J Implicit Hypotheses
“No one searches without a plan,” said Santiago Ramon y Cajal,51 and you can't have a plan without a hypothesis about where and how you should search and what you'll encounter.
A related concept, voiced by Popper and other philosophers, is that all observations are “theory-laden.”52 You can't make a scientific observation without having a theoretical framework for it. The Critical Rationalist philosopher Bryan Magee gets this point across to beginning students by instructing them to “Observe” and record their observations.53 The students, naturally, are baffled—what should they observe; what should they record? The world has much too much information to observe it all at once. “Observation is always selective... it needs an object.” You have to break down the world into manageable categories—birds, colors, temperatures—and for that you need hypotheses: what constitutes a bird, a color, a temperature? And you need to know how to observe, what instruments you need, and what hypotheses the instruments depend on, etc. Observations, he concludes, “are always interpretations of the facts observed; that they are interpretations in the light of [hypotheses].” But the hypotheses that shape the observations are rarely laid out explicitly; they are implicit.At first the notion of implicit hypotheses seems paradoxical: hypothesis formation and testing demand focused, conscious thought, so how could a hypothesis not be explicit? Implicit hypotheses exist in two forms: as unstated experimental frameworks and as deep assumptions (deep implicit hypotheses).
2.J.1 Implicit Experimental Hypotheses
Authors of scientific papers frequently do not state their hypothesis (Chapter 9) even when they have one. Teasing out an implicit experimental hypothesis buried within a paper requires careful reading. A common modern tactic is to note in the Abstract or Introduction to a report that a certain phenomenon is “poorly understood.” (I regretfully confess to having used this formula.) Saying something is poorly understood is almost entirely vacuous.
The fundamentally uncertain state of all scientific knowledge guarantees that all phenomenon have poorly understood features. In making the remark, the experimenters are usually just announcing the topic of the paper and, perhaps, hinting at their implicit hypothesis and its predictions. The authors ordinarily go on to lay out their experiments logically and give a rationale for each critical test. Here and there, they may sprinkle in “predictions,” even if the predictions aren't part of a coherent argument. The Discussion section integrates the major findings into a coherent whole, puts them into a larger context, and occasionally proposes a summary hypothesis (or model) that suggests what the experimental hypothesis had been all along. With some effort you can find the implicit hypothesis, and, indeed, trying to find them can be an educational exercise (Chapter 14).We could, for example, deduce our mechanic’s implicit hypothesis when he was trying to diagnose why the car wouldn’t start. By reading his report afterward; (“Because the reason for the car’s not starting was poorly understood...”), he pressed the Start button, heard no sound, and tried the lights and the radio; then, when still nothing happened, he opened the hood and checked the battery terminals, cables, etc. Without being told what his reasoning was, we could figure it out, assuming, of course, that we already knew how cars worked. Otherwise, we might be puzzled about why he tried the lights and radio (did he really think that they had anything to do with starting the engine?). Reading a report based on an implicit hypothesis requires extra effort and presumes that the reader can see that the conclusions do follow from the results. While such a report is easier to understand than one without any apparent plan, it is generally harder to follow than one with an explicit hypothesis, especially for readers who are not experts.
2.J.2 Deep Implicit Hypotheses
Implicit hypotheses are also part of the background knowledge that we bring to any project.
Since all scientific knowledge is ultimately uncertain, a “fact” is essentially a hypothesis—“A fact is where the investigation rests.” As with other tested-and-not-falsified hypotheses, we assume that they’re true, but, unlike implicit experimental hypotheses, we are not currently investigating them. They are deep implicit hypotheses.When neuroscientists study the brain, one such deep implicit hypothesis was based on decades of accumulated information about what cells are and what they do. The brain processes information at chemical synapses between cells, as we discussed earlier. There are two fundamentally different kinds of cells in the brain, neurons and glial cells (aka glia). For nearly 100 years neuroscientists interested in information processing focused on neurons, the highly electrically excitable signaling cells that control everything from muscle contractions to visual perception. Although there are about as many glia as neurons, the glia were mostly ignored; their electrical excitability was too sluggish to keep up with the rapid signaling capabilities of neurons. The deep implicit hypothesis was that glia were the go-fers and support staff of the brain, cleaning up the local microenvironment and guiding the slow process of brain development, but having no role in the glamourous job of information processing. Ignoring glia allowed neuroscientists to make enormous strides in understanding neuronal communication.
Still, the assumption that neurons alone carry out information processing was an implicit hypothesis, and hypotheses, even deep implicit ones, exist to be tested. When the neurons-only hypothesis was made explicit and tested, it failed miserably. It turns out that if you disrupt glial cells so they can't function, you prevent neuronal synapses from operating normally. In fact, in recent years, we've learned of so many ways in which glia influence neuronal signaling that some neuroscientists believe that the key communication structure in the brain is the “tripartite synapse,”54 a microstructure made up of pre- and postsynaptic neurons together with their nearby glial partners.
You'll often hear that you've got to have a completely open mind to do a scientific investigation. It is an admirable objective that you won't be able to achieve. We can't escape the constraints of our minds, which are constantly at work constructing our understanding of the world by spinning webs of deep implicit hypotheses. A tell-tale sign of this activity comes in the form of a surprise or an unexpected result. “Expectation” is another name for a prediction made by a deep implicit hypothesis, a bit of background knowledge that you took for granted. Surprise is what happens when a prediction of a deep implicit hypothesis is falsified.
Besides forming background knowledge, the deep implicit hypothesis also shows up in the choice of the research technique. “To a man with a hammer, everything looks like a nail” fits when it comes to scientific experiments; your choice of measuring instruments shows which measurements you think are worth making. Microscopists believe that secrets lie in cellular structure, while electrophysiologists measure electrical potentials because they believe that cellular activity is the key feature of biological function. Science takes place within a web of deep implicit hypotheses.
Now that we've surveyed some properties of hypotheses, in the next chapter I'll return to the main one, falsification, and address a few of the questions or objections that philosophers have raised and that you might have about it. Popper's falsification program was novel and completely at odds with the philosophical mainstream of his day, so it is not surprising that it ran into a lot of opposition.
2.