p245

The following comments are always incomplete.

Darwin process

The so-called Modern Synthesis that tried to synthesize evolution and genetics regarded the changes in genetic information was only adaptational and disregarded historic constraints on the genesis of variations and the effects of cataclysms. However, 

(1) Most base sequence variations are neither advantageous nor harmful significantly, and their fates are mostly stochastic ( determined by chance ) {the prevalence of neutral modes at the molecular level).

(2) As can be seen from the likely asteroid impact at the KT boundary, there must have been tremendous cataclysmic disturbances that derailed the `normal’ history ( importance of cataclysms ).

(3) Deep homologies indicate restrictions to mutations (the existence of  phylogenetic inertia ) [For example, as to eyes see: W. J. Gehring and K.Ikeo, Trends Gen.  15 , 371 (1999)]

 Thus, as Gould was criticizing, the reliability of the results of the so-called Modern Synthesis has been questioned.

Has Darwinism been demonstrated?

Darwinism is an assertion that only the Darwin process can initiate any Pasteur chain, and

can increase complexity along it. Actually, what Darwin and Wallace really asserted were mainly the following two points:

(I) All the organisms can be arranged into a single tree according to the parent-offspring relation (into universal phylogenetic tree).

(II) The Darwin processes can start Pasteur chains, and can increase complexity along it.

The core of Darwinism is the rejection of the necessity of the designer. If Darwin process is the sole cause of changes, then all the novelties came only from randomness. Consequently, without scale interference no complex system can emerge. 

 From the point of view of this book, the merit of Wallace and Darwin is the proposal of a mechanism to produce complex systems without design. Whether the mechanism was actually used as a sole means does not matter. Of course, this statement is a bit too extreme, but, in any case, the proposal itself is regarded as a great contribution. 

 Needless to say, logical possibility need not actually happen. To deny the actual relevance of Darwinism is to assert that the process of selection of the results of mutation cannot produce the organisms we now see on the earth. There are two kinds of this assertion, the assertion of impossibility in principle, and that of conditional impossibility.

(I) [Impossibility in principle] The set of the organisms that can be produced by randomness (i.e., the totality of the outcome of mutations from which actual organisms have been selected) is not large enough. In other words, random mutation cannot do much (noise is not a universal information source). Since the total set is not large, selected results cannot include all what we see.

(II) [Conditional negation] Even if random mutations can in principle produce anything biological, it is impossible to do so within 4 Ga. That is, noise can be a universal information source, but it has not enough time to exercise its power [The so-called developmental constraints advocated by Gould is understood, usually and erroneously, as this type of objection to Darwinism.].

[It is a well-known fact that Thomson (later Lord Kelvin) tried to demonstrate that the age of the earth is at the longest a few million years to destroy Darwinism. See K\"{o}rner’s Fourier transformation text book devoting one chapter to this problem, or Gould: SET p492, which starts with “ In 1866, William Thomson, the future Lord Kelvin, published one of the most arrogant documents in the history of science---a one-paragraph paper (with an appended calculation) boldly entitled “The ‘Doctrine of Uniformity’ in Geology Briefly Refuted. ””

(I) above cannot be an effective argument as can be seen from the following argument. It is trivially  true that mutation cannot produce any organism made of organic compounds that is powered by thermal nuclear reactions, so obviously there are things mutations (noises) cannot create. However, all the organisms that ever lived on earth can be plausibly connected by chains of mutations. That is, if a starting point and a destination are given, it can be imagined that an evolution path can be traced with mutational steps whose effects are not large. We could say there is always a mutational homotopy just as is considered for the

evolution of eyes [1]. Actually, we can conceive numerous such paths [2]. That is, noise is not omnipotent, but it is at least as powerful as to produce any organism that we know. Thus, negation in principle of Darwinism as an agent of evolution is untenable [3].

[1] Nilson, “Pessimistic estimate of time required for an eye to evolve,” Proc. Roy. Soc. B 256, 53 (1994).

[2] Dennett calls this a difficulty of too many possibilities: C. Dennett,  Darwin’s dangerous idea . This book is with a flavor of pan-adaptationism, and the following severely critical review is quite appropriate: H. A. Orr, “Dennett’s dangerous idea,” Evolution  50 , 467 (1996).

[3] However, to cover the genesis of life with this argument, a large amount of uncertainty comes in. Therefore, INH avoids this question. However, more plausible scenarios have been woven.

Noise is called noise because its effect does not have any guarantee of being beneficial to the organism.  Therefore, there are only two strategies to refute (II). In order to demonstrate that random noise alone is enough (no need to deny the possibility of other information sources):

(A) demonstrate the efficiency of parallel search under realistic conditions,

or

(B) demonstrate that the actual search space is not as wide as is assumed

or, in order to exclude information sources other than noises:

(C) demonstrate that non-random processes are harmful to increase complexity.

