07 April 2021


I’m grateful that:

  1. I had a book in me that I thought was worth writing.
  2. Pelagic Publishing gave me a chance to write it.
  3. I made it through a pandemic to see it in print.

I hope people find it helpful.

03 April 2021

Why “descriptive” science is downplayed

On Twitter, Rejji Kuruvilla asked:

I’m sorry, but WHY are descriptive studies a problem in grant or ms review? If you don’t provide the 1st description or visualization of a biological process, how do you provide the basis for hypothesis or mechanism-driven science?

Oh, I feel this. I complained about this since my grad school days. One of the biggest scientific endeavors that closed out the twentieth century, the human genome project, was pure description. I love this from Niko Tinbergen, in his most famous paper (1963):

Contempt for simple observation is a lethal trait in any science(.)

Here’s what I think is going on.

First, I suspect “descriptive” as a critique might mean any one of three things.

  1. Not hypothesis driven.
  2. Not investigated by controlled experiment.
  3. Studies only a single level of organization.

The bias against description is a symptom of the fact that basic biological research is has been supported by medical agencies. In the United States, the budget for the National Institutes of Health dwarfs that of the National Science Foundation. (Interestingly, this isn’t the case in Canada.)

Medical agencies aren’t funding science for science’s sake. They are not interested in making discoveries that broaden our understanding of the natural world. They have sick people they want to make better. They want treatments. They want results.

To their credit, most medical funding agencies recognize that investing in basic science pays dividends. That’s why they support it at all. But their priorities are not those of curiosity-driven science.

So it is no surprise that such agencies would strongly favour hypothesis-driven research. Because as much as I love basic description and serendipitous discoveries, I absolutely recognize that hypothesis-driven research –  particularly strong inference of pitting competing hypotheses against each other – is ferociously efficient at generating new knowledge.

I don’t think hypothesis-driven research is enough. But even I have to say that I don’t think any other approach generates knowledge as quickly or as consistently.

If you get the “too descriptive” critique, you can’t fix it just by working in the word “hypothesis” into your proposal at every opportunity. The “descriptive” critique is not necessarily about whether you have a hypothesis at all, but whether you address a hypothesis that is actively investigated by your research community. 

You can have a perfectly hypothesis driven project, but is the hypothesis doesn’t addresses what the community cares about, it will still get called “descriptive.”

Another aspect of the critique is that “descriptive” studies are contrasted against “mechanistic” studies. Again, I think this is a symptom of the “medicalization” of research funding.

This semester, I was unexpected assigned to teach part of a course in human pathophysiology. This is way outside my expertise, and I’ve been forced to learn about medical topics more than I ever have before in my life. And after a few months of digging into bone, muscle, and hormonal disorders, it’s surprised me how often developing treatments drill down to understanding molecular interactions. 

A description like “There’s too much hormone” is necessary. But treatments are often based in, “This drug blocks the receptor to this hormone” and “This drug blocks synthesis of this hormone.” In other words, the research spans multiple levels of analysis. When people talk about “mechanism,” they usually mean that they are looking for a level of analysis that is at more finely grained, by at least one step. 

If you are studying an organism, they want an explanation at the level of organs. 

If you’re studying organs, they want an explanation at the level of tissue.

If you’re studying cells, they want an explanation at the level of molecules.

(At least in biology, we usually stop there and don’t require explanations at the quantum level. Thank goodness.)

So seeing the challenges of the problems and the successes gained from these these molecular approaches helps me see why funding agencies like them. They have a proven track record.

I’ve also found that many students struggle to articulate hypotheses. I wonder if early career researchers writing their grants might also be struggling with this.


Tinbergen, N. 1963. On aims and methods in ethology. Ethology 20(4): 410-433.