What is fitness?

Phil Madgwick
Aug 27, 2017 · 5 min read

Hi Tam, thanks for responding and pointing out your article — I greatly enjoyed reading it. I can sympathise with the ‘expected fitness’ or propensity interpretation. For some reason I have always associated this with Sober’s The Nature of Selection, which I have somewhat mixed feelings about.

Gould and others’ argument is a kind of intuitive or colloquial definition of fitness. I do not know how much you have encountered this when quizzing other evolutionary biologists, but this kind of approach to definition is hugely popular within science as a whole — and evolutionary theory is no exception. Scientists are generally very snobby about philosophy because, I think, philosophers are unafraid to present flawed-answers to insoluble problems. The ability of scientists to avoid getting bogged down in definitional problems is usually maintained with appeal to common-sense, and this is basically what I see Gould/Gardner doing. Whilst I do not share the tradition of dismissing the utility of philosophy, I am content with the common-sense approach to definition in most practical discussions, especially about fitness.

In defence of this point of view against the argument that there is no a priori definition of fitness, I think I simply don’t agree but this may be just semantics. If you look at how evolutionary biologists use the term in the adaptationist paradigm, typically fitness is treated as being approximated by some continuum of the particular trait that is under investigation (excluding other features that are not highlighted as relevant in the study). For example, see this favourite paper of mine about tail length in widowbirds. The question that is being asked could be phrased: How does proposed-adaptation X effect fitness? ‘Tail-length’ and ‘mating success’ are just proxies for ‘X’ and ‘fitness’ respectively.

It is a legitimate question to ask how well ‘mating success’ approximates fitness, but the precise meaning of fitness is not taken as something that is a necessary question for evolutionary theory to precisely resolve because empirical research can only make use of proxies (because fitness is a non-physical property that cannot be directly measured like, say, a barometer measures pressure). I think Gould/Gardner, from this line of thinking, would not be overly fussed with the need to use the word ‘expected’, though I also suspect that both would accept the substitution if it avoided debating semantics.

Now, obviously, the lack of clear definition is very dissatisfying, particularly to biologists like myself who can see the utility of philosophy. The problem, as I see it, is not one that is of concern to empirical researchers (in the main), but comes down to keeping the language that we use to describe natural selection conceptually clear. The intuitive way I like to think about fitness is using an ‘asset-price’ model, where I would equate fitness to a company’s total share price. The idea being that a share has a value even when shares are not actually being traded, and that value is some measure of present and future ‘worth’ in the same way that fitness is. I would avoid tautology simply by suggesting that the ‘asset price interpretation’ of fitness is operating in an ideal world that we cannot measure; thus the concept of fitness has autonomy from survival as a reflection of ‘future survival’ even if we cannot measure it.

From a mathematical point of view, the tautology problem has been absolutely resolved by similar logic (but without my intuitive interpretation). The ‘forces’ of evolutionary change are usually given as reducible to mutation, drift and selection. The quip ‘survival of the fittest’ is then taken to be equivalent to Fisher’s ‘fundamental theorem of natural selection’, which in simple terms demonstrates that natural selection increases mean fitness. By definition, ‘survival’ incorporates all three ‘forces’ and not natural selection alone, hence the term is not a tautology (even though Williams (1966) claimed it was, I think because of his fondness for the fundamental theorem that is necessarily a tautology like all mathematical theorems). This, I think, is the sense that Gardner and others would take in ignoring the tautology debate as ‘offering nothing new’ to existing biological theory.

Hopefully that clarifies my own, from my experience, other evolutionary biologists’ thoughts on the matter. I appreciate that you may find this ‘philosophical laziness’ infuriating for want of clarity.

I might add one more comment upon reading your paper. You state: “Natural selection is a theory that predicts differential survival and reproduction of organisms with the highest expected fitness.” I am not sure whether I disagree with the phrasing or the content, but it would be good to hear your response to what I am about to say. I would say that natural selection is often used (metaphor aside i.e. in intentional language sensu Dennett 1995) as the mechanism of adaptation. As such, natural selection is not really ‘tested’ in the way that I think your language is imagining.

In the hypothetico-deductive method, as used by Darwin, a process in described in terms of purely deductive logic. The connection of the logic to data then requires ‘bridging premises’ to realise the concepts that are discussed in the deductive logic. I think this is well-accepted (see Mayr and Ghiselin, who are notable amongst others). In this sense, the logic of natural selection is as much of a tautology as Fisher’s fundamental theorem would have you believe.

But, we do not test ‘evolution by natural selection’ as some grand theory of how life evolves. Instead, we look to specific examples where some particular trait in question is hypothesised to be an adaptation (i.e. under the scrutiny of natural selection). In systems amenable to quantitative genetic analysis, the strength of selection can even be estimated.

Testing of the grand theory of natural selection would presumably only work at a macro-evolutionary scale, but such questions become irrelevant because the answer to all traits is ‘some mixture of chance and necessity’ (sensu Monod 1971; meaning mutation-drift and selection). Instead, where reasonable, traits are assumed to be adaptations enabling inference about ecology or life-history traits that cannot be deduced from a face-value interpretation of the organisms whose skeletons are preserved in the fossil record.

I suppose the point I am driving at is that natural selection is, in my experience, a theory that predicts (and/or explains) adaptation — not fitness. With Brave New World irony, if a philosopher is someone who dreams of fewer things than there are in heaven or earth, I think an evolutionary biologist is someone who dreams of more things than there are in heaven but only focuses on what is on earth. (This is something I will develop further in response to your other question.) As we have both been struggling to find suitable disentanglement of these two terms, I think this distinction is meaningful.

Please do reply. I hope that my confession of dismissiveness and my comment on your paper does not come across as either rude or aggressive. I find it is sometimes hard to clearly say what I mean without an unsympathetic reader making such an accusation!

)

Phil Madgwick

Written by

Evolutionary biologist PhD (Milner Centre/University of Bath) on the general theory of Social Evolution, and have a personal interest in applications to AI.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade