Strong Inference - KnowledgeContext

Strong Inference John R. Platt Science, New Series, Vol. 146, No. 3642. (Oct. 16, 1964), pp. 347-353.

Stable URL: Science is currently published by American Association for the Advancement of Science.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in technology. For more information regarding JSTOR, please contact support@.

Mon Jul 16 12:46:57 2007

16 October 1964, Volume 146, Number 3642

SCIENCE

Strong Inference

Certain systematic methods of scientific thinking may produce much more rapid progress than others.

J o h n R. Platt

Scientists these days tend to keep up a polite fiction that all science is equal. Except for the work of the misguided opponent whose arguments we happen to be refuting at the time, we speak as though every scientist's field and methods of study are as good as every other scientist's, and perhaps a little better. This keeps us all cordial when it comes to recon~mendingeach other for government grants.

But I think anyone who looks at the matter closely will agree that some fields of science are moving forward very much faster than others, perhaps by an order of magnitude, if numbers could be put on such estimates. The discoveries leap from the headlinesand they are real advances in complex and difficult subjects, like molecular biology and high-energy physics. As Alvin Weinberg says ( I ) , "Hardly a month goes by without a stunning success in molecular biology being reported in the Proceedings of the National Academy of Sciences."

Why should there be such rapid advances in some fields and not in others? I think the usual explanations that we tend to think of-such as the tractability of the subject, or the quality or education of the men drawn into it, or the size of research contracts-are important but inadequate. I have begun to believe that the primary factor

T h e author is professor of hiopliysics and physics a t the University of Chicago, Chicago, 111. This is the text of a n address given before the Division of Physical Chemistry of the American Chemical Society in September 1963. under the title "The New Raconians."

in scientific advance is an intellectual one. These rapidly moving fields are fields where a particular method of doing scientific research is systematically used and taught, an accumulative method of inductive inference that is so effective that I think it should be given the name of "strong inference." I believe it is important to examine this method, its use and history and rationale, and to see whether other groups and individuals might learn to adopt it profitably in their own scientific and intellectual work.

In its separate elements, strong inference is just the simple and oldfashioned method of inductive inference that goes back to Francis Bacon. The steps are familiar to every college student and are practiced, off and on, by every scientist. The difference comes in their systematic application. Strong inference consists of applying the following steps to every problen~ in science, formally and explicitly and regularly:

1 ) Devising alternative hypotheses;

2) Devising a crucial experiment (or several of them), with alternative possible outcomes, each of which will, as nearly as possible, exclude one or more of the hypotheses;

3) Carrying out the experiment so as to get a clean result;

1') Recycling the procedure, making subhypotheses or sequential hypotheses to refine the possibilities that remain; and so on.

It is like climbing a tree. At the first fork, we choose-or, in this case,

"nature" or the experimental outcome chooses-to go to the right branch o r the left; at the next fork, to go left or right; and so on. There are similar branch points in a "conditional computer program," where the next move depends on the result of the last calculation. And there is a "conditional inductive tree" o r "logical tree" of this kind written out in detail in many first-year chemistry books, in the table of steps for qualitative analysis of a n unknown sample, where the student is led through a real problen~of consecutive inference: Add reagent A; if 1011get a red precipitate, it is subgroup alpha and you filter and add reagent B; if not, you add the other reagent, B'; and so on.

On any new problem, of course, inductive inference is not as simple and certain as deduction, because it involves reaching out into the unknown. Steps 1 and 2 require intellectual inventions, which must be cleverly chosen so that hypothesis, experiment, outcome, and exclusion will he related in a rigorous syllogism; and the question of how to generate such inventions is one which has been extensively discussed elsewhere (2, 3 ) . What the forn~alschema reminds us to do is to try to make these inventions, to take the next step, to proceed to the next fork, without dawdling or getting tied up in irrelevancies.

Jt is clear why this makes for rapid and powerful progress. For exploring the unknown, there is no faster method; this is the minimum sequence of steps. Any conclusion that is not a n exclusion is insecure and must be rechecked. Any delay in recycling to the next set of hypotheses is only a delay. Strong inference, and the logical tree it generates, are to inductive reasoning what the syllogism is to deductive reasoning, in that it offers a regi~larmethod for reaching firm inductive conclusions one after the other as rapidly hs possible.

