C Think About Thinking

Cognitive ease⁶ is a state in which people are less careful, less critical, more likely to accept superficially reasonable statements and, therefore, more likely to fall for cognitive illusions or otherwise make mistakes.

14. C.1 Beware the Curse of Knowledge

In a Sense of Style,⁷ Steven Pinker defines the curse as “a difficulty in imagining what it is like for someone else not to know something you know.” Now think back on a time when you sat through a lecture in which almost every word, from the initial rationale of the talk to its conclusions, left you feeling baffled. It is likely that the talk was given by someone firmly in the clutches of the curse; someone who didn’t realize that, although you were an intelligent, interested listener, you did not have all of the background knowledge that he had. Had he been attuned to what you (and probably 80% of the audience) might not know, everyone, in­cluding the speaker himself, would have gotten much more out of the experience. I suspect that many scientific authors do not state their hypotheses explicitly be­cause they suffer from the curse of knowledge.

Obviously, you want to avoid the curse when you are trying to communicate, and this can be hard to do. Pinker points out that completely abolishing the curse would demand not just putting yourself in others’ shoes, though that should help, but putting yourself in others’ brains, which is impossible. He emphasizes that you can alleviate the effects of the curse by avoiding jargon and too many abbreviations, knowing your audience, and other tips.

While all this is good advice, it doesn’t explain why we are susceptible to the curse in the first place. We know full well that others don’t really know what we know—that’s why we’re trying to communicate with them! so why are we so ob­tuse about the need to get on the same wavelength with them?

Although it is not the primary issue for him, Pinker notes that we are una­ware of how abstractly we think about things that we know well. We pack large amounts of information into idiosyncratically organized chunks of knowledge and forget to unpack them for everyone else. In the same vein, we don’t explain the logic behind our reasoning. In Chapter 11, we found that our reasoning and our knowledge are often unconscious, hidden from ourselves as well as from others. Without special effort to find out, we often don't know what we know. We take for granted that we fully understand the details of those high-level abstractions in our brains, though we sometimes don't. In other words, we can be victims of our own curse of knowledge.

The first step in overcoming the curse, says Pinker, is to appreciate how “devilish” it is. Likewise, becoming aware of the curse of knowledge in science, making our thoughts and reasoning concrete and explicit, thinking about what we're doing in forming and testing hypotheses, is a first step toward improving scientific thinking.

As a complementary suggestion, I'd like to advocate for a writing style that may help reduce the abstract nature of scientific communications. Along with an increasing number of scientific journal editors, I recommend that we use the “active voice” in writing (see Box 14.1). That is, do not write, “the following experiment was done... ;” write “we did the following experi­ment... ” (and don't use “we” if you mean, “I”). If you take ownership of your ideas you will probably think about them and present them in more direct and easier-to-follow ways than when you're filtering them through the passive voice.

14. C.2 Take an Outside Point of View

Daniel Kahneman tells the story⁸ of a committee that he was once on that was charged with writing a new statistics textbook.

We need to take an outside view of our own thinking, trying to see it as an external observer would. This does not conflict with the advice to use the active voice when actually writing, of course. This is an extension of the meta-cognitive view that I mentioned earlier; we need to be thinking about our own thinking in order to find ways to make it better.

Box 14.1 The Passive Voice and the Hypothesis

“The data [evidence, results] suggest...,” “the data show...,” “the data imply....” If you took it literally, this way of talking would make no sense. “The data” are what we get from doing experiments—columns of numbers, graphs, images from high-powered microscopes, and so on; data just sit there, not suggesting, showing, or implying anything. We human beings look at the data; we interpret, infer from, think about them; and we finally communicate our thoughts about what they mean. When we assign “the data” the leading role, we're using what's called the passive voice in English. In general, when the entity that is carrying an action is not the subject of a sentence and may be left out of it entirely, the sentence is passive.

The passive voice has adherents.“Considered to be objective, impersonal, and well suited to science writing, the passive voice became the standard style for medical and scientific journal publications for decades” (http://www. biomedicaleditor.com/active-voice.html).

“In conclusion, the use of the passive voice encourages precision and pro­bity and when used correctly can generate as much passion and stimulation as the skilled use of the active voice. The active voice encourages careless­ness, partisanship, and as used by many of its adherents, does no favors to the English language or science” (http://www.sci.utah.edu/~macleod/writing/ passive-letters.html).

On the other hand, there are those who object strongly to the passive voice: “Most scientists use passive voice either out of habit or to make themselves seem scholarly, objective, or sophisticated. Scientists have not always written in passive voice. First-person pronouns such as I and we began to disappear from scientific writing in the United States in the 1920s when active voice was replaced by today's inflexible, impersonal and often boring style of scientific writing” (Randy Moore, Editor, The American Biology Teacher).

