Hypothesis Based Research

"It is what we think we know that keeps us from learning."

Claude Bernard¹

Surgery, as all of medicine, arose from empirical arts. Surgeons, pragmatic as a group, have always tended to believe and report what they can or potentially could directly sense (see, feel). Other branches of medicine, on the other hand, are more accustomed to dealing with the unseen: inadequate insulin throughout the body, micro-organisms buried somewhere within an organ. Perhaps in part owing to this fact, medical sciences accepted earlier hypothetical thinking: a search for biological explanation if you will, explanation of those things they could not directly observe. Whether or not this is the correct explanation, one needn't extensively survey medical and surgical literature to realize surgeons have typically described, rather than explained the results of their endeavors, and used either simple (and often simplistic) explanations for their observations. Recent events, however, suggest a "maturation" of approach in the "orthopaedic sciences." My arguments for maturation largely arise from the history and philosophy of science, and my thoughts emanate from those of others, although I accept sole responsibility for the synthesis.

Maturation requires recognition and acceptance of limitations. Engineering approaches have historically served to "explain" the locomotor apparatus owing to the obvious analogy with inanimate structures. Yet, if engineering has provided one of the greatest strengths of orthopaedic sciences, just as surely such analogies frequently mislead. Engineering, as surgery, arose from empirical endeavors. Mechanistic explanation wandered into engineering only in the 1700 and 1800's. (I hasten to add substantial introduction of mechanistic explanation ventured into surgical thinking nearly a century later). Numerical methods, the very foundation of engineering, worked extremely well for most structural purposes. I will therefore explore this limitation in some detail.

"Entia non sunt multiplicanda praeter necessitatem."

"The number of entities used to explain phenomena should not be increased unnecessarily."

Occam's Razor

William of Ockham (ca. 1280-1349)

William of Ockham (Ockegem) taught us to seek the simplest explanation consistent with observations. As a means to predict behavior, Occam's razor often applies in the physical sciences. One need only consider the example of Newtonian physics which predicts astonishingly well for phenomenon except at the speed of light and at the level of atomic particles. But in biology, which Elsasser notes is "unfathomably complex," (Elsasser, 1991) simple explanations rarely suffice owing to the almost inevitably elaborate chains of (causal) physical or chemical reactions underlying all biological responses. Simplistic explanations abound in the surgical sciences.

For example, "Wolff's Law" suggesting principal tensile and compressive stresses govern bone adaptation remarkably fails in predicting as readily as Newton's Laws succeed (Wolff, 1892). This is not to say seeking simple explanation is not an efficient means of advancing biological thought, only that appropriately simple explanations or hypotheses are most efficient. While Wolff's Law focused attention on the role of the mechanical environment on connective tissue adaptation, I would argue Wolff's simplistic law has done more harm than good, because for more than a century it diverted attention from the non-continuum nature of tissues and from temporal aspects of the mechanical environment likely more critical than merely load magnitude². Perhaps more importantly, Wolff's writing seemingly confused prediction with explanation³. Owing to the pervasiveness of explicit or implicit mechanical (i.e., mathematical) explanation in orthopaedic sciences, I will explore this point in some detail, but imagine we had instead emphasized a "Roux's Law" based upon Wilhelm Roux's earlier and less restrictive teleological ("Zweckmässig") concept of "functional adaptation." (Roux, 1881)

In suggesting the most appropriately simple explanations, we must fundamentally understand science. "Science," as noted by the contemporary philosopher Robert Richards, "is not a disinterested examination of the structures of reality. Objective truths can be captured only in the thicket of cultural belief, refined experience, and honed intuition." (Richards, 1992) We never know how Nature works, we only develop explanations or models of how Nature works. In the sense all explanations are models, numerical models are like all others. However, in the sense numerical models do not address causal chains of events, they are distinct from current (molecular) biological models.