"But what is so novel zlhout this?" someone will say. This is rhe method of science and always has been; why give it a special name? The reason is that inany of us have alnlost forgotten

16 OCTOBER 1964

it. Science is now a n everyday business. E,quipn~ent, calculations, lectures become ends in themselves. How many of us u r l t e clown our alternatives and crucial experiments every day, EocusIng on the e,xclu\ion of a hypothesis? We niay write our scientific papers so that it looks as if we had steps 1, 2, and 3 in mind all along. But in between, we do busywork. We become "method-oriented" rather than "problem-or~ented." We say we prefer to "feel our way" toward generalizations. W e la11 to teach our students how to sharpen up their inductive inferences. And wc do not realize the added power that the regular and explicit use of alternative hypotheses and sharp exclusions could give us at every step of our research.

The differcnce between the average scientist's informal n~ethods and the methods of the strong-interence users i\ so~ilewhat l ~ k e the difference between a gasoline engine that fires occasionally and one that fires in steady sequence. If our motorboat engines were as ejrdtic as our deliberate inlellectr~alefforts, rvlost of us would not get home for supper.

Molecular Biology

'The new molecular biology is a fielci where I thlnk this systematic nietliod of inference has become wide4pread and effective. It is a complex Field; yet a succession of crucial experiments over the past decade has given us a surprisingly detailed understanding of hereditary mechanisms and the control of enzynle formation and protein synthesis.

The logical structure shows in every cxperitiient. In 1953 James Watson and Francis Crick proposed that the DNA molecule-the "hereditary substance" in a cell---is a long twostranded helical molecule ( 4 ) . This suggested a number ot alternatives tor crucial test. D o the two strands of the helix stay together when a cell divides, or do they separate? Matthew Mcselson and Franklin Stahl used an ingenious isotope-den5ity-labeling technique which showed that they separate (5). Does the D N A helix always have t ~ 7 0 strands, o r can it have three, i i ~atomic niodels suggest? Alexander Rich 5howed it can have either, depending on the ionic concentration (6). These are the kinds of experi-

ments John Dalton would have liked, chemist or scientist in any field ac-

n h e r e the combining entit~csare not customed to less closely reasoned arti-

atoms but long n~acronlolecular cles and less sharply stated inferences

strands.

will find it a salutary experience to

O r take a different sort of question: dip into that journal almost at random.

Is the "genetic map"-show~ng the sta-

tistical relationship of different genetic

characteristics in recombination exper- Resistance to

iments-a one-dimensional map like Analytical Methodology the DNA ~ilolecule (that is, a linear

map), as T. H . Morgan proposed in This analytical approach to biology

191 1, o r does it have two-d~mensional has sometimes become almost a cru-

loops or branches? Sey~ilour Benzer sade, because it arouses so much re-

showed that his hundreds of fine micro- sistance in many scientists who have

genetlc experi~ilentson b a c t e r ~ awould grown up in a more relaxed and diffuse

fit only the n~athematical matrix for tradition. At the 1958 Conference on

the one-dimensional case ( 7 ) .

Biophysics, at Boulder, there was a

I3ut of course, selected crucial ex- dramatic confrontation between the

periments ot this kind can be found two points of view. Leo Szilard said:

in every field T h e real difference in "The problems of how enzymes are

molecular biology is that tornial in- induced, of how proteins are synthe-

cIuctive inference is so systematically sized. of how antibodies are formed,

practiced and taught. O n any given are closer to solution than is generally

rnornlng at the 1aboratory of Molecu- believed. If you d o stupid experiments,

lar Biology in Cambridge, England, and finish one a year, it can take 50

the blackboards of Francis Crick or years. But if you stop doing experi-

Sidney Brenner will commonly b e ments for a little while and ihinlc how

found covered with logical trees. On proteins can possibly be synthesi~ed,

the top line will be the hot new result there are only about 5 different ways,

just u p from the laboratory o r just in not SO! And it will take only a few

hy letter or rumor. On the next line experiments to distinguish these."

will be two o r three alternative ex- One of the young men added: "It

planations, or a little list of "What h e is essentially the old question: How

did wrong." Underneath will be a se- snzall and elegant a n experiment can

ries ot suggested experiments or con- you perform?"

trols that can reduce the number o t These comments upset a number of

possibilities. And so on. The tree grows those present. AII electron rnicroscopi5t

during the day as one m a n or another said. "Gentlemen, thi5 is off the track.

comes in and argues about why one This is philosophy of science."

a t the experiments wouldn't work, o r Szilard retorted, "1 was not quarrel-

how it should be changed.

ing with third-rate scientists: 1 was

T h e strong-inference attitude is evi- quarreling with first-rate scientists."

dent ju\t in the style and language in A physical chemist hurriedly asked,

which the papers are written. For ex- "Are we going to take the official

ample, in analyzing theories of anti- photograph before lunch o r after

body formation, Joshua Lederberg lunch?"

