At the same time, and on the other hand, any writing project like this has a dynamic of its own — I am doing a ‘reconstruction’ of certain aspects of these thinkers’ ideas, not trying to give them an exhaustive ‘close reading,’ but their texts take me in places I didn’t expect, and I find myself forced, if I am to write at all, to follow certain of their inflections, even if they take me someplace else from where I wanted to go. So in that sense, my writing isn’t personal at all.

It massively complicates it: you need to compute the effect of the mutation at time T-1 on the fitness function at time T-1 in order to know what fitness function to apply to the effects of the mutation at time T.

I think there’s an underlying assumption that the rate at which mutations affect the fitness function is much lower than the rate at which the fitness function selects mutations, so there’s at least enough stability for the model to hold over a portion of history even if the landscape does change slowly over time. Forget what if anything Dawkins says about this, but I doubt he ignores it completely.

Your post makes an interesting suggestion, which I don’t have any ready answer to. Sometimes Dawkins, Dennett and others seem to be proposing the fitness landscape as a multidimensional phase space which gets explored (to a certain limited extent, as you note, given the problem of local optima) by the evolutionary algorithm (Dennet explicitly proposes that natural selection is an algorithm). In the sense that the phase space is conceived as preexisting (there is an enormous, but ultimately finite set of possible codon sequences , etc) I am not sure that Deleuze would accept it as virtual in his sense — when all the possibilities already preexist, there is no creation going on, and in this way it would be very different from the relation of a mathematical problem to its solution. To make the parallel to actualization in Deleuze’s sense, you’d need something more like a complexity theory approach to phase space (e.g. Stuart Kauffman) rather than Dawkins’ more atomistic approach. I suppose one could argue this in terms of how a change to an organism changes the environment as well, so that there isn’t a steady background fitness landscape, but every move on the fitness landscape changes the topography of the landscape itself… I don’t know enough math to have any idea whether this makes the problem less tractable to a Dawkins/Dennett algorithmic approach, boosts it to another level of magnitude in computational terms, etc.

Your larger question — “n what significant way does Deleuze’s Bergsonian vitalism add anything to the non-vitalist “adaptationist” account of how the “problem of light” might find a solution?” — is a really serious and important one. Can all the issues of emergent order, of interactions between the genome and other non-genetic biological factors in development, etc., all be comprehended in more reductionist, “adaptationist” terms? Do the recent developments in “evo-devo,” for instance, obviate the criticisms of adaptationism made by Susan Oyama and other developmental systems theorists? Can Kauffman’s emergent order, or Lynn Margulis’ vision of symbiosis as driving evolution, be recuperated in adaptationist terms? This seems to be the big issue right now, and I haven’t a clue as to how it will play out.

I’d say, first read SYMBOLISM: ITS MEANING AND EFFECT, which is a very short book, under 100 pages.

Then, part 3 of ADVENTURES OF IDEAS.