In  my question … hp?id=2358 , Haldane's rabbit was mentioned. Having looked further into that reference, I can see this chap, Karl Popper, followed similar questions on whether 'evolution' is scientific or not in its nature.

I am, actually, now more confused that ever over what "the theory of evolution" is actually good for!

As an illustration, please may I ask the following question: Is there a prediction that the "theory" of evolution can make that can be observed to happen in the future, whilst observing an environment of some stated conditions, such that the prediction makes a definitive statement about how something will end up or will occur?

In "fact vs theory" ( … hp?id=1172) it was said that the "theory" of evolution stood alongside other theories such as gravity, relativity and atomic. But this seems to be false because these all make a clear statement how given objects will behave under given conditions that we can see directly (or observe by scattering, in atomic theory) and be clearly measured, and it says they will happen tomorrow just as they will happen in a million year's time.

So, at the risk of pre-empting a few possible answers, if I were to observe a species extant in some environment, along with an assortment of other flora and fauna, and those environmental conditions went through some change, then is the species I am watching predicted to genetically change? I suspect there are exceptions to this already well known (e.g. primitive animals that have been the same for many millions of years that have survived extremes of conditions) but I also expect there are some that follow this. So the only prediction evolution seems to be able to make is "species may change under changing conditions, or they may survive if they are fit enough".

This cannot be an answer to my question, because it would be predicting "A OR NOT(A)", which is always true and is a tautology.

If the answer is the prediction that "evolution predicts only those fit enough will survive, unchanged" then it would require some ability to state whether a species is "fit enough" today. But it seems to me that the "theory" of evolution says that "those that survive are the fit ones". This is therefore a circular argument. The "theory" of evolution cannot state that "the fittest survive" AND "those that survive are the fittest". At some point, to be scientifically provable, "fitness" must be an independently measurable quality independent of whether it survives or not.

My question then is; without tautologies or self-circular definitions of "fitness", what "measurable" prediction can the theory of evolution make?

best regards,

Chris MB.

I'm no philosopher, but by my understanding evolution by natural selection is a tautology. This stems from the way we define of selection and evolution. There is, as far as I can see, no escape from this mathematically but this probably isn't the venue to discuss that - I would really recomend reading some texts about selection that are written by evolutionary biologists  (e.g., Graham Bell's book). Your question above implies a number of misunderstandings about how selection drives change. For example, setting aside the levels of selection arguments (which I DO actually think are important) we do not generally believe that selection among species plays that big a role. Most selection is about fitness differences between individuals (or even genes if you take a Dawkins type view of things).

In terms of predictive power - evolutionary theory is routinely used to make both qualitative and quantitative predictions. Our predictions may relate to how fast and in what direction a trait (e.g. body size in African elephants) will change, or to what an optimum trait value should be (e.g., just how many eggs should a cuckoo lay?). We also predict how gene frequencies will change in a population over time, how parasites will respond evolutionarily to medicines used against them, and a host of other things. These predictions are then tested by empirical studies across diverse fields of biology; population genetics, medical genetics, animal breeding, ecology etc etc. In this way we don't just look at the power of evolutionary theory to explain what has happened, but we also assess its predictive power for telling us what will happen.

Hope that helps and sorry for rambling a bit!

Last edited by Alistair Wilson (5th Feb 2009 15:28:12)

Predictions need to be based on specifics, not general ideas.

For example, an evolutionary thought experiment was carried out by Alexander (1974) to understand what conditions might be required for a mammal to be eusocial. Subsequently a mammal was identifed from an environment that matched the conditions of the prediction and after a study of its behaviour it was found to be eusocial (the naked mole-rat).

There are other examples, particularly when addressing the fossil record. Feathered theropods were predicted using evolutionary principles before they were discovered, four-winged gliding dinosaurs were predicted by Beebe nearly 90 years before Microraptor gui was found.

Predictions derived from evolutionary principles can be tested, as long as they are formulated in a testable way.

Alexander RD. 1974. The evolution of social behavior. Annual Review of Ecology and Systematics; 5:326-83.
Beebe C.W.A. 1915. Tetrapteryx stage in the ancestry of birds. Zoologica, 2: 38-52.


Thanks for starting an interesting discussion about the nature of evolutionary theory. To a large extent, the difference between the relatively clear, crisp predictions of relativity theory and quantum theory and the more vague and contingent predictions of evolutionary theory is just the difference between physics and biology. Plants and animals are a lot more complicated than protons and electrons, and the evolution of any given species is guaranteed to depend on a large number of variables relating to both the surrounding environment and the species itself.

Nevertheless, evolutionary theory does make some general predictions. The most basic one is that the species making up the Earth's fauna and flora will change their characteristics as future generations unfold, and that scientists will find evidence of their having changed in the past. (As should be clear from Paolo's post, a "prediction" in the scientific sense does not have to be about the future - evolutionary theory also makes predictions about what evidence of past change scientists should find when they look at the fossil record, or at currently living organisms.) This change might occur through divergence into new species, or through modification of existing species.

