``This isn't true, Chuck. It is very easy for non-adaptive or even anti-adaptive traits to survive, because it isn't this trait or that trait that survives, but whole complex bundles of traits. Gould's hypothetical example is a brain trait that had adaptive use for a paleozoic fish, ... Carrol
-----------
Now, back home with the book in front of me.... I have to say, I got sick of this pretty quickly after forcing myself to go back over Kuhn...
I don't want to pick away, but the key word was `common' attribute. However the point was that adaptive was not a useful term. Hence, anti-adaptive or non-adaptive are not particularly any more useful. In this context, Gould as you point out uses exapation (43p), for structures coopted for utility, arising from different sources of origin, noting also the non-adaptive. But he also notes his own, growing doubts about adaptationism more generally (44p). And, that was the main point I was trying to make, even if I got the argument wrong.
In any event, if you continue on he re-crosses this same terrain coming from other directions. This is because the problem is not gotten around by pointing to various counter examples, since they inevitably require a prior appearance (an appearance usually assigned as adaptive). The whole argument of the appearance of non-adaptive traits in the orthodox view depends on random or quasi-random mutations that are supposedly culled by natural selection. But Gould is also forced to modify or add to this source through structural morphology (I can't find the exact term at the moment).
So then if you are allowed anti-adaptive traits, where's the natural selection pressure to cull them? If you allow traits to arise beyond the reach of natural selection, then in what sense does such selection operate as selection? The answer is moving up to species.
The idea that there are bundles of traits comes from the problem of the difference between genotypic and phenotypic expression. There isn't a direct causal chain between genotype and phenotype. This is where the whole one-gene, one-protein controversy arises and then breaks down. So without that causal chain, in what sense can natural selection work to cull genes since it operates on phenotypic expression. So there is a lot of loose territory in there.
And so it goes around and around and these arguments form some of the ground work for the revision. Here is what Gould says (41p):
``Against this orthodox background [some of the above]---or, rather, within it and quite unconsciously for many years---I worked piecemeal, producing a set of separate and continually accreting revisionary items along each of the branches of Darwinian central logic, until I realized that a `Platonic' something `up there' in ideological space could coordinate all these critiques and fascinations into a revised general theory with a retained Darwinian base..''
The `up there' leads to considering whole species as the object of selection in the context of macro-evolution. This is completely counter to first the orthodox focus on individuals and second, on the reductionist orthdoxy which focuses on genes.
In tackling the problems of adaptation, Gould uses the functional or operational version, of emergent fitness as a substitute for emergent characters.
Instead of discrete traits or bundles of discrete traits, what Gould substitutes or rather integrates is a morphospace with pathways and structural constraits. (see the reponse to Ian)
Maybe this will drag me back to read this.
Chuck Grimes