( 8 ) gives a list of nine propositions

But this did not cleflect the dispute.

"subject to denial," discussing which A distinguished cell biologist rose and

ones would be ''most vulnerable to said, "No two cells give the same

experimental test."

properties. Biology is the science of

T h e papers of the French leaders heterogeneous systems." And h e added

F r a n ~ o i s Jacob and Jacques Monod privately, "You know there are rci-

are also celebrated tor their high "logi- entirts; and there arc people in science

cal density," with paragraph aftcr para- who are just working with these over-

graph of linked "incl~ictive 5yllogisn1s." simplified model systems-DNA chains

Brit the style is widespread. Start with and in vitro systems---who are not

the first paper in the Journal o f Mo- ctoing science a t all. W e need their

lec~llarBiology f o r 1964 (9). and you auxiliary work: they build apparatus,

immediately find: "Our conclusions . . . they make nzinor studies, but they are

might be invalid if . . . (i) . . . (ii) not scientists."

. . . o r (iii). . . W e shall describe ex- Ts which Cy Levinthal replied:

periments which eliminate these al- "Well, there are two kinds of biolo-

ternatives." The average physicist or gists, those who are looking to see

SCIENCE, VOL. 146

if there is one thing that can be understood, and those who keep saying it is very complicated and that nothing

can be understood. . . . You must

study the sirrlplest system you think has the properties you are interested in."

As they were leaving the meeting, one man could be heard muttering, -'\?'hat does Szilard expect me to doshoot myself?"

Any criticism or challenge to consider changing our methods strikes of course at all our ego-defenses. Hut in this case the analytical metl~odoffers the possibility of such great increases in efroctiveness that it is unfortunate that it cannot be regarded more often as a challenge to learning rather than as a challenge to combat. Many of the recent triumphs in molecular biology have In fact been achieved on just such "oversimplified model systems," very much along the analytical lines laid down in the 1958 d~scussion.They have not fallen to the kind of men who justify themselves by saying, "No two cells are alike," regardless of how trut that mav ultimately be. The triumphs are in fact triumphs of a new way of thinking.

High-Energy Physics

This analytical thinking is rare, but it is by no means restricted to the new biology. High-energy physics is another field where the logic of exclusions is obvious, even in the newspaper accounts. For example, in the fanlous discovery of C . N . Yang and T . D. Lee, the question that was asked was: Do the fundamental particles conserve mirror-symmetry or "parity" in certain reactions, or do they not? The crucial experiments were suggested; within a few months they were done, and conservation of parity was found to be excluded. Richard Garwin, Leon Lederman, and Marcel Weinrich did one of the crucial experiments. It was thought of one evening at suppertime; by midnight they had rearranged the apparatus for it. and by 4 a.m. they had picked up the predicted pulses showing the noncon5ervation of parity (10). The phenomena had just been waiting, so to s ~ e a k ,for the explicit forniulation of the alternative hypotheses.

The theorists in this field take pride in trying to predict new properties or

new particles explicitly enough so that from the new accelerators and the

if they are not found the theories will million-dollar costs of operation have

fall. As the biologist W. A. H. Rush- forced a similar analytical approach.

ton has said ( I ] ) , "A theory which It pays to have a top-notch group

cannot be mortally endangered cannot debate every experiment ahead of

be alive." hlurray Gell-Mann and time; and the habit spreads throughout

Yuval Ne'eman recently used the parti- the field.

cle grouping which they call "The

Eightfold Way" to predict a n~issing

particle, the Omega-Minus, which was Induction and Multiple Hypotheses

then looked for and found (12). But

one alternative branch of the theory Historically, I think, there have been

would predict a particle with one-third two main contr~butions to the de-

the usual electronic charge, and it was velopnlent of a satisfactory strong-

not found in the experiments, so this inference method. The first is that of

branch must be rejected.

