Bayesian analysis can move in either direction in time. If I know that a new allele confers resistance to streptomycin, for example, I can predict that it will provide greater fitness to a bacteria population in my body when I’m taking antibiotics. If a new allele reduces the leg-to-body ratio in male humans, I can predict that it will provide greater fitness to the lucky Valentinos who have it.
That said, any predictions would be probabilistic rather than binary. The overall fitness of a trait in a population in a particular environment is highly complex. And as our friend Steve stated earlier,
Not the entire difference, as Hayek pointed out in his Nobel lecture. The complexity of physical systems nevertheless remains tractable to science because the variables are geberally independent, and most can be safely ignored in an experimental or modelling situation.
However, the complexity of human systems (he was an economist) is not so because the complexity is what he called “organised”, in that all the variables affect all the others. In that case, unless one knows all the individual factors completely (impossible), one can intrinsically only make generalised approximations, rather than predictions: in other words, it’s not “more difficult”, but “qualitatively dissimilar”.
The problem is, one cannot know for sure just how approximate ones conclusions are: you don’t know what you don’t know until you know it, and so even Bayesian procedures don’t protect one from being wildly wrong.
Biology, of course, lacking willful humans, is somewhere in between physical and humans science, but is closer in its complexity to the human than the physical science situation. That’s been my point all along.
This example appears another case where direct comparison with human history in the form of war makes the point I started with. Imagine in a war that some boffin stupidly suggests painting some of their warplanes white, and some stupid politician agrees to a study to check if it will save money on camouflage. The outcome, if not a foregone conclusion, is easily ascertained by the losses that would occur to the white planes. Science rules, OK.
But another boffin questions - more usefully - whether, say, USAF or RAF decals detract from camouflage and should be removed. If they attempt a trial, the multiple confounding factors of minor advantage, losses from friendly fire and especially the fact that even effective camouflage has a marginal role in overall success make the trial inconclusive in principle. In fact, the effect of removing decals would be a black box because of the nature of the real world: the fitness of decals is outside science… or if not that example, there is a level of complexity beyond which science cannot go.
The best that could be modelled is, perhaps, some experiment in which pilots’ ability to spot decaled and undecaled planes in a photograph is tested - but the model is so far from reality that it cannot help the war effort.
In other words, the slam dunk cases for selection like lactase persistence may demonstrate the existence of a mechanism, and may in certain cases be of practical value. But as soon as one knows that multiple causes are in operation - many of them entirely contingent - other competing traits, neutral retention of traits, ecology, animal behaviour, forest fires, cold winters, predation of young etc etc, there will come a point where all models give increasingly unreliable answers.
If that point is unrecognised, there is a danger of treating a black box as if it were science, rather than the stuff of theology, philosophy - or just imagination.
Jon, I fully agree with everything you wrote except for the last sentence quoted above. I think its a matter of degree, and that it makes no sense to postulate boxes called science (where predictability is the rule), other non scientific boxes like theology, philosophy and imagination. So there are highly complex systems in scientific models, (and yes, I mean the models themselves, not just the reality) which are difficult to allow predictions. I gave some examples in a previous comment. A great deal of biology, but not all of it, fits into this category, as does the vast majority of humanities like history, literature, art. But not all of it. As a musician, you know that the beauty of music can derive partly from contingency but also from some fairly rigourous rules of harmony, dissonance, melodic scales, and so on. I remember learning the laws of harmony (as practiced by JS Bach) as if they were laws of physics (parallel fifths are forbidden comes to mind).
Even your warfare model is a mix of contingency (just think of the weather issue during the Battle of the Ardennes) and some pretty useful and predictive modeling. I would cite Tolstoy’s long treatment of the laws of war in War and Peace, the curriculum of military academies that include a great deal of theoretical principles based on empirical observations (these can get extremely technical, like positioning of naval forces, and use of artillery). Yes, there are always exceptions. Even Tolstoy’s dictum that larger armies defeat smaller ones does not always hold. But the same is true in all of biology, and to a lesser extent in chemistry as well.
