Wednesday, November 21, 2012

Elisabeth Lloyd: Gould and adaptation: San Marco 33 years later

This is a talk by Elizabeth Lloyd from a conference in Italy last May on "Stephen J. Gould’s Legacy: Nature, History, Society." Thanks to Ryan Gregory for posting all the videos [Stephen Jay Gould conference in Italy — full series of talks].

Lloyd explains the problems with the adaptationist approach to the logic of research questions using a case study on the evolutionary origins of female orgasm. She points out that Gould & Lewontin were right 33 years ago when they called attention to the adaptationist bias and that things haven't changed very much.

This is an excellent example of a philosopher of biology, Elizabeth Lloyd, making a substantial contribution to science and the philosophy of science



20 comments :

  1. "...case study on the evolutionary origins of female organism"

    Well, I'm not a female organism, but I suspect that the study is a bit more specific than the entire organism ...

    ReplyDelete
  2. Unbelievably bad talk. Every faucet of it. Any grad student would be quartered over such poor performance, visuals and logic. How can anyone take seriously a talk that shows a graph with two peaks being commented as "as you can see, there are no peaks"?

    ReplyDelete
    Replies
    1. Please tell me you are kidding, because if not, you are so fucking ridiculous. Apparently you never learned how to comprehend what graphs are actually displaying.

      Delete
    2. I'd agree the talk was horrific - I'd be livid if one of my students gave a talk like that. But I also have to agree with anonymous - the 'peaks' in her orgasm graph (never thought I'd combine those two words) are minor variations. The graph was crap - error bars are not just for fun - but I highly doubt those peaks represent anything significant. Statistically, its a flat line.

      Her take home message cannot be repeated enough - HOW we ask questions can greatly influence what results we get and how we interpret those results. An important lesson, both in evolutionary biology & in science in general.

      But even a good message doesn't excuse such a poor talk.

      Delete
    3. Anonymous said:
      "Apparently you never learned how to comprehend what graphs are actually displaying

      Oh dear... For something like female orgasm, the N is probably HUGE. If so, chances are very high that the points on that graph are statistically significantly different from each other.

      Of course, if the presentation was actually competent, the graph would have had error bars on it and we wouldn't be having this discussion. Personally, I can think of nothing wrong with the actual distribution being bimodal.

      Larry:
      Google/Blogspot keep eating my posts randomly. I would see a message posted then most of the time the post disappears and later one some fraction of them reappears again and another gets lost forever. Really strange that there is some kind of randomness to this bug.

      Delete
    4. Anonymous said:
      "Apparently you never learned how to comprehend what graphs are actually displaying"

      Sigh... Where to begin? Since the graph showed data on female orgasm, it is safe to assume that N is very large. Which means that it's very likely that the differences between the points are highly statistically significant. Got it?

      Of course, had the talk been done competently, the graph in question would have had error bars on it and we wouldn't be having this discussion to begin with.

      Now go and see Table 2 here:
      http://rsbl.royalsocietypublishing.org/content/1/3/260.full
      Clear bimodal distribution, not unlike what that graph shows. And since the N > 5000, it is a safe bet that it is significant at very low p level.

      Delete
    5. Just wondering why you thought the talk was so horrific.

      Delete
    6. DK says,

      Google/Blogspot keep eating my posts randomly. I would see a message posted then most of the time the post disappears and later one some fraction of them reappears again and another gets lost forever. Really strange that there is some kind of randomness to this bug.

      Your comments sometimes get sent to my spam box and I have to "rescue" them when I check for errors. Nobody else posts comments that get labelled as spam. I don't know why yours are different.

      Sometimes it's because you insert a link but not all comments with links are treated as spam.


      Delete
    7. Yes I'd agree plot looks somewhat strange; the original paper:
      https://genepi.qimr.edu.au/contents/p/staff/CV402.pdf
      seems to have seven categories that could be plot on that graph; whereas she also puts an additional point (which I guess is the "do not do" category, i.e. to be excluded from analysis). The next point is how to characterise the difference between "Never" and "Rarely (<20%)" from a rigorous point of view--the same distinction is made at the other end. About the only thing I can think of is to combine the two into a single category. Otherwise you end up with different sized bins and everything gets confusing. Once you do that you get a nice bimodal distribution, like:

      http://i49.tinypic.com/28rct44.jpg

      I enjoyed the rest of the talk though!

      Delete
  3. If that's an "excellent example", I'd hate to see a really bad one!

    [Adaptation does happen. Get over it.]

    ReplyDelete
    Replies
    1. You'll be along shortly with the list of people who deny that adaptation happens of course. No?

      Delete
  4. LLoyd's talk is so confusing because she does not distinguish between biological role and biological/selective advantage of a trait.

    ReplyDelete
  5. I didn't find it confusing at all -- it was very clear and making a most excellent point.

    But I'm biased -- I'm a coauthor on her rebuttal paper in Animal Behaviour that she mentions in her talk.

    ReplyDelete
    Replies
    1. I think you missed the point. :-)

      The talk was supposed to be confusing to adaptationists because it challenges them to think critically. They are psychologically primed to find some way to criticize her talk because otherwise they might have to concede that she (and Gould & Lewontin) have a point.

      Delete
    2. Or maybe not. I totally agree that the by-product idea makes sense. I don't consider it proven but I think chances that it's correct are higher. Given the facts, adaptationists' best option is to claim balancing selection. Had Lloyd addressed that possibility at all?

      None of it changes the fact that the talk is completely dreadful.

      Delete
    3. @Larry:
      Or maybe not. I totally agree that the by-product idea makes sense. I don't consider it proven but I think chances that it's correct are higher. Given the facts, adaptationists' best option is to claim balancing selection. Had Lloyd addressed that possibility at all?

      None of it changes the fact that the talk is completely dreadful.

      Delete
    4. What's so "dreadful" about it? Seems reasonable to me.

      Delete
    5. Like DK, I think the byproduct explanation is fairly likely to be correct. I would point out one flaw in her argument (and presumably in the rebuttal paper, with apologies to PZ Myers) - since she (correctly) takes the twin study to task for making an invalid comparison, I must do the same for her comparison of the one-peak result of directional selection with the flat-ish graph of female orgasmic success.

      It is inappropriate to compare the maximum speed of a wolf (its potential) with the achieved orgasmic success of women - a closer comparison would be with the average speed of a wolf (as in, observe a wolf at various (waking) time points, and chart out its speed). This would give a much more spread out graph for the wolf - indeed it would probably be rather bimodal, with the highest peak at zero, and another one near its top speed.

      If one must compare maximum speed, the valid comparison would be potential to achieve orgasm in women, assuming optimum conditions. A woman who rarely achieves orgasm may be fully capable of achieving orgasm all the time, but happens to have a partner who doesn't let her achieve that potential, or has other hindrances to achieving this potential.

      Hmm, though as I think about this, "optimal conditions" may be too high a bar, not required by an adaptationist explanation... but I would still suggest that using a broad sampling of women's orgasmic achievement without allowing that some of those women might be capable, in different circumstances, of achieving much higher success, requires explicit justification.

      Delete
  6. I like her by-product hypothesis, and I think it explains better female orgasm than adaptationist hypothesis for which there is not much evidence. She has a point, and a good one, no doubt. That does not mean that she is a good speaker and that making the distinction I mentioned would have made the talk much clearer. That something has a biological role in an organism does not mean that it arose by natural selection for functional reasons.

    ReplyDelete