Natural selection, of course, is widely regarded as the most important mechanism of evolution. If we accept natural selection as a cornerstone of evolutionary theory, we can make another prediction: that populations of organisms, over time, will tend to develop features that make each individual more likely to survive in its environment. As the environment changes, adaptations to the new conditions are likely to appear and to spread through the population. As you recognised in your original post, the response to a changing environment is hard to predict in any individual case. For example, imagine a population of deer inhabiting an area where the climate is becoming progressively colder. The deer might grow thicker coats or otherwise adapt to the cold, but if the right mutations fail to come along (which is purely a matter of chance) they might be unable to adapt and simply go extinct. They might also be able to withstand the new conditions (perhaps in reduced numbers) without significantly changing their characteristics, at least for a while. Finally, they might simply move to somewhere warmer.

In other words, you're right that a wide range of outcomes, in any given case, would be compatible with the theory of evolution. (Again, that's the messiness of biology.) However, the theory does predict that adaptation to changing conditions will be a common phenomenon, and that this will occur through the accumulation of individual genetic mutations that spread through the population in question. There's probably no single observation that could make scientists suddenly reject the whole body of interlinked propositions and ideas that we call evolutionary theory, but we would have to worry if anomalous observations began to show up too frequently. Anomalies might include:

1. Examples of species that appear to have survived for many millions of years without substantial change. (Some species like this are known, but it ought to be a relatively uncommon phenomenon.)

2. Examples of extremely sudden (miraculous?) change in the structure of a species. (In theory, a mutation could appear that produced a very large and advantageous change all at once, but this should be VERY rare.)

3. Examples of populations that become LESS well adapted to their environment over time: those hypothetical deer I mentioned earlier should not be evolving thinner coats in response to the cold, or giant ears that would provide a large surface for heat loss. (However, random fluctuations in gene distribution, called "genetic drift", can be important in evolution, and I suppose this might produce a counteradaptive trend now and then through sheer chance.)

4. Examples of fossils that are outliers in time, appearing in the geological record either well before their close relatives evolved or well after they went extinct. (But the fossil record of some groups is poor enough that surprises like this are bound to turn up occasionally.)

Fortunately, anomalies of all these types (and others I haven't mentioned) really are extremely rare. When biologists look at past and present life on Earth they see evidence of modification, diversification and extinction almost everywhere. Change sometimes occurs by fits and starts on shorter time scales, but over spans of millions of years or longer the fits and starts mostly smooth out and the pattern becomes quite stately and gradual except for the occasional mass extinction. The changes that do occur are often obviously conducive to survival in the environment the species inhabits, and virtually never obviously counterproductive. On the whole, the fossil record reveals lineages made up of transitional forms that occur in a logical sequence: for example, the evolution of whales can be traced from their land-dwelling fossil relatives through a series of intermediates that became progressively aquatic.

One final point, to finish off this already over-long post: fitness CAN be predicted independently of survival, although the complexities of biology preclude any great precision. A mutant polar bear with black fur would stick out like a sore thumb during those long Arctic winters, and presumably would have difficulty catching much prey. It's often not too difficult to guess whether a particular change in an organism's structure would be likely to increase or decrease its fitness, although guessing the level of impact might be more difficult.

A quick reprise of the Popper business first.

Popper advocated a particular type of argument in his earlier work. He then made the famous statement about evolution not being science, because it was not "hypothetico-deductive" and incapable of setting up "critical experiments". He later changed his views, possibly because some biologists explained what evolution DID predict, and actually came to see the evolution of philosophical ideas as a type of selection process. Anyone who uses Popper's earlier statements probably hasn't read his books, which I admit can be tough going. I recommend Michael Ruse's books, as they deal with evolution from a philosopher's point of view but are not laden with jargon.

Biology and astronomy have the problem that they are both historical sciences and much of the work done in these fields is often related to coming up with the best general explanation of classes of phenomena. The technical name for this is inference to the best explanation, but to give you an idea of how this works, look no further than Sherlock Holmes. Although Holmes claims to use deduction, he in fact uses induction (technically abduction). He looks at the data and comes up with the best explanation for all the facts. Sometimes Holmes gets it wrong, because another explanation is possible, or the facts do not lead to an inescapable conclusion the way they do in genuine deductive logic.

To address the more practical part of your question, it is hard to make future predictions about a lot of issues. Conservations biologists are quite good at predicting extinction, based on current environmental needs of species (see the work of J. H. Brown and his group at Arizona) but the difficulty is that there are very few cases in which we have any significant grasp on the interaction between  genes, both with each other (epigenetics), and the environment which results in the organisms we see (phenotypes). Evolution in the lab is one thing, evolution in the field is another with a lot of variables.

The more fruitful approach to think about evolution and its predictions is to ask whether observed examples of structures, behaviours and other aspects of biodiversity are in agreement with the concept that:

1. More individuals (insert your level of selection here) are born than can survive
2. Variation exists among those individuals
3. That variation is heritable to some degree

With some exceptions, explicable by other theories such as sexual selection, much of life on Earth can be explained by application of this formula.

"Hope is a duty from which palaeontologists are exempt."
David Quammen