My point is mostly semantic. We agree on the nature of science and other fields of study, and we agree on the role of contingency and complexity. My point is simply that its a mistake to label highly contingent, complex, non predictable phenomena (or the models we construct to study them) not “scientific” a priori. In fact treating black boxes as they were science is often the first step to a better understanding of the natural word.
Eddie asked a while ago when the propensity definition of fitness came into play. I ran into this pretty clear statement of that conceptual approach from Thomas Hunt Morgan in 1916:
[quote]
If through a mutation a character appears that is neither advantageous nor disadvantageous, but indifferent, the chance that it may become established in the race is extremely small, although by good luck such a thing may occur rarely. It makes no difference whether the character in question is a dominant or a recessive one, the chance of its becoming established is exactly the same. If through a mutation a character appears that has an injurious effect, however slight this may be, it has practically no chance of becoming established. If through a mutation a character appears that has a beneficial influence on the individual, the chance that the individual will survive is increased, not only for itself, but for all of its descendants that come to inherit this character. [/quote]
What’s notable is that Morgan is considered a mutationist, rather than a Darwinian, and this understanding of the probabilistic nature of fitness was largely lost in the Modern Synthesis.
I do find it a little odd that Eddie has been defending the Darwinian and Neo-Darwinian understanding of fitness against its critics, given his usual embrace of those critics.
“What work does the word ‘advantageous’ do that isn’t already covered by discussing ‘fit,’ ‘unfit,’ and ‘neutral’?”
I don’t see the utility of quibbling over roughly synonymous terminology in someone else’s discipline. But if I’m blind to a crucial difference between “fitter” and “more advantageous,” do carry on.
The above article is a critique Dawkins view of The Selfish Gene and The Extended Phenotype. I agree with his criticism of Dawkins, but not his position of how evolution takes place, which is through Natural Selection based on ecology, not genes.
Shortly after I began to post on BioLogos I brought up the concept of Niche Construction theory which I researched for my book.
I was informed by others on the web site that Richard Dawkins and friends had refuted this view. after some digging on the web I found that some supporters of Dawkins in Europe called a conference to denounce Niche Construction as counter to Dawkins’ view of the Extended Phenotype, which they did.
Dawkins magnanimously said that it was not wrong, but might be considered as an aspect of the Extended Phenotype, which explains how genes control behavior. .
I think that I sent you a copy of my book and if so look on pp. 51-52 of Darwin’s Myth.
The only difference is that “fitter” is the word with a technical definition in the models we use. (Actually, in my neck of the woods at least, we usually use the selection coefficient, s, rather than the fitness (fitness = 1 + s)).
Many commentators on this forum have rued the relative lack of mathematics in biological theories, compared to the theories of other sciences. My feeling is that if biologists can define fitness mathematically in the context of a math-based model, I’m all for it. In fact, I’d readily give up 100 photos of long-necked giraffes and whale fins and whatnot for one good set of equations. And it seems that biologists agree with that approach.
Sure there is: “however slight this may be”. It’s a crude quantification, but it’s still quantification. In any case, so what? The definition of fitness changed by including probability, not by being quantified. Darwinians had no problem with quantifying fitness.
Right. It’s almost as if the particular word doesn’t matter at all.
It’s not clear to me that any communication is occurring in this thread. Let me start again from the beginning.
Darwin had the notion that some traits made an organism more suited to its environment – that they made it more fitted for it – and that this suitability would make it more successful reproductively. White bunnies really are more suitable for life on the tundra than brown bunnies. That notion has not changed. Morgan had the same notion, and so do I. That’s why we continue to use Darwin’s term today.
