Complexity--Yes! Irreducible--Maybe! Unexplainable--No!
A Creationist Criticism of Irreducible Complexity

Terry M. Gray
Department of Chemistry
Calvin College
Grand Rapids, Michigan

Mike Behe has given us a nice introduction to his notion of "irreducible complexity" and the inference of intelligent design of these complex systems that can be made from it. The argument, simply put, is this: in complex systems each component plays an essential role in the functional whole such that only when all the components are present and functioning correctly does the functional whole exist. Consequently, it is impossible for natural selection to produce such a system in a stepwise fashion since all the parts must already be present to in order to have a function to select.

I want to respond to this argument in four ways: two theological and two scientific. Let me outline them from the start:

1. Within a Christian theistic framework evolutionary explanations do not necessarily imply the absence of design.

2. Concrete responses to the biochemical arguments

3. Self-organization as an explanation of the origin of irreducible complexity

4. Premature appeal to special divine activity to explain the world around us damages the Christian theistic apologetic

Evolutionary explanations do not imply the absence of design

At the outset I want to make clear two things. First, in criticizing the "intelligent design" argument I am in no way questioning the orthodox Christian doctrine of design. The entire universe, as the product of God's creative and governing hand, is designed. Design is a consequence of God's creative activity. Even when we can account for features of the world with scientific explanations, a theistic view of the world sees these features as due to secondary causes for which God is the ultimate cause both in creation and providence. What I am disputing is the primary claim of the "intelligent design" hypothesis that there are clear cases of phenomena whose origin cannot be explained by ordinary causes.

Second, I want to assert that I have no "in principle" objection to the kind of divine action proposed by the advocates of the "intelligent design" hypothesis. Many of the miracles recorded in the Bible require such divine activity. However, in my opinion, the appeal to God's special activity to explain various difficulties with a naturalistic account (such as the origin of life, irreducible complexity, the Cambrian explosion, etc.) is unnecessary and premature.

The biochemistry: a few general comments

Before giving some concrete examples that illustrate why I am not pursuaded by the "irreducible complexity" argument, let me make a few general comments. First, I am not going to claim that I know how complex biochemical systems originated. My opponent is absolutely correct in his claim that little is known for certain and in much detail and that there is very little in the professional literature on this subject. What I will be presenting are broad outlines and hints of an explanation. They are most certainly wrong in the details, but they have convinced me that the "intelligent design" inference from "irreducible complexity" is fundamentally wrong-headed.

Second, it may be the case that the detailed evidence for the origin of specific examples of complex structures is hidden too deeply in the evolutionary past. Take, for example, the symbiosis theory of the origin of complex eukaryotic cells. The proposal is that the primitive eukaryote became a host to various prokaryotic organisms and that this symbiotic relationship stabilized in subsequent evolution. The symbiotic origin of mitochondria and chloroplasts from primitive bacteria and blue-green algae has nearly universal acceptance. Attempts to account for the origin other features of eukaryotes, such as motility systems, in terms of the symbiosis has met with less success. The point I wish to make is that these symbiosis theories merely push the question back in time and make the original evolution of these features extremely ancient. Since some of the irreducibly complex features of eukaryotic cells are shared by all eukaryotic cells, then, given an evolutionary scenario, the details of the origin of those features are exceedingly primitive and it may well be impossible to reconstruct the details. If that were the end of the matter, then my disagreement with irreducible complexity argument would simply be a matter of preference with no basis in reality. I don't think that that is the case. However, I will be modest enough to say that we may never be able to know all the details in the way that we might want to know them.

Finally, by way of introduction, to what extent have we really opened the black box of these complex cellular and molecular structures? Can we really say, as Mike Behe has said, that we know enough to know what can't be? I will be presenting below some evidence for the evolution of a relatively simple complex molecular machine, the hemoglobin molecule. In order to tell this story I need to know all the components of the machine, their ultrastructure and their detailed atomic structure, comparative sequences from many different organisms and from related gene families, and the structure large stretches of the genomes that code for these. I don't think that it is possible to tell a decent evolutionary story without all of this information. To my knowledge there aren't many systems for which we have this breadth of information. Hemoglobin is "easy" because it seems to be a recent evolutionary development. We have only recently figured out what all the components are for the systems Mike has described. We do not have an x-ray crystal structure of the ciliar motor. It is not clear that what we know about cilia extends much beyond mammals or vertebrates. There are not extensive sequence comparisons of relevant motility apparatus components among all the diverse groups of protazoans (where I'd expect to find the most important information regard the evolutionary origin of these machines) as there are for hemoglobin. It just seems to me that it is way too premature to say that we know what could not be or what could not have happened concerning these complex molecular assemblies. Mike is free to conclude whatever he wants to but in my opinion he is probably going to have to take it all back as more information comes in.

The biochemistry: hemoglobin

Hemoglobin is a complex molecular machine. The elucidation of the detailed molecular mechanism of reversible oxygen binding and its relationship with respiratory physiology is one of the few triumphs of reductionistic biochemistry. Hemoglobin is made of 4 protein subunits, each one having a heme group and an oxygen binding site. 2 of the 4 subunits are identical and are denoted alpha and the other 2 are identical and are denoted beta. Alpha subunits and beta subunits are very similar in three dimensional structure and also similar in amino acid sequence. An alpha subunit and a beta subunit assemble together to form an alpha/beta dimer. Two alpha/beta dimers make up the intact hemoglobin tetramer. There are two forms of hemoglobin: the deoxy form with no oxygen bound and the oxy form with oxygen bound. A key difference between the deoxy and the oxy form appears to be a rotation of about 15° of one alpha/beta dimer relative to the the other alpha/beta dimer. In the detailed structure it is as if there were two interlocking gears and in going from one form to the other they slipped a cog. The physiological observations are explained very nicely by this model. Oxygen does not bind very readily to the deoxy form of hemoglobin, so at low oxygen concentrations the fraction of possible oxygen molecules bound is very low. This corresponds to the condition of the blood in the tissues away from the lungs. However, when 2 of the four possible oxygen binding sites have oxygen bound, the oxy form becomes the most stable energetically and the whole tetramer switches over to the oxy form. Now the remaining two sites have a high oxygen affinity and bind very readily. So at the critical