A Motivation for the Book

Why write a book about the scientific hypothesis? In particular, why a book about defending the hypothesis?

First, most of us, ordinary citizens and scientists alike, do not really know what a hypothesis is and yet it touches nearly every aspect of science.

Second, despite its traditional place in science, the hypothesis today is suf­fering simultaneously from neglect and attacks from critics who will tell you that it is no longer relevant; that, in fact, it is detrimental to science.

The book has two major goals: to explain what a hypothesis is and what it does, and to show why its critics are wrong.

Ordinary citizens and scientists need to be able to evaluate the claims of sci­ence about global climate change, the safety of vaccines, the risks of cancer, or any of the many other science-related challenges that keep confronting us. To as­sess competing claims, we need to know how to distinguish stronger, more solid, and reliable claims from weaker and less reliable ones. The claims that science makes, no matter what they're called, are hypotheses—putative explanations for phenomena—and to be able to assess the claims we need to know how to eval­uate hypotheses. It is not difficult to do once you know how. But it is not an ob­vious or easy thing to do, especially in the face of the criticisms and contradictory information that we're bombarded with. The purpose of this book is to make rea­soning with and about the hypothesis and understanding science easier.

Why is the book so long? I could use a loose analogy with what the Nobel Prize- winning physicist, Richard Feynman, once said. He claimed that a seem­ingly simple idea—“the atomic hypothesis... that all things are made of atoms— little particles that move around in perpetual motion, attracting each other when they are a little distance apart but repelling when being squeezed into one an- other”—contains “an enormous amount of information about the world,...

To give your imagination something to work with, I wanted to go into the details of scientific thinking that your teacher never told you about, probably be­cause she had never been taught them herself. What I found was that the subject is a lot deeper—and so the book is a lot heavier—than anticipated.

In the beginning, I didn't set out to write a book of any kind, though. Actually, the project was started by a simple question:“What is my hypothesis?”

Although the graduate student appeared earnest, I wondered at first if she was joking. She was very bright and was well advanced on the research phase of her PhD thesis in a colleague's laboratory. Her project focused on a novel aspect of the neuroscience of rat behavior that might be related to human mental illness. She had done many experiments and gotten masses of high-quality data. Her experiments were focused on specific topics; this was not an open-ended “dis­covery science” approach. Surely, I thought, she must have a good idea of why she is doing each experiment, what the likely outcomes are, how the series of experiments hangs together—in short, an idea of what her hypothesis was? As we discussed her project, it seemed that the different pieces of the investigation were running along in parallel; near each other but not really intersecting.

“What is the working title of your dissertation?” I asked, because a well-chosen title usually announces its major theme.

“I'm not sure at this point.”

We took a step back and started talking about why she was interested in the topic, what the central unanswered questions were, and what she thought that her results were leading to. But her question had touched a nerve. A clearly for­mulated hypothesis would have helped her organize her thinking, yet she wasn't sure she had one, even though, like all students in our program, she had learned about hypotheses in her first-year graduate Proseminar course.

Unfortunately, this vignette was not unique. Even very good students often had trouble integrating the concept of a hypothesis into their thinking. We, the faculty, seemed to be failing them somehow. We were doing a good job of getting across the mechanics of laboratory research. The students became adept at putting together specific experiments, carrying out procedures, and collecting and analyzing data; however, they often missed a broader sense of what was going on in an investiga­tion, whether it was their own or someone else's. What exactly was the problem?

I got one clue when a senior colleague and I began co- teaching a section on sci­entific thinking in Proseminar. He conducted an exercise in which the students had to pick out true hypotheses from a list that included scientific statements that were not hypotheses. It quickly became apparent that he and I disagreed on the answers. Although we soon papered over the problem, I found this in­cident profoundly disturbing. We were both strong advocates of hypothesis­based scientific thinking, and I had naturally assumed that we had a common understanding of what a hypothesis was. The experience taught me otherwise. Evidently the concept was not as universally well understood as I had thought, and if we senior scientists did not agree on what a hypothesis was, then what on earth were we teaching? No wonder the students were struggling!

Another clue came in the form of a blog post² entitled, “Hypothesis over­drive?” by Jon Lorsch, Director of the National Institute for General Medical Sciences at the National Institutes of Health and an important biomedical sci­ence thought leader.

