The Critical Rationalist Vol. 01 No. 04 ISSN: 1393-3809 31-Dec-1996
5.2 ... or Metaphysics?
5.1.4 Misconception 2: Words
(76) The final misconception which arises does not strictly involve a tautology, and is only incidentally inspired or supported by Spencer's phrase. However, it does involve a degree of "circular" reasoning, is sometimes said to be tautologous, and is commonly associated with the strictly tautologous misconception(s) of the previous section. It is therefore appropriate to consider it here.
(77) I consider this misconception to be the most significant and pernicious of all, and will discuss it in some detail.
(78) I shall present the "argument" in what I consider to be its plainest form, but caution that it is rarely if ever expressed in such explicit terms.
is proposed as a solution of
--i.e. as an explanation for the growth of
adaptive complexity in the phylogenetic tree.
to successfully solve would require
that it predict (stochastically or otherwise) the growth
of adaptive complexity.
actually predicts is a
growth of S-value in certain lineages.
can be said to solve
only by (re-)defining adaptive
complexity to be the same as S-value (fitness). This
effectively uses a circular definition, which is just
such that the problem to be solved (originally
) becomes simply "whatever problem
can be solved (by )".
(79) Arguments of this sort are tacitly involved whenever there is a debate about the relationship between "fitness" and "adaptation". Darwinism is first taken to be a theory of the growth of "adaptation" (informally conceived of in terms of adaptive complexity, or "fit" between organism and environment--which is to say knowledge); but it is then noticed that Darwinism per se can actually only explain the growth of "fitness" (in the sense of S-value); so it seems that it can work as an explanation only if adaptation is defined as equivalent to fitness (and we forget our original informal notion of adaptive complexity, or adaptation to an environment). This does not turn Darwinism into a tautology in any strict sense (though it does involve a kind of circularity); but if this step is taken then Darwinism loses its ability to solve the kind of problem we were originally interested in--for the terms of the problem no longer appear in the theory. It is thus greatly diminished in scope and significance.
(80) Popper raised essentially this problem in what was (as far as I am aware) his earliest consideration of the status of Organismic Darwinism--his Herbert Spencer lecture, Evolution and the Tree of Knowledge:
... survival, or success in the sense of an increase in numbers, may be due to either of two distinguishable circumstances. A species may succeed or prosper because it has managed, say, to improve its speed, or its teeth, or its skill, or its intelligence; or it may succeed or prosper merely because it has managed to increase its fecundity. It is clear that a sufficient increase in fecundity depending fundamentally on genetical factors, or a shortening of the period of immaturity, may have the same survival value as, or even a greater survival value than, say, an increase in skill or in intelligence.... But be this as it may, it should be possible, I think, to [measure] the success in the adaptation of the individual organisms of a species ...
Without some distinction such as this ... we are liable to lose sight of the original problems of Lamarck and Darwin, and especially of the explanatory power of Darwin's theory ...
(81) Popper is essentially pointing out that the temptation to equate adaptation (or adaptive complexity) with fitness (or S-value) must be resisted at all costs, for otherwise we lose contact with the problems we wish to solve. Lewontin (1978) has discussed this problem in very similar terms. Hull has also recently made much the same point, concluding that the requirement to identify adaptive complexity (which he actually calls "fitness"!) independently of S-value (which he terms "differential perpetuation") cannot be circumvented "without evolutionary theory degenerating into an empirically empty formalism" (Hull 1980, pp. 318-319).
(82) Now I have, indeed, been careful not to define adaptive complexity in terms of S-value (or fitness). Granted, I have not attempted any formal or detailed definition of adaptive complexity; and I have particularly eschewed any attempt to establish a metric for it. But, as already discussed in section 3, this vagueness is not untypical in the field; and I would argue that my general formulation in terms of inborn knowledge (Popper 1961, pp. 258-259) is still a more definite ontological commitment than is usual.
(83)
It must be repeated that misconception 3 does not strictly
involve a tautology. It is in this light that we must read
Hodge's (1983, p. 58) claim that it is
"a mistake to defend natural selection against the tautology
objection by proposing criteria of fitness independent of
reproductive success". I suggest that Hodge is here referring
only to the strictly tautologous arguments already
discussed (particularly the two cases of misconception 2), and
not to misconception 3, which has quite a separate
character. Dunbar (1982, p. 10), on the other
hand, rejects the argument that cannot be
tautological because it is empirically testable (i.e. the kind
of argument offered by Hodge), saying that "this claim misses
the point entirely". I contend that Dunbar is effectively taking
up a position precisely complementary to that of Hodge, confining
his attention exclusively to misconception 3 and ignoring or
dismissing misconceptions 1 and/or 2. My position is that,
despite the apparent contradiction between Hodge and Dunbar, they
are actually both correct, so far as they go--but they are
discussing different problems (I consider Dunbar's analysis
in more detail in paragraph 94).
