they include, forexample, the possibility that the battle lines against in·natism need, yet again, to be redrawn.

Let us return then to Dictyostelium. Our earlymodel was deficient in many ways, yet our central pointis still valid: the aggregation of a population of single­celled amoebae (and its subsequent development intomulticellular organisms) proceeds spontaneously, with·out the need for distinctive founder cells; the populationemerges as the product of decentralized and local inter­actions among molecules secreted by individual cells.In other words, despite the elucidation of its genetic

organization, Dictyostelium has survived as a simpleand compelling model of a self-organizing dynamicalsystem. Given the current state of controversy I returnto my touchstone to ask: can we therefore say that thereis no developmental program for this organism?

That depends, of course, on what we mean by aprogram. The most relevant definition given by the Ox­ford English Dictionary has two parts: first, ~a definiteplan or scheme of an intended proceedings," and sec­ond, Man outline or abstract of something to be done."6Both suffer, in this context, from an objectionable degreeof anthropomorphism. There is no ~intention" guidingthe development of an organism, nor is there anywherean agent "doing" the work. Nevertheless, developmentalprocesses proceed along rather well-defined tracks andconclude with quite predictable outcomes. Remarkablylittle is left to chance in a developing organism-in fact,it might be said that one of the fundamental character­istics of biological development is the capacity to resistthe effects of the myriad vicissitudes the growing em­bryo inevitably encounters. 7 Thus, to the extent that wecan rid the notion of program of its anthropomorphicconnotations and think of it simply as a plan or schemeof a proceedings with a definite outcome, a plan thatneed not be located in a particular structure or homun­culus but that may instead be distributed throughoutthe system, I would argue that the very reliability ofmost forms of biological development demands the ex­istence of a program.

The key idea of a plan or program for reliable de­velopment is that the organism (or machine) must beable to resist the disturbances that can throw it offcourse, either by suppressing or by adapting to them. Inother words, such a program must have contingencybuilt into it-instructions, if you will-for how to re­spond to a range of different kinds of input. In the caseof slime mold, the single cell needs to have a change-of­state plan-a plan for changing certain key cell param-

304 Evelyn Fox K.n.,

eters when the food runs out. More sophisticated organ­isms are equipped with programs (or built-in strategies)for changing state in response to changes in a far largerset of parameters. Acomputer program may not be sucha bad image after all, but think of it as a program for sur­vival. Such a program (or strategy) no more requires an­ticipation than does any other function that has beenevolved by the process of natural selection.

The main challenge for the notion of a program lo­cated inside the individual cell lies elsewhere: if we areto grant the existence of a program inside the individual(undifferentiated) slime mold cell, that program mustnot only allow for the change of st-tlte in that cell to betriggered by starvation but also for the reproduction ofthe cell. Without reproduction, there will be no popu­lation, and without a population of cells, there can beno aggregation. But after fifty years of work, most ofwhich has been inspired by John von Neumann's earlyefforts, this problem too seems to have been resolvedfor programs. Today, programs for reproduction-invirtual even if not in physical space-have become ubiq­uitous. There are of course still problems. and theseproblems-primarily having to do with the programs'lack of robustness-are largely responsible for their con­finement in virtual space. In this sense, real organismsremain far ahead.

As an object-to-think-with, slime mold has provento be an immense resource. I am grateful for this oppor­tunity to pay it homage. Some have resisted its message,but there is, too, the danger of overusing it. There arelimits to what it has to teach us. The particular route tomulticellularity it manifests is, after all, a rather primi­tive one; furthermore, it bears little resemblance to thedevelopmental process by which most complex organ­isms come into being. Slime mold may be equipped witha program for adapting to the scarcity of food, but thedevelopmental programs of higher organisms must dealwith a far larger range of variability, and evolution has

Slime Mold 305

L 1-,

What makes an object evocative?l As I write, Bodies. anexhibition of preserved humans from China, is on tour.internationally. Its objects, poised between death andnew animation, raise questions about the sanctity ofwhat has lived, the nature of art, and the human beingswho once were the objects on dispT!ly. Thinking aboutthe uncanny, about thresholds and boundaries helpsus understand these objects with their universal pow­ers of evocation.

And yet, the meaning of even such objects shiftswith time, place, and differences among individuals.1

Some find the preserved bodies the fearsome creaturesof night terrors. For others, they seem almost reassur­ing, an opportunity to contemplate that although deathleaves matter inert, a soul may be eternal.

To the question "What makes an object evocative?"this collection offers pointers to theory (presented asepigraphs) and the testimony of its object narratives,voices that speak in most cases about familiar objects­an apple, an instant camera, a rolling pin. One role oftheory here is to defamiliarize them. Theory enables us,for example, to explore how everyday objects becomepart of our inner life: how we use them to extend thereach of our sympathies by bringing the world within.

As theory defamiliarizes objects, objects familiar·ize theory. The abstract becomes concrete, closer tolived experience. In this essay I highlight the theoreticalthemes ofeach of the six parts of this collection (with spe­cial emphasis on objects and the inner life) in the hopethat theory itself will become an evocative object. Thatis, I encourage readers to create their own associations,

Sherry Turkle

WHAT MAKES AN OBJECTEVOCATIVE?

'..~

Evelyn Fox Keller is Professor Emeritus of Historyand Philosophy of Science in the Program in Science,Technology, and Society at MIT.

equipped them with an extraordinary repertoire of waysof adapting to such variability. The world challengesthem anew each and every day and in ways that couldnot possibly be met with a single tool, or even a few, orperhaps not even with a finite number of tools. Slimemold, in its capacity for self-organization, illustratesone strategy for survival, and it is undoubtedly a versa­tile and fertile object-to-think with. But ultimately morecomplex living beings find the need of a far larger reper­toire of strategies than this little organism can possiblybe expected to display.

30(; Evelyn FoIC Keller