Category Archives: Science

Science

Scientific Writing and E-Prime

E-Prime is a modification to the English language in which the verb “to be” is not allowed. I first learned of it via Catallaxy (disclaimer: I occasionally post on Catallaxy), where Cato links to this article.

I love the idea of E-Prime. A few attempts to speak or write in it convinced me that it makes it difficult to hide assumptions. E-Prime is therefore very appropriate for scientific writing, in which it is important to clarify what assumptions were made. Unfortunately, removal of the verb “to be” causes awkward and sometimes repetitive writing, particularly in “Methods” writing. It gets rid of the passive voice, oh-so-beloved pseudo-humble writing style of many a scientist.

Normal Methods

The model was paced eight times at a basic cycle length (BCL) of 300 ms. A shock was then applied with a field perpendicular to the direction of propagation over a range of coupling intervals (CIs) following the last paced beat. Once reentry was initiated, the model was allowed to run until action potential duration (APD) stabilized to a variation of less than 1 ms per beat.

E-Prime Methods

We paced the model eight times at a basic cycle length (BCL) of 300 ms. We then applied a shock with a field perpendicular to the direction of propagation over a range of coupling intervals (CIs) following the last paced beat. Once we initiated reentry, we allowed the model to run until action potential duration (APD) stabilized to a variaition of less than 1ms per beat.


Not too awkward there, but to be is replaced by we (verb)ed, resulting in gratuitous use of the Royal We. Zeus forbid we use the first person pronoun “I” in a scientific paper. Imagine several more paragraphs of this, though — it can get annoyingly repetitive quickly.

Perhaps it is more important to focus on the use of E-Prime in results. After all, one of the best examples I’ve seen of the utility of E-Prime is in describing elementary particles (as done in the linked article):

A proton appears to behave like a particle when observed by instrument 1. A proton appears to behave like a wave when observed by instrument 2.

vs.

A proton is a particle. A proton is wave.

Even for people who are well-versed in the mathematics and various experiments used to generate the statements above, one is clearer than the other. In fact, I would go so far as to say that the E-Prime version (first one) is a set of observations, while the “to be”-inclusive version (second) is a set of (contradictory) conclusions.

Now, assuming that I want to write my papers in E-Prime, how do I buck the system in which people are used to passive voice and still get published?

Disclaimer: I did not attempt to make this post E-Prime-compliant, though this disclaimer follows the rules of E-Prime.

What would have happened?

Simulations of real systems make it possible to test “what if” questions, and compare the results with what did happen. For example, I am working on a study in which a stimulus pulse turns on and off while a spiral wave rotates about an obstacle. We can ask, “What happened to the spiral wave following application of a stimulus pulse?” The answer depends on many factors. The image below shows the results of an experiment that tests this question:
095_08_pulse_only
It’s hard to say what happened to the wave. We see recovery, and then eventually activation. We might expect that the additional current supplied by the pulse (green, units not shown) would lead to an earlier activation. How can we say? What if we re-run the same simulation, but do not apply a stimulus pulse?
095_08_both
On inspection, pulse application postponed activation (in the red trace) by extending the duration of the wave and thus, the refractory period. Without the “what-if” data (black) for comparison, the effect of the stimulus pulse seemed less clear.

I give here only one example, but this concept applies to simulations in general. Deterministic simulations have the advantage of perfect repeatability, something much sought after but difficult to achieve in experimental studies.

Clinical Applicability

We do a lot of really cool and scientifically interesting stuff in both our lab and the field as a whole. We have now developed our technology to the point where we can take an MRI scan of a heart and develop a complete model, with accurate fiber directions from diffusion tensor data, in a matter of days if not hours. However, it’s easy to get caught up in the technology and the science and lose sight of clinical endpoints.

This past week, we had the good fortune to be visited by a research scientist from a device company. He pointed out that despite the great sophistication of our knowledge about various CEP minutiae, there remain basic problems with pacing and defibrillation therapies, and with our knowledge of cardiovascular disease in general.

And that brings up a very important point. At times, I’ve felt like we’re patching the symptoms instead of fixing the problem — by researching heart attacks that result primarily from cardiovascular disease, we’re redirecting effort that could be spent on eliminating cardiovascular disease in the first place. For example, it seems vegetarians may have around a 20% smaller risk of death from ischemic heart disease than non-vegetarians. Occasional meat eaters fall between meat-eaters and vegetarians. (Please note that I don’t think that one study is conclusive — I haven’t done a thorough literature search.) To keep things honest, this study suggests that in British vegetarians, differences in mortality may be due to other lifestyle factors.

Here are some other articles on the subject, with their conclusions:

After reading over these abstracts, my suspicions and some things I have heard have been confirmed: It’s really hard to get conclusive evidence out of human studies.

Even if everyone on the planet simultaneously switched to the optimum diet for minimizing heart disease, whatever that is, there would still be a need for treatment of ischemic heart disease in the form of defibrillator devices for the next several decades, as in many people the damage is already done. With that in mind, it is important that we as a research community focus our work on developing helpful therapies rather than keeping to our ivory towers.