What’s been subject to change is the precise definition of fitness. As long as you’re thinking about large populations, and (more importantly) that you’re thinking about variation as a cloud of infinitesimal changes always accessible to the population, then it’s fine to collapse the two parts of the notion and just define “fittest” as “having the most offspring”, because natural selection is always operating. That worked for Darwin and for the architects of the Modern Synthesis, because that’s how they were thinking of variation.
Once you start dealing with small populations, though, and the reality that variation occurs as discrete mutations and that some of these will be absent or rare, then you have a problem. You have to deal with the fact that mutation, survival and reproduction are all stochastic processes. This means that the definition of fitness no longer matches the core notion, because sometimes the more suitable organism doesn’t survive. Sometimes the white bunny gets eaten by the tundra cats sooner than the brown bunny. You can keep the definition – whatever survives is more fit – if you insist, but you’ll lose the core idea of natural selection and adaptation. Brown bunnies really aren’t better adapted to life on the tundra. If you do that, you’ve now got a concept of fitness that has no obvious utility for anything.
Alternatively, you can recognize that the original notion didn’t depend on the precise definition, and is in fact captured better by a slight tweak to that definition: what is best suited to the environment is the type that tends to leave the most offspring. That way fitness is still directly connected to Darwin’s insight about natural selection and adaptation, we have a definition we can use in mathematical models, we’ve extended the Modern Synthesis, and everyone is happy. Everyone except you, that is, for reasons that I do not at all understand.
It’s precisely because traits don’t map one-to-one to alleles that you have to consider the fitness of alleles, at least if you’re interested in the fate of individual mutations. A “trait” is an often arbitrary way to describe an aspect of an extremely complex organism, and different traits can overlap or be highly correlated; it may not be possible to have one trait without also having another. An allele, though, is a real thing, a real state of the individual’s DNA. The fitness of the allele represents the entire effect on reproductive fitness of any and all phenotypic changes that the allele causes.
Now, if you happen to be more interested at the moment in a trait than in molecular evolution, it’s also fine to talk about the fitness of the trait, however many alleles may have contributed to it. But sometimes all you know is that an allele has increased in frequency, without having a clue what traits are involved.
Either way, the fitness describes the expected reproductive success of one class of individuals compared to that of another.
I completely agree, at least for species like humans. If you look at my quoted words in context, you will see that I was explaining why Darwin and the modern synthesists didn’t consider stochastic effects in NS. To state it more precisely, in the situations they were thinking about, the actual reproductive success is never significantly different than the expected reproductive success, and thus they did not need to distinguish between the two. I was not suggesting that there is no difference between the two, or that the situations they were thinking about covered all real-world populations.
No, I admitted nothing of the sort. See my last paragraph. I wasn’t telling you how I think about fitness; I was explaining why Darwin and others didn’t fully work out the concept of fitness.
Quite true – provided there are enough of both color bunnies in the Arctic. In practice, we’re usually interested in traits or alleles that have proved to be more fit because they succeeded. That doesn’t mean we should be satisfied with a muddled concept of fitness, though. A muddled concept won’t help us when we are considering cases where expected reproductive success is very different from actual success. [quote=“Eddie, post:195, topic:5784”]
It’s only when we ask why white rabbits are more fit that theoretical reasoning about various traits and their advantages becomes important. But to me, it’s the latter questions that are far more interesting.
[/quote]
They’re more interesting to lots of other people, too, including lots of biologists. That’s why one research program is to identify beneficial alleles and then figure out what traits they affect. That’s actually quite a hard problem, but there have been successes. Regardless, the fact that there are other questions to answer doesn’t mean we should think sloppily about fitness.
Great. Here is the portal for the complete 1000 Genomes data. Tell me which alleles out-reproduced others.
They’re more fit in the technical sense of the “fit” as well. Does this mean you’ve decided that the technical sense of “fit” isn’t so far from the common-sense meaning after all? (Also, if it’s so obvious as to be hardly mentioning, why did my mentioning it trigger this entire lengthy exchange?)