Indeed, some journals are trying to swing the pendulum back toward the active voice:

“Nature journals like authors to write in the active voice (‘we performed the experiment...') as experience has shown that readers find concepts and results to be conveyed more clearly if written directly” (https://www.nature. com/authors/author_resources/how_write.html).

“Choose the active voice more often than you choose the passive, for the passive voice usually requires more words and often obscures the agent of action.

The dispute about passive and active voice focuses mainly on science writing style, and style is not my concern. Rather, I worry about how language shapes our thinking. I suspect that a strong bias in favor of the passive voice makes us less inclined to be forthright about our hypotheses. Hypotheses do not spring up spontaneously; scientists invent, draw predictions from, and test them. Perhaps, in order to avoid the disembodied and faintly pompous passive construction, “The hypothesis suggested itself that...,” we excise the hypothesis from our writing. If there is no hypothesis, then we don't have to account for where one came from. In any case, by omitting the actor—the scientist—from the report, the passive voice obscures the reasoning process involved. In contrast, by adopting the active voice, “I hypothesize...” or words to that effect, we make the reasoning plain, and direct responsibility and give credit where it belongs: the scientists themselves.

14. C.3 Do a “Premortem”

Seventy-three seconds after its launch on January 28, 1986, the Space Shuttle Challenger blew up in mid-air, killing the seven crew members and vir­tually halting the US space program for more than 2 years. Following the tragedy, the Rogers Commission conducted a thorough review to determine what caused it and how it could have been prevented. An investigation into the causes of a failure, conducted after the failure has occurred, is called a postmortem.

Kahneman recommends undertaking a similar, thorough investigation of your plans before you act. He calls this a premortem.⁹ The idea is to assume that you already put a plan into effect and now, a year later, have learned that it failed. That's the starting point of the exercise: it definitely failed. You have work out what went wrong. Ideally, you enlist coworkers and ask them specifically to iden­tify the fault(s) in the plan.

You can apply the premortem technique to your scientific thinking as well. Imagine that you've written a paper or a grant application that went through a re­view process and was rejected with the comments that it is “hard to follow,” “dif­fuse,” “not well-organized,” etc. Ask yourself—or better, ask your colleagues—to read it and tell you why it was rejected. Take their advice seriously.

14. C.4 What Can You Explain Now that You Didn't Set Out to Explain?

In his biography, Genius: Richard Feynman and Modern Physics, James Gleick reports¹⁰ that Feynman would often ask a young theorist: “What can you explain now that you didn’t set out to explain?” Good ideas have considerable “reach” in David Deutsch’s terms (Chapter 10.C); they go well beyond the circumstance in which they first arose. Einstein’s General Theory of Relativity got a big boost in acceptability when astronomers realized that it could account for a slight anomaly in Mercury’s orbit that had troubled them. Einstein had been unaware that his theory had such reach and was pleased to learn that it did.

On a far more humble scale, I mentioned (Chapter 2) an experiment that a colleague and I did where we found that when we stimulated a principal (P) cell, incoming signals from inhibitory (I) cells transiently disappeared because a mys­terious signal from the P cell briefly silenced the I cell. Later, we found that there were two separate groups of I cells: one whose signals disappeared, one whose signals didn’t. Why the difference? We had no idea. Meanwhile, another group of investigators discovered that cannabinoid receptors were only present on one type of I cell.¹¹ This was an isolated finding, what looked like a naked fact, until the mysterious signal that the P cell was sending back to the I cells was discov­ered to be an endocannabinoid.

Suddenly, everything fell into place: our temporarily disappearing I sig­nals must be coming from the I cells bearing cannabinoid receptors. The endocannabinoid hypothesis, which began as an explanation of how P cells could regulate the output of I cells, now suggested that there was a defined network of I cells that would be uniquely affected by cannabinoids because only they had the cannabinoid receptors. This, in turn, could mean that brain activity affected by cannabinoid molecules—the ones in marijuana in addition to the natural ones— would be disproportionately regulated by these particular cells. Researchers now suspect that the endocannabinoid hypothesis reaches far beyond its origins to brain activity involved in highly complex human behaviors, everything from the development of play behavior in young males¹² (rats, anyway) to epileptic seizures,¹³ as well as the fear responding that we just reviewed.

The overarching message is that you can hardly answer Feynman’s question— “What can you explain that you didn’t set out to explain?”—without thinking a lot about your data and your hypothesis.

14.