Thursday, March 28, 2013

Just like pornography!

In a famous obscenity case, an infamous Supreme Court justice Potter Stewart said he couldn't define 'pornography' but he knew it when he saw it (that was before the internet, so he actually had to do some work to get his, um,  exposure, so to speak).

There is something similar in relation to scientific explanations that really have transformative power:  when it happens, you may or may not be able to explain it in its details, but you recognize it.

Goya, The Nude Maja
Every day in evolutionary and biomedical genetics, we see a stream, indeed, a flood of reviews, overviews, commentaries, papers, blog posts, and op-eds promising progress on understanding biological complexity.  Papers with sentences like
"Here we develop a model that shows how considering [sequence, systems biology, epigenetics, copy number variation, evolution, new functional equations, neuroimaging, high throughput analysis, new 'omics' data, methylation, acetylation, ..... (pick your favorite)], major advances in understanding the biology of complex traits and diseases.  Our method ....."
But, then, where is all this promised progress?  It might not exactly be obscene, but are such bevies of claims just posturing and careerism, what we are forced to do to succeed in academic careers these days?  It may be apt way to say so, because what we really see these days is incremental change, some of it progress but most of it trivial or useless, yet fed by the constant pressure for more Large Scale high-powered computational this or that. And that leads to all the self-congratulation.  But it's as paradigm shifting as Goya's Maja is pornography. 

If you think about the major advances in science that by most counts really were progress, the so-called revolutions or real (rather than self-flattering) paradigm shifts that have happened in science, from Galileo, to Darwin/Wallace, Einstein, the discovery of the nature of DNA sequence, or continental drift, these changes were very similar:
1.  Many diverse things that had been given separate, forced, or hand-waving explanations fell dramatically and quickly into place 
2. This was almost instantly recognized 
3.  The new ideas were conceptually very simple
The new theory may have involved some technical details, like fancy math or biochemistry and the like, but the ideas themselves were over-arching, synthesizing, and simplifying.

As Thomas Huxley famously proclaimed after learning Darwin's explanation of the mechanism for evolution:  "How extremely stupid not to have thought of that!"  

Once you see it, you realize it's import.  In genetics and biomedicine today, people are always saying it....but we're not yet seeing it.

3 comments:

Kenny A. Chaffin said...

Just want to say I agree.

S.H. said...

A wonderfully precise analysis of scientific innovation, but it's also sad.
The three properties you mention that are characteristic of break-through innovations (diverse ideas falling into place,instant recognition, simplicity) is also the chief reason for why "revolutionary" novelties (perhaps, at a more modest level than in your examples) face insurmountable hurdles in grant applications or manuscripts: they are so plausible and intuitive that they trigger what psychologists call the "We-knew-it-all-along" phenomenon (which has also been studied for reaction to scientific breakthroughs*). A symptom indicating that the reviewer has succumbed to such "hind-sight bias"** is the emphasis of his/her critique on apparent lack of novelty while he/she fails to cite prior art and finds no technical weaknesses in a proposal or paper but keeps reiterating the subjective notion that it's not novel.

*P. Slovic and B. Fischhoff. (1977). On the psychology of experimental surprises. J Exp Psychol Hum Perception Performance 3, 544-551

**N. J. Roese and K. D. Vohs. (2012). Hindsight Bias Perspect Psychol Sci 7, 411-426.

Ken Weiss said...

Thanks for this fine and thought-provoking comment! In evolution and relativity and plate tectonics, the WKIAA reaction was I think one that (in grantsmanship terms) would lead to testing the new idea or making predictions from it, or showing how extant data fit the new insight.

So it would not lead to 'do not fund' decisions. I think the kinds of instances you are referring to are somewhat different. I'll try to explain, since I'm not sure about my own thinking here:

Einstein's general relativity led to the famous Eddington check on the prediction of the bending of light. Plate tectonics led to the discovery of things like ocean trenches and use of biogeography and geology to reconstruct Gondwanaland, etc.

In genetics and evolution, we had a theory for polygenic control that made Mendel's and Darwin's ideas consistent. Complex traits were caused by multiple contributing genes.

So why would one bother with gene mapping? The informally stated but gene-centered rationalization was that we need to find 'the' genes, or that we didn't really know traits were polygenic, etc.

So, without really so-stating things, GWAS was started. Some hits, like BRCA, fueled an it-must-be-simple fire (and funding opportunity). Without criticizing the legitimate enthusiasm of that finding, one can say that after a number of GWAS had shown the truly complex nature of traits, we could and perhaps should have stopped those extensive and expensive studies and re-thought the question about the role of genes.

Instead, there was a scaling-up (and there still is).

This is really just what Kuhn would call 'normal science', trying to fit a square (it's simple) peg into a round (it's actually complex) hole. We hand-wave about 'complexity', but we're not really facing up to the problem we face.

In that sense, for some reason, the WKIAA reactions to GWAS requests are at best mixed: huge studies are still being launched. Rather than say WKIAA, the claim is that more and bigger will get us to the real truth, but we don't know what that is til these huge studies have been done.

To the extent this accurately portrays the current fundscape, it is the opposite of WKIAA. We should be saying, if GWAS, say, were truly transformative: make some unusual predictions, like Eddington did for physics, that follow from the purported insights being written about.

Generally, I think that while we are clearly learning things, and can predict implications, both sides of this are, at present, incremental rather than insightful.

In this sense, there's no kind of paradigm shift going on these days. The claims of just such transformative insight--the topic of our post--are the result of our need to keep our labs in business despite no truly major new ideas.

What, or even whether, truly new ideas are called for is a separate question.