(84)
At this point my argument is that misconception 3 is mistaken in
concluding that adaptation or adaptive complexity should, or
must, be defined in terms of S-value (fitness). I therefore
insist on retaining essentially independent definitions of
adaptive complexity and fitness (and thus retain in its
interesting form). What then are we to make of the original
criticism--that cannot actually solve this problem, for
it does not predict the growth of adaptive complexity
(so-defined)?
(85) This brings us to the nub of the problem, which is to ask
how much we can sensibly ask of a proposed solution of
. It seems to me that the error lies precisely in
supposing that a solution must (or even can) take the form of
some general theory which predicts a growth in adaptive
complexity. This derives in part from a fundamental
misunderstanding of what actually says.
recognises that there has been a sustained growth in adaptive
complexity, in at least some evolutionary lineages, and asks for
an explanation of how this could be (preferably one which does
not assume the pre-existence of an even more complex creator). It
does not say that growth in adaptive complexity
must occur (in general or in particular lineages); or
that it had to occur in the particular way which it did; or that
it must continue occurring. But only the latter kinds of
problem would call for a solution which incorporates a general
prediction of growth of adaptive complexity.
(86) We might characterise the general difficulty here as a
supposition that entails some kind of guaranteed,
monotonic, "progress" (in adaptive complexity). It is a
harking back to the "great chain of being". As Gould
puts it: "The familiar iconographies of evolution are
all directed--sometimes crudely, sometimes subtly--toward
reinforcing a comfortable view of human inevitability and
superiority" (Gould 1989, p. 28). The idea of
necessary progress in evolution is so deeply entrenched that it
is very difficult to free oneself of it. Not even Darwin himself
was completely immune. Although he explicitly emphasized that
the absence of "perfection" in biological
organisms should be interpreted as positive evidence in favour of
the operation of natural selection (at least as compared with a
theistic theory), we still find him remarking, in the concluding
pages of The Origin, that "... as natural
selection works solely by and for the good of each being,
all corporeal and mental endowments will tend to progress
towards perfection" (Darwin 1859, Chapter XIV, emphasis
added).[11]
(87) Once, however, we free ourselves from the
preconception that organismic evolution is a story of
necessary or inevitable "progress"--and
excise this idea from --then the way is
open to consider as offering at least a
partial solution.
(88) Briefly, as has been commented upon several
times, cannot and does not predict the
growth of adaptive complexity. However, given that
adaptive complexity has grown, can
retrospectively offer a generalised explanation for
it--namely that the growth of adaptive complexity
"happens" to have been correlated with the
growth of S-value--and, crucially, can do
this without postulating the existence of any
"designer" (prior knowledge) directing the
overall course of evolution.
(89) in itself, does not explain why
adaptive complexity should be correlated with S-value
either in general, or in any particular case; not does
it explain how (heritable) adaptive complexity can
increase at all, in general or in any particular case.
In addressing it does entail that
some (heritable) increases in adaptive complexity have
occurred, and that, of these, at least some have been
correlated with S-value; but it specifically
denies that such increases in adaptive
complexity have been "designed" or
"anticipated", or that they have been
universally correlated with S-value.
(90) Now, of course, does
predict more or less monotonic "progress" in
one very specialised sense: increase in S-value (within
some lineages). But that much is true of all systems
where (S-lineage) selection processes arise, and does
not impinge at all on the central problem of the growth
of adaptive complexity. No doubt we can conduct a
certain amount of evolutionary theorizing without ever
referring to adaptive complexity: but we can certainly
never solve . The crucial extra step, which
is rarely made explicit is to say that, given
, adaptive complexity can grow in
evolutionary lineages if and only if at least some
("unjustified") increases in adaptive
complexity occur, and, of these, at least some are
correlated with a net increase in S-value (i.e. are
selectively retained).
(91) Of course, the reasoning given here only works at all if it is accepted that an increase in adaptive complexity might be correlated with greater S-value. If I had defined adaptive complexity in some (strange) way which was intrinsically opposed to S-value (say, necessarily involving unconditional altruism on the part of the S-lineage) then the argument could not go through. But my actual definition--which corresponds to something like "inate knowledge"--does precisely have the characteristic that, ceteris paribus, it may be expected to be correlated with higher S-values. But the ceteris paribus clause is absolutely crucial here; to ignore or omit it would effectively mean a reversion to equating adaptive complexity with S-value, and thus defining away the real problem.