In particular, as scientific explanation of Nature, mathematics fails. Naomi Oreskes and her colleagues, compellingly argued in Science: "Verification and validation of numerical models of natural systems is impossible . . . The primary value of models is heuristic." (Oreskes, et al., 1994) That is, we can never prove the veracity or truthfulness and we can never prove the logical internal consistency in an open model system (i.e., one where all parameters are not identified and known in adequate detail). At best, we can confirm a model, meaning observations predicted by the model are consistent with those actually observed. This view is a far cry from that held by most of us accustomed to numerical models. If I do not entirely agree with her assessment of model utility, her assertion of impossibility of verification and validation is unquestionably correct.

Let me quote rather extensively from Whitehead: "Mathematics is thought moving in the sphere of complete abstraction from any particular instance of what it is talking about. So far is this view of mathematics from being obvious, that we can easily assure ourselves that it is not, even now, generally understood. " (Whitehead, 1925) Whitehead wrote this statement many decades ago, yet it is as true to day, as then. He continues: "For example, it is habitually thought that the certainty of mathematics is a reason for the certainty of our geometrical knowledge of the space of the physical universe. This is a delusion which has vitiated much philosophy in the past, and some philosophy in the present. This question of geometry is a test case of some urgency. There are certain alternative sets of purely abstract conditions possible for the relationship of groups of unspecified entities, which I will call geometrical conditions . . . . (Furthermore) . . . the certainty of mathematics depends upon its complete abstract generality. But we can have no a priori certainty that we are right in believing that the observed entities in the concrete universe form a particular instance of what falls under our general reasoning." Whitehead further commented, ". . . the equations of physics provide little or no basis for the layman's belief in causal connection as conceived by earlier philosophical discussions of the matter." The layman's intuitive view of causality, he realized, was somehow abandoned by many mathematicians, philosophers, and scientists. He continues, using a somewhat confusing double negative, "The criticisms of this rationalistic view of causal connection gradually undermined these convictions so that today it is not too much to say that they have been abandoned by a considerable number of philosophers and are in no way operative in the practice of scientists." Scientists, and I would add engineers and surgeons (as exemplified by Wolff, and more recently Pauwels and others), had somehow come to accept a seemingly magical connection between mathematics and causality⁴, and, Whitehead believed, made a fundamental error. Whitehead understood mathematics was a creation of humans, not something inherent in Nature. We apply our abstract creation to Nature; Nature does not apply mathematics to Her domain. Nature is ignorant of mathematics.

Let me also cite at some length, Julius Weinberg, from his essay "Causation:" "The development of natural science since the seventeenth century has tended to emphasize functional determination, for example, as expressed in the generalizations of physics and chemistr y, rather than regularities of succession⁵..." (Weinberg, 1973) He correctly notes, "The conviction that some kind of uniformity governs the play of events in the natural world has been one of the most influential beliefs of man since the beginning of human reflection. Attempts of various kinds, as we have seen, were made to base this conviction on the deliverances of reason. In particular, the belief that the causal maxims could be established by the purely logical (i.e., mathematics), dominated almost the entire history of the subject." Weinberg concludes, "The criticisms of this rationalistic view of causal connection gradually undermined these convictions so that today it is not too much to say that they have been abandoned by a considerable number of philosophers and are in no way operative in the practice of scientists." Weinberg's assertion of abandonment of mathematical causality, in contrast to Whitehead's belief they were persistent, is unfortunately overly optimistic. Many scientists, particularly in mathematically based fields still cling to ancient ideas. Ironically, Newton well understood mathematics did not imply causality: "Hypotheses non fingo, " he remarked, in the Principia Mathematica: "I propose no explanation."

The point made by Oreskes, Whitehead, Weinberg, and many others is mathematics plays no role in causality. Mathematics has documented its powerful ability to describe and even predict regularities in Nature; it cannot, however, establish causality. This distinction must be made, since mathematical models can never explain Nature's mechanisms. In the interest of causality, we must distinguish description or prediction from explanation.