The blog set-up implied that a hypothesis was a simple thing. The on­line discussion cast doubt on this implication: one respondent said “My claim is that... hypotheses are generated from data... and then refined/eliminated with further data,” while another said “you have to have some ideas worth testing,... to organize your research These are working hypotheses.” Finally,

one says that, “First, the ‘hypothesis'... is often not hypothetical at all, but only phenomenology dressed up as hypothesis; a hypothesis is a universal state­ment that cannot be directly verified by observation.” This was confusing. Is the experiment-stimulating function related to working hypotheses? How do working hypotheses compare to universal statements, are there other kinds of hypothesis, and how does phenomenology fit in?

The actual meaning of the word “hypothesis” was not the only sticking point.

Respondents were also anxious about what happens if one's “overarching hy­pothesis” was disproved. Was it true that the only thing that you'd have learned was that the hypothesis was wrong? One scientist acknowledged that “Knowing how a biological system doesn’t work is certainly useful, but most basic research study sections expect that a grant will tell us more about how biological sys­tems do work, regardless of the outcomes of the proposed experiments.” In con­trast, one commentator counsels students “that if you do propose a new ‘model' then... try to shoot [it] down... before someone else does.” Both writers were alluding to the concept offalsification promoted by the philosopher Karl Popper, which states that tests of a hypothesis can never prove its truth but can, in prin­ciple, prove its falsehood. The principle of falsification is widely recognized throughout science, and so it was disconcerting to see so much disagreement about its function and worth.

Several comments touched on philosophical matters that the hypothesis involves—the nature of scientific answers; of certainty of knowledge; the writings of Karl Popper, John Platt, and David Hume—that many of us scientists don't know very much about.

Many respondents took it for granted that the whole conception of hypothesis- driven research is ill-advised. Lorsch's premise was that “By putting questions ahead of hypotheses, getting the answer becomes the goal rather than ‘proving' a particular idea.... [putting] questions first and [including] multiple models to ex­plain our observations offers significant benefits.” A number of respondents liked this proposal; however, no one explained how making the switch would bring the clarity to scientific thinking that it was supposed to. The unstated assumption running through the comments was that non-hypothesis- related approaches are free from the drawbacks of the hypothesis. Lorsch's questions suggest that there are sharp divisions between asking questions and testing hypotheses or between models and hypotheses. But is this true? What is the evidence that science can answer questions more reliably than it can test hypotheses?

Furthermore, while the post focused on “question-driven science,” respondents also mentioned “discovery science,” “exploratory science,” “fishing expeditions,” “curiosity-driven science,” and “big science” as alternatives to the hypothesis-driven approach. Although these alternatives were not defined and the contrasts were not spelled out, the conclusion was clear: these alternatives and the hypothesis were mutually incompatible and a scientist would have to choose one or the other.

In short, the blog and its follow-up revealed that (1) the hypothesis is a gen­uinely hot topic and the scientific community has pervasive concerns about it, (2) there is no consensus on what the concerns are or how to tackle them,

(3) differences in interpretations of a hypothesis contribute to the confusion,

(4) much antagonism toward the hypothesis arises from nebulous worries re­garding bias or other problematical behaviors, and, finally, (5) to a large extent frustration with the hypothesis is prompted by anxieties about how others, par­ticularly reviewers of scientific grant applications, interpret it and is only tangen­tially about the hypothesis itself. To be sure, this was a self-selected, nonrandom sample of respondents, and we can't be sure how broadly representative their views are. Still, the variety and passionate tone of the responses made me suspect that there is a problem regarding what many people would have said is one of the bedrock tenets of science. The main message that I took away was that I needed to know more about the hypothesis—its divergent roles and attributes—and about the relationships between hypotheses, questions, exploratory science, etc.

I was beginning to get a good sense of why our graduate students had trouble with the hypothesis: the scientific community itself had trouble with it. A final piece of the puzzle was discovering that, in addition to facing neglect and mis­understanding, the hypothesis also faces active opposition. It turns out that scientists have written entire books that feature extensive attacks on the hypo­thesis. Surely, I thought, the hypothesis has served as the bulwark of scientific thinking for hundreds of years, hypothesis-based science has produced formerly unimaginable advances in knowledge—things can't be all that bad!

I.