(92) None of this is to say that the growth of
adaptive complexity cannot be explained (or even
predicted); it simply claims that there cannot be any
general theory of it (as always, this is just
another way of denying the existence of a logic of
induction). asserts precisely that, for all
particular historical cases of an increase in
adaptive complexity, there is a particular
explanation, involving an "unjustified"
variation in adaptive complexity (which happened to be
an increase) which was selectively favoured. It may or
may not be possible to organise these particular,
historical, explanations into a smaller number of more
general cases: but I take the view that it will not be
possible to translate these particular explanations,
nor generalisations of them, into predictions for
continuing growth of adaptive complexity into the
future. The aggregation of all the particular cases (if
such could be individually established) would then be
the complete (historical) "explanation" of
all growth of adaptive complexity in the biological
world. But does not assert that such growth
had to occur, nor that it will continue into the
future, nor even that it would necessarily recur on
another "similar" planet.
(93) The confusions and misconceptions discussed here have centred on the distinction between "adaptation" and "fitness". For this reason I have tried to avoid these terms in my own general presentation of Darwinism: the only lingering remnant is the "adaptive" in "adaptive complexity". I have retained this in deference to the existing biological literature, but I suggest that it might actually be better to eliminate even this concession. In speaking of the "adaptation" of biological organisms it seems almost impossible not to think in terms specifically relating to their success in living and procreating--which is to positively invite a reversion to the relatively sterile concept of S-value.
(94) A particularly "good" example of this is Dunbar's analysis (Dunbar 1982). He is very clear about the need to distinguish "adaptation" and "fitness". Furthermore, he seems to adopt much the same kind of distinction as I have suggested above, interpreting adaptation in terms of "problem-solving" (Dunbar 1982, p. 11) (following Lewontin 1978, among others). He seems to recognise the essential independence of the two concepts when he cites Dobzhansky (apparently favourably) as saying that "we cannot draw inferences about fitness from a knowledge of adaptation, nor of adaptation from a knowledge of fitness" (Dunbar 1982, p. 14). However, he then goes on to explicitly deal with the alleged circularity of Darwinian explanations, in the following terms:
The relationship between the concepts of adaptation and fitness might seem to confirm the worst fears of the anti-Darwinians. Each appears to depend on the other in a way which makes them virtually inseparable. It is, however, crucial to appreciate that they are not definitionally interdependent: adaptation is not defined in terms of fitness, nor vice versa. Adaptation (and hence reproductive success) is defined with reference to individuals, whereas fitness is defined with reference to genes and is thus a characteristic of populations.(Dunbar 1982, p. 16-17)
Dunbar here seems to suggest that the distinction between adaptation and fitness is (merely?) a distinction between properties of organisms and (consequent) properties of lineages. The best interpretation I can offer of this is as a somewhat tortuous reference to what I have called misconception 2 above--effectively a failure to distinguish I-survival and L-survival. Having thus retreated from the real issue--misconception 3--Dunbar finds that he must admit that a significant circularity may still remain in Darwinian theory. It is perhaps not surprising that he then resorts to the philosophical relativism of Kuhn and Feyerabend as his final defence of Darwinism against circularity. I, of course, take the view that such a conclusion is unsatisfactory and unnecessary--that the conceptual independence of fitness and adaptation (S-value and adaptive complexity) can and should be recognised, and this can be done without depriving Darwinism of its power as an explanatory schema. However, it does underline the point that the terminology of "fitness" and "adaptation" may be critically flawed. Thus, if one confines oneself to discussion in terms of (inate) knowledge instead of adaptation or adaptive complexity, it may be easier to remember that there is no necessary connection between these things and their retention or growth under natural selection. It should then be clear that any connection which may exist will have to be individually argued for in each particular case.
(95) In short, I propose that the term "adaptation" (and "fitness" too, for that matter) be henceforth considered harmful, and be avoided in formulating and presenting the theory of Organismic Darwinism.
5.2 ... or Metaphysics?
5.1.4 Misconception 2: Words
The Critical Rationalist Vol. 01 No. 04 ISSN: 1393-3809 31-Dec-1996
Copyright © 1996 All Rights Reserved.
TCR Issue Timestamp: Tue Dec 31 17:37:08 GMT 1996