It would be easy to appear overly critical of both engineers and orthopaedic surgeons in their failure to recognize the lack of link between mathematics and causality. As I earlier mentioned, both arose from empiricism, where prediction is critical, but explanation is not. Traditionally, engineers and orthopaedic surgeons have rarely been interested in how some physical phenomenon works, but they have been critically interested in whether it works. Only recently has the question of "how?" become common with biomechanicians and surgeons for at least two reasons: First, as a practical matter, if some intervention works well, how it works is less critical, and second, only recently have the problems of surgery been amenable to intervention at mechanistic levels. Today, an increasing number of medical interventions depend upon intimate knowledge of causality: the capacity to interfere with the cause of some ordinarily complex chain of events. Mathematical models will not help us here, except in a descriptive (i.e., statistical) or predictive sense. Associations suffice for prediction, but not causal explanation. A mathematical model of bone adaptation which accurately predicts bone behavior in select situations should not be thought to explain behavior. That does not mean mathematical models are of no utility, quite the contrary: we cross oceans on planes developed through numerical theory; predictive numerical models are likely to afford the potential to ascertain optimal surgical reconstructions in the foreseeable future. Rather, we must recognize they are inherently limited for causal explanation. This limitation has few practical effects when we have no means of beneficially intervening in causal connections; it has major effects when, as is now the case, we do.

Beyond recognition of such inherent limitations in science, fields of knowledge, just as humans and societies, go through developmental stages on the journey toward maturity. The eminent and late philosopher of science, Thomas Kuhn, remarked, ". . . once current views of nature were, as a whole, neither less scientific nor more the product of human idiosyncrasy than those current today." (Kuhn, 1970) Consider alchemy which developed in the dark and middle ages to a high state, then faded in the Renaissance being replaced by chemistry and physics and even pharmacology. Empirical arts progress through periods of observations, then predictions, and finally explanations using ever more efficient and sophisticated paradigms. Surgical sciences are now experiencing this sort of scientific maturity.

"When I use a word," Humpty Dumpty said in a rather scornful tone, "it means just what I choose it to mean - neither more nor less."

Through the Looking Glass

Lewis Carroll

If I may paraphrase Humpty Dumpty, "When I use the term scientific maturity, I mean just what I choose it to mean - neither more nor less." (Carroll, 1998) What do I mean by scientific maturity? Certainly moving from observation to explanation. Yet, I imply ever more efficient conduct and effective reporting of science through hypothetical thinking.

And what do I mean by "hypothetical thinking" or "hypothesis"? An hypothesis is simply an explanation of a group of relevant yet selected biological or clinical observations, or a prediction or deduction which necessarily and logically arises from some explanation. I might equally employ the terms "theory," "view," "model," or any number of related words and without hierarchical implications, but, taking license from Humpty Dumpty, I choose "hypothesis" or "explanation." As will become apparent, any testable hypothesis has an unambiguous answer. I would argue one of the principal signs of scientific maturation includes reporting of hypothetical thinking.

I recently ascertained whether past abstracts of the Orthopaedic Research Society posed and meaningfully addressed hypotheses: that is, focused statements explaining observations so formulated as to have unambiguous answers. I considered the titles and first paragraph or two, as well as the final paragraph of the first 25 abstracts in each year of the ORS Transactions. Realizing legitimate hypothesis- and design-driven studies (i.e., those developing technological approaches) might be differently reported, I then characterized the work, liberally interpreting design-driven not only for devices, but also experimental or technical design. For these, I included clear objectives or goals.

In the early years I encountered few statements I deemed effective hypotheses or objectives (Figure 1). When the abstracts did explicitly mention purposes, they were most often descriptive: "We characterized, or investigated, or determined, or quantified...x, y, or z": observations, if you will, not explanations. In many cases, abstracts provided little or no rationale for their observations. I do not imply investigators did not think of some hypothesis - only they reported none.

Figure 1.

Percent of ORS abstracts having clear objectives (solid bars) or hypotheses (open bars) from 1976 until 1997. Note an increasing percent, particularly after 1993. Note, however, only 25% of all abstracts in 1997 had clear objectives or hypotheses.