Francis Bacon (13). H e wanted a

The logical tree is so much a "surer method" of "finding out nature"

part of high-energy physics that some than either the logic-chopping or all-

stages of it are con~monly built, in inclusive theories of the time or the

fact, into the electronic coincidence laudable but crude attempts to make

circuits that detect the particles and lnductlons "by silnple enumeration."

trigger the bubble-chamber photo- He did not merely urge experiments,

graphs. Each kind oi particle should as some suppose; he showed the fruit-

give a different kind of pattern in the f ulness of interconnecting theory and

electronic counters, and the circuits can experiment so that the one checked

be set to exclude or include whatever tbe other. Of the many inductive pro-

types of events are desired. If the dis- cedures he suggested, the most im-

tinguishing criteria are sequential, they portant, 1 think, was the condi-

may even run through a complete logi- tional inductive tree, which pro-

cal tree in a microsecond or so. This ceeded from alternative hypotheses

electronic preli~ilinaryanalysis. like hu- (possible "causes," as he calls them),

man preliminary analysis of alterna- through crucial experiments ("ln-

tive outcomes, speeds up progress by stances of the Fingerpost"), to exclu-

sharpening the criteria. It eliminates slon of some alternatives and adoption

hundreds of thousands of the irrele- of what is left ("establishing axioms").

vant pictures that formerly had to be HIS Instances of the Fingerpost are

scanned, and when it is carried to its explicitly at the forks in the logical

limit, a few output pulses, hours apart, tree, the term being borrowed "from

may be enough to signal the existence the fingerposts which are set up where

of the antiproton or the fall of a loads part, to indicate the several di-

theory.

rections."

1 think the einphasis on strong in- Many of h ~ csrucial experiments pro-

ference in the two fields I have men- posed in Book IT of T h e h e w Orgnnon

tioned has been partly the result of ale st111 fascinating. For example, in

personal leadership, such as that of order to decide whether the weight of

the classical geneticists in ~ilolecular a body is due to its "inherent nature,"

biology. or of Szilard with his "Mid- as some had said, or is due to the

west Chowder and Bacteria Society" attraction of the earth, which would

at Chicago in 1948-50, or of Max decrease with distance, h e proposes

Delbriick with his summer courses in comparing the rate of a pendulum

phage genetics at Cold Spring Harbor. clock and a spring clock and then

But it is also partly due to the nature lifting them froin the earth to the top

of the fields then~selves.Biology, with of a tall steeple. H e concludes that if

its vast informational detail and com- the pendulum clock on the steeple

plexity, is a "high-information" field. "goes more slowly than it did on ac-

where years and decades can easily count of the diminished virtue of its

be wasted on the usual type of "low- weights . . . we may take the attrac-

information" observations or experi- tion of the mass of the earth as the

ments if one does not think carefully cause of weight."

in advance about what the most im- Here was a method that could sepa-

portant and conclusive experiments rate off the empty theories!

would be. And in high-energy physics, Bacon said the inductive method

both the "information flux" of particles could be learned by anybody, just like

16 OCTOBER 1964

learning to "draw a straighter line or

more perfect circle . . . with the help

of a ruler or- a pair of compasses." "My way of discovering sciences goes Ear to level men's wit and leaves but !ittle to inclividual excellence, because it pertorms everything by the surest rules and demonstrations." Even occasional mistakes would not be fatal. "Truth will sooner come out from crror than from confusion."

Tt is easy to see why young minds leaped to try it.

Nevertheless there is a difficulty with this method. As Bacon emphasizes, it is necessary to make "exclusions." H e says. "The induction which is to be available for the discovery and demonstration of sciences and arts, must analyze nature by proper rejections and cxclusions; and then, after a sufficient number of negatives, come to a conclusion on the affirlnative instances." "[To n-ian] ~t is granted only to proceed at first by negatives, and at last to end in afirniatives after exclusion has been exhausted."

Or, as the philosopher Karl Popper says today, there is n o such thing as proof in science--because some later altertiative explanation may be as good or better-so that science advances only hy disproofs. There is no point in making hypotheses that are not falsifiable, became such hypotheses do riot say anything: "it must be possible for an empirical scientific systeni to he rcfuted by experience" (14).

T h e difiiculty is that disproof is a hard doctrine. If you have a hypothesis and 1 h'ive another hypothesis, evidently one of them must be eliminated. The scientist seems to have no choice hut to be either soft-headed or disputatious. Perhaps this is why so many tend to resist the strong analytical approach--and why some great scientists are so disputatious.