 It is hard to improve the plausibility of (A) further than its current plausibility (or mere improvement is not enough). The argument along this line often gives an impression that it is too convenient. For example, it is impossible to give a meaningful estimate of the probability of creation of life [1]. Since we cannot calculate the nonequilibrium noise statistics theoretically, the probability to evolve or to generate life is unthinkable in our right minds . [L. E. Orgel (of te Miller-Orgel reaction) emphasizes in his “The origin of life - how long did it take?” (Origin of Life and Evol Bio.  28 , 91 (1998)) that there is no way to estimate the upper or lower bound of needed time to generate self-reproducing systems.]  Therefore, although (A) may be sufficient, to pursue (B) and (C) must be the main strategy to refute (II).

 There are two attempts in the direction of (B). It is almost impossible to say that the search space (the space of all possible products of mutations) is small, but (B1) we have only to search in its factor space with respect to some equivalence relation, and/or (B2) the biologically meaningful subspace can be reached relatively easily. An example of (B2) is Kauffman’s idea of “order for free,” and an example of (B1) is the idea suggested by Fontana et al., from the study of the RNA evolution that in an appropriate neighborhood of any heritable state is a biologically meaningful state, so no global search is needed to evolve [2]. Detailed studies by Kauffman and collaborators of self-organization of the gene and polypeptide networks have demonstrated indeed some order reminiscent of the order in biological systems can emerge spontaneously [3]. However, because an important feature of ontogeny is that spontaneous development is patently impossible, and that of phylogeny is a long time massive parallel ‘computation,’ the relevance of Kauffman’s idea is likely to be

limited (at least it is irrelevant to the bottleneck of genesis of life).

 It is often claimed that the reason why there are so many anti-Darwinian claims is simply that their advocators do not study evolution properly. The claim contains a lot of truth, but it is still too early to dismiss all of them. Recall the arguments that surrounded atomism in the late 19th century; in this case the opponents were not ignorant; there were many philosophical questions, e.g., positivism. Clearly recognize that the assertion that Darwinism is sufficient to explain all the organisms we see now has not yet been demonstrated

empirically nor logically. 

 Let us consider (B) in more detail.  It is not surprising that evolution would proceed in the direction to make evolution more efficient (or make random search more efficient). To this end

(1) Reduce the cost of detecting errors (make errors more quickly detected)

Although random noise is the source of innovation, it also destroys established order and functions. Therefore, it is desirable to remove such harmful noise effects as efficiently as possible. This is about developmental constraints, inner selection, etc. That is, damaged systems should be excluded as early as possible in their development. However, this alone is not enough to increase the search efficiency of `good systems.’ The systems that satisfy developmental constraints must function well; good beginnings must make good endings. This is only possible by imposing high correlations among traits. This is closely related to:

(2) Make parallel search efficient

For example, we human beings must not have been selected to recover from a heart surgery, but the mechanism that makes this recovery possible has been tested by repairing scratches. The machinery to repair scratches could be tested without big costs (to the extent the law of large numbers can be exploited). There must be many such things Needless to say, the organization of the system in this way itself is an outcome of random mutations, but the evolution of the mode of evolution is a logical necessity, so improving the random search capability is also not surprising. Without doubt there are situations where Lamarckian mechanism is advantageous, so Darwinism should have evolved emulators of Lamarckism as well. This may be called 

(3) Emulated Lamarckism.

Waddington’s genetic assimilation is a typical example [If a particular phenotype is selected, even if it does not have any genetic basis (i.e., it is only a phenotypic variation), that phenotype tends to have a genetic basis. This is not the inheritance of an acquired trait, but due to the selection selecting a set of gene alleles that produce the phenotype readily. Waddington’s autobiography is a good introduction. See also a recent review: M. Pigliucci and C. J. Courtney, “Genetic assimilation and a possible evolutionary paradox: can macroevolution sometimes be so fast as to pass us by?” Evolution  57 , 1455 (2003)].

(4) Regulation of (the effect of) mutation rates

The evolution of evolution mechanism (i.e., the evolution on the meta-level) is thinkable, so it must exist. For example, increasing mutation rates when the environment is not favorable [This indeed happens, as is well known]. Also, if mutation is not desirable, its effect should be buffered. In this way mechanisms to reduce the harmful mutation rate must exist [and it is the case].

 Without Darwinism no organisms can be maintained for a long time as we will discuss soon. What theoretical or mathematical physicists could do is not to collect empirical data.  They should think more general and inevitable things. 