In more recent years, I encountered increasing numbers of reasonable hypotheses or objectives. In 1993, ORS abstract instructions to authors requested explicit objectives, questions, or hypotheses. Thus, in part, their presence in the abstracts relates to explicit instructions, although I would argue this, too, implies maturation.

Now, I make no claim for scientific validity of my little study. Rather, I argue we are more effectively reporting our investigations. Effectively conveying findings is as critical as efficiently conducting studies, since the most brilliant theory or experiment remains incomplete until effectively communicated. These data also suggest we are moving from an era of observation to an era of explanation. Might I add, we have a way to go in pursuing explanations: seventy-five percent of the 1997 ORS abstracts remained more or less isolated observations.

"The least questioned assumptions are often the most questionable."

Paul Broca

Meaningful collaboration is surely another critical sign of maturity, for it suggests diverse sophistication of observation. More importantly, true collaboration implies a willingness to consider the often disparate ideas of others, an implied willingness to question assumptions: assumptions of the sort the eminent 19th Century French neuropsychologist, Paul Broca, considered most questionable. A molecular biologist may challenge an assumption long accepted by an engineer, or vice versa. Knowledge seldom advances without abandoning cherished assumptions. Collaboration blurs manmade inter-disciplinary boundaries: boundaries of which Nature is unaware. How many scientific articles in our literature now include the authorship of engineers, molecular biologists, geneticists, even immunologists, psychologists, and (of all people) surgeons? How many included such diverse authorship even twenty years ago?

The Weber brothers, in 1836, published one of the very first collaborative scientific endeavors, which was not coincidentally, the first scientific work on locomotion. (Weber and Weber, 1836) Neither Wilhelm, a mathematician, nor Eduard, an anatomist, had any background in locomotion. Neither had a particular interest in locomotion. Scientific interest played no role in their motivation. Rather, they commented in their foreword, "And though we are convinced the choice of our objective requires no justification, we do not wish to hide our true motivation . . . It was the joy we found in a unified endeavor; an endeavor to which each of us brought unusual strength and abilities and which became all the more valued and treasured because of the very lack of resources. Man is never more capable or persistent in scientific study than when there is mutual participation and excitement, not only at the completion of the work, but during its entire course." Imagine our endeavors if we simply sought and found those with whom we enjoyed working.

Such meaningful collaboration implies trust and respect. We collaborate precisely because others know what we do not know, and others can do what we cannot. We cannot truly know another's limits, and while trust and respect derive partly from experience, without trust and respect true collaboration is impossible. Surely the ability to trust implies maturity.

In presuming maturation of a field, what changes can we anticipate? While I suggest more efficient conduct of science as well as more effective reporting, I will address primarily the former and note but six means to conduct more efficient explorations.

First, we must formulate the most effective hypotheses. The pre-eminent philosopher of science, Sir Karl Popper recognized, "We do not know: we can only guess . . ." (Popper, 1968) Accordingly, he introduced the concept of "high" and "low informative content hypotheses": High informative content hypotheses were those for which the answers were most uncertain: guesses if you will. They must be unique. From these we are most likely to gain critical information. Low informative content hypotheses, on the other hand, were those for which the answers were almost certain. From formulations such as, "We hypothesize we can develop a model to do whatever" we gain little new insight or information. We can always fit some model to select data, but that process, in and of itself, does not guarantee a meaningful general model, so the hypothesis as such is trivial. Poorly formulated hypotheses or questions fail to efficiently and effectively answer questions. Well-formulated questions, on the other hand require time and effort and, I hasten to add, creativity. Focus on explanation rather than technical development, and emphasis on high informative content hypotheses confers efficiency. I am not suggesting we must not advance technology, only that focus on explanation is most efficient.