Fortunately, it seems to me, this difficulty can be retnoved by the use of a second great intellectual invention. the "tiiethod of niultiple hypotheses," which is what was needed t o round out the Baconian scheme. This is a rnethod that was put forward by T. C. Chamberlin (15), a geologist at Chicago at the turn of the century, who is best known for his contribution to the Chamberlin-Moulton hypothesis of the origin of the solar systeni.

Chaniberlin says our trouble is that when we make a single hypothesis, wc become attached to it.

"The moment one has offered an

original explanation for a phenome- t h o ~ ~ g h t "that Chamberlin described,

riot1 which seems satisfactory, that mo- the reason for the sharpness, the ex-

ment affection for his intellectual child citement, the zeal, the teamwork-yes,

springs into existence, and as the ex- even international teamwork-in mo-

planation grows into a definite theory lecular biology and high-energy phys-

his parental affections cluster about his ics today. What else could be so ef-

offspring and it grows more and more fective?

dear to him. . . . There springs up When niultiple hypotheses become

also unwittingly a pressing of the the- coupled to strong inference, the sci-

ory to make it fit the facts and a entific search becomes an emotional

pressing of the facts to make them powerhouse as well as an intellectual

fit the theory. . . .

one.

"To avoid this grave danger, the Unfortunately, 1 think, there are

lnethod of multiple working hypotheses other areas of science today that are

is urged. It differs from the simple sick by cotnparison, because they have

working hypothesis in that it distributes forgotten the necessity for alternative

the effort and divides the affections. hypotheses and disproof. Each man

. . . Each hypothesis suggests its own has only one branch-or none-on the

criteria, its own means of proof, its logical tree, and it twists at random

own lnethod of developing the truth, without ever coming to the need for

and if a group of hypotheses encom- a crucial decision at any point. W e

pass the subject on all sides, the total can see from the external syn~ptotns

outcome of means and of methods is that there is something scientifically

full and rich."

wrong. The Frozen Method. The Eter-

Chaniberlin thinks the tiiethod "leads nal Surveyor. The Never Finished. The

to certain distinctive habits of ~ n i n d " Great Man With a Single I-lypothesis.

and is of prime value in education. T h e Little Club of Dependents. T h e

"When faithfully followed for a sufi- Vendetta. Thc All-Encompassing The-

cient time, it develops a mode of ory Which Can Never Be Falsified.

thought of its own kind which niay Some cynics tell a story, which niay

be designated the habit of conlplex be apocryphal, about the theoretical

thought. . . ."

chemist who explained to his class,

This charming paper deserves to be "And thus we see that the C-CI

reprinted in some more accessible bond is longer in the first compound

journal today, where it could be re- than in the second because the percent

quired reading for every graduate stu- of ionic character is stiialler."

dent-and for every professor.

A voice from the back of the rootii

It seems to me that Chatiiberlin has said, "But Professor X, according to

hit o n the explanation-and the cure the Table, the C-Cl bond is shorter

-for many of our problen~s in the in the first con~pound."

sciences. The conflict and exclusion "Oh, is it?" said the professor.

of alternatives that is necessary to "Well, that's still easy to understand,

sharp inductive inference has been all because the double-bond character is

too often a conflict between men, each higher in that compound."

with his single Ruling Theory. But T o the extent that this kind of story

whenever each Inan begins to have is accurate, a "theory" of this sort is

multiple working hypotheses, it be- not a theory at all, because it does

conies purely a conflict between ideas. not exclude anything. It predicts every-

It beco~iiesmuch easier then for each thing, and therefore does not predict

of us to aitii every day at conclusive anything. It becotiies simply a verbal

disproofs-at .strong inference-with-

formula which the graduate student

out either reluctance or cotiibativeness. repeats and believes because the pro-

I n fact, when there are multiple hy- fessor has said it so often. This is not

potheses which are not anyone's "per- science. but faith; not theory, but

sonal property" and when there are theology. Whether it is hand-waving

crucial experiments to test them, the or number-waving or equation-waving,

daily life in the laboratory takes on a theory is not a theory unless it can

a n interest and excitement it never be disproved. That is, unless it can

had, and the students can hardly wait be falsified by some possible experi-

to get to work to see how the de- mental outcome.

tective story will come out. It seetiis In chemistry, the resonance the-

to m e that this is the reason for the orists will of course suppose that I

cleveloptiient of those "distinctive hab- am criticizing then?. while the molecu-

its of mind" and the "complex lar-orbital theorists will suppose I am

SCIENCE, VOL. 146

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download