 From the mathematical point of view, the ‘excellence as a theoretical framework’ of Darwinism must be clearly appreciated. The reader might say this is a matter of taste. The matter of taste is fundamentally important to any theory.

Side issues of evolution

Various topics have been talked about as if they shake the very foundation of Darwinism, but they are all side issues .

(A) Punctuated equilibrium theory

Darwin asserted that big changes are all due to the accumulation of small changes . This implies that the large-scale evolution can be studied (or at least its mechanisms can be studied) on our time scale. That is, large scale evolution can be studied in the laboratory. This opens up the possibility to study evolution as a topic of the ordinary natural science. Needless to say, the assertion is neither based on empirical facts nor logical deduction. According to this logic, all the organisms are gradually changing at every moment as functions of time. Eldredge and Gould pointed out that paleontological data showed that the steady gradual changes are rare and that evolution processes often consist of long stases without appreciable changes and short phases (on the geological time scale) of rapid changes, and proposed that large scale evolution is not gradual but consists of stases punctuated by large changes. If we combine the fact that many species are stable to the extent that they were misunderstood as created by God, and the idea that evolution process is on the average not very different now from the past, the punctuated equilibrium is a logical consequence [As to this point Gould says on SET p435, “ We know, after all, that most species are stable in both current and paleontological perspective. If a lineage were as mutable as O. Lamarckiana, we would never be able to designate Linnaean taxa. ” This must represent naturalists’ opinion].（Notice, however, the idea of punctuated equilibrium can be found in the  Origin of Species .)

(B) Neutral theory

Neutral theory points out the possibility that natural selection need not have any adaptive significance.  The neutralism controversy is a controversy completely within Darwinism. Mathematically, we can summarize it as the question whether the law of large numbers is applicable or not. (If we may apply the law of large numbers, then almost surely there is no neutral gene). If we take account of the fact that the number of combinatorial possibilities of genes (whatever definition you may use) is astronomically more numerous than the total number of organisms, it is obvious that we can never apply the law of large numbers to the statistics of gene combinations. Thus, neutrality is just entailed by the random nature of mutations, and reinforces the basic idea of Darwinism.

(C) The effect of contingency

Contingency has non-negligible effects on large scale evolution. Whether amino acids are $d$ or $l$ is often mentioned as an example. This is just consistent with Darwinism, but we should not lightly have recourse to this. [Gould regards the number of legs (4) and the number of fingers (5) on a leg as examples of frozen accidents. However, since the number of fingers of newts can mutate, it is not easy to swallow that the number 5 is accidentally fixed. There is a possibility that the number of legs of large animal being 4 is adaptive. For example, observe with how many points large running centipedes or rapidly moving sidewinders touch the ground. If the inertial effect is smaller, it is likely to have more legs as hexapods. 4 and 6 seem to be the mechanically smallest even numbers.]

(D) Structuralism

This is nothing but to admit the importance of mesoscopic structures between genes and phenotypes, a very natural idea in the modern physics. This simply makes Darwinism more flexible.

(E) Controversy as to the level of selection

In contrast to the idea represented by  Selfish Genes that the selection works at the level of genes, there is an idea represented by Gould that selection works at each level of organization of organisms (for example, on the species level [a long argument can be found in Gould SET Chapter 8]). This controversy is, if we use physicists’ parlance, whether collective coordinates are required or not. It is possible that even when we talk about individual genes, they may be something like dressed particles, or renormalized particles [A precursor of the idea may be found in P. Hammerstein, “Darwinian adaptation: population genetics and the streetcar theory of evolution,” J. Math Biol. 34, 511 (1996)].

Language and thought

There is a paper asserting that the portions of brain used for logical inference and for language are different: 

Logical inference is not embedded in natural language

Montiet al., The boundaries of language and thought in deductive inference

Proc Natl Acad Sci  106  12554 (2009)

(Stupid Question) Is human thought fully embedded in language, or do some forms of thought operate independently? 

*3T fMRI study

*When contrasted with matched grammaticality judgments, logic inference alone recruited

‘core' regions of deduction [Brodmann area (BA) 10p and 8m], whereas linguistic inference alone recruited perisylvian regions of linguistic competence, among others (BA 21, 22, 37, 39, 44, and 45 and caudate). In addition, the two inferences commonly recruited

a set of general `support' areas in frontoparietal cortex (BA 6, 7, 8, 40, and 47). 

*Logical inference is not embedded in natural language and confirm the relative modularity

of linguistic processes.

 As can be seen, this paper compares two distinct types of thoughts that use language, and does not mean that thoughts come from the part of brain different from that used for the manipulation of language.