Second, Popper additionally and compellingly argued we can never prove any explanation, we can only disprove them. Henrí Poincaré, the French mathematician and philosopher recognized this point when he said, "If a phenomenon admits of a complete mechanical explanation it will admit of an infinity of others which will account equally well for all the peculiarities disclosed by the experiment." As Poincaré, Popper, Oreskes, and many others (including Phaedrus, the protagonist of Zen and the Art of Motorcycle Maintenance (Persig, 1974)) recognized, all sets of observations may be equally well explained by an infinite number of choices limited only by our creativity. Thus, even with what we consider experimental or theoretical support and appropriate confirmation, we can never insure our own explanations are the correct ones. In fact, Popper went so far as to suggest we should design experiments to disprove our hypotheses, since we could never entirely "prove" them. This does not mean our explanations cannot be of practical effect: quite obviously our explanations suffice to clone species and create gene therapy. However, attempting to disprove explanations raises different approaches than attempting to prove explanations.

Third, formulating explanations based upon the largest practical number of observations lends efficiency. One needn't review many manuscripts to realize even technically well-formulated hypotheses are typically based upon few observations. Since added observations may effectively falsify explanations, considering more at the outset is both efficient and effective. While more sophisticated observations obviously advance a field, more effective hypothetical thinking never depends entirely on technical sophistication.

Fourth, efficient hypothetical thinking demands succinct, unambiguous questions or hypotheses that minimize the possibility experiments will fail to clearly address critical issues. Imagine we formulate only questions that can be answered "yes" or "no" by our theoretical or experimental design; imagine we always formulate hypotheses which can be plainly supported or refuted. Seemingly small differences in wording of explicit questions or hypotheses can make large differences in experimental design or choice of variables. Furthermore, no hypothesis can be tested which is not posed in terms of independent and valid dependent variables. Other hypotheses may have heuristic value, but cannot be tested. We should practice proposing questions and hypotheses in measurable terms. Good questions inevitably imply scientific design, and vice versa.

Fifth, once we have posed a high informative content question or hypothesis, we should conduct thought experiments. Imagine every conceivable outcome. Plot out potential results. Imagine the implications of each. This procedure will often result in reformulating the question owing to its heuristic insights.

Sixth, in advance of initiating our experiments, as we formulate and reformulate our hypotheses, we should imagine the most simple approaches which will address the real questions. We might even alter approach to take advantage of some more simple, yet powerful technique.

Imagine we formulated only high informative content hypotheses. Imagine in these formulations we identified and considered ten relevant observations, not merely one or two or three. How many explanations could we quickly eliminate from among the immense pool? Imagine our questions had unambiguous answers. Imagine we used only the minimum essential approach. How many unnecessary experiments could we avoid? How many resources would we save? And how much better the answers.

Decades ago, when technical sophistication was not what it is today, Sir Bertrand Russell commented: "One of the troubles of our age is that habits of thought cannot change as quickly as techniques with the result that as skill increases, wisdom fails." Technology seductively and perniciously focuses attention on observation, not explanation. In a culture that worships technology, we should pay heed to Russell's observation. Wisdom dictates we spend greater time formulating hypotheses as technical sophistication increases. Imagine we spent even a fifth of the time formulating hypotheses or questions as we did solving technical problems. What scientist spends one full day in the workweek merely reflecting? Yet, is not the mind our most powerful tool?

These six suggestions enhance efficiency. Yet effective science requires more than efficiency, more than wisdom: it demands creativity. The creative scientist must prepare for resistance to novel ideas, and unfortunately at times even personal attacks. Thomas Lounsbury astutely commented, "We must view with profound respect the infinite capacity of the human mind to resist the introduction of useful knowledge." We humans have infinite capacity to resist accepting new and useful knowledge, particularly when the ideas arise from a "competitor." As Persig's Phaedrus realized, ". . . no one is willing to give up the truth as he sees it . . ." (Persig, 1974) Most of our scientific views are not arbitrary, but based upon considerable past, if not current reflection. Thus, it is not surprising we do not easily discard what we have learned with such effort to believe. But to repeat Bernard's admonition: "It is what we think we know that keeps us from learning." Any framework we maintain rests on limited or even filtered information, not any complete or even coherent framework, and this fact alone should make us be more willing to abandon tenaciously-held views. The requirement for creativity is both bane and joy of the scientist. Bane because we stubbornly cling to old, comfortable, and uncreative habits of thought; bane because creativity is the most difficult task we confront as scientists; bane because we encounter such resistance to novelty. Joy because creativity affords unexcelled, if often only internal rewards for the scientist. Joy because in the long run, little in science matches creativity for efficiency and effectiveness.

Scientific maturity, however, implies more than even efficiently, wisely, and creatively conducted explorations. Maturity implies certain attitudes and conduct in science, ethics if you will. As Kuhn emphasized, the history of science teaches us sooner or later our explanations will be replaced as new concepts, connections, observations, or paradigms appear. Without negating their heuristic value, we must consider all explanations tentative.

The ephemeral nature of our explanations and our inherent inability to prove them should imbue a spirit of tolerance and humility: a humility driven by historical, philosophical, and ethical imperatives. Humility, I hasten to add, is unfortunately far from universal among scientists.

Bronowski further argued the value of science to society resided principally in its search for truth. (Bronowski, 1965) "Truth," he recognized, "is the drive at the center of science; it must have the habit of truth, not as dogma but as a process . . ."—truth as a process to be transmitted to and incorporated within society. Truthfulness accompanies maturation of a field: truthfulness not only in the design, conduct, and reporting of our work, but in interactions with colleagues—an implicit acknowledgment our approaches, observations, explanations may someday be deemed inadequate or even seriously flawed. The concept of truthfulness applies to oneself: the scientist who deeply admits his or her own ideas will be replaced will less likely resist the new ideas of another, and less likely deceive him—or herself.

Openness accompanies truthfulness. "Scientific knowledge, like language, is intrinsically the common property of a group or else nothing at all," Kuhn suggested. (Kuhn, 1970) Common property, as described by Kuhn, is nowhere more obvious than in the Human Genome Project where newly-discovered gene sequences belong to the community, not the individual. One needn't reflect long to recognize the critical value of common property and the candor it implies.

What have trust, respect, humility, truthfulness, and openness to do with scientific maturity? Each of these qualities intrinsically follows maturation. And, do they not contribute to the wisdom suggested by Russell? Our challenge is to conduct ever more mature investigations: restricted resources and boundless technology demand ever more efficient conduct and effective communication in a spirit of humility, tolerance, truthfulness.

But even creatively, wisely, efficiently conducted investigations do not suffice. One must responsibly and effectively convey those explanations to the broader community. The implications must be clear for the widest audience, for we may never predict which explanations, which observations, will affect someone in a diverse field. (I equate parochial reporting—not to be confused with detailed, focused investigations—with irresponsibility.) Contemporary searching capabilities insure our ideas potentially reach a large audience, but they will never reach the appropriate persons unless we write in broad and inclusive terms.

I make no pretense I fully understand how to teach what I intellectually and even to some degree emotionally understand. Einstein commented, "The real difficulty, the difficulty that has baffled the sages of all times, is rather this: how do we make our teaching so potent in the emotional life of men that its influence should withstand the pressures of the elemental psychic forces in the individual?" Our psyches contain immense forces opposing trust, opposing respect, opposing complete truthfulness, opposing openness—forces resisting the finest conduct of science. And those forces act towards ourselves as well as towards others. Yet, we must recognize and resist those forces to practice the most wise, efficient, effective, and mature science.

Once we as teachers employ efficient and ethical science and effective reporting, we must actively teach the processes of science to students as best we can. Example of the mentor provides a critical key for any student. However, we cannot presume students innately grasp those processes, or that mentors somehow magically transfer them merely by example. We cannot presume maturity occurs in a vacuum. Just as fields mature owing to newly-identified thought which changes processes, we must teach these processes. We must encourage and support the creativity of our students, as well as our own. We must teach students fair, respectful, thorough, constructive peer review. As experienced by the Weber brothers, we must communicate to the young the "Joy of Science": pleasure in the process, not merely in techniques or outcomes. The ethics and responsibilities we identify and teach to students must withstand the elemental psychic forces within all of us, for only with such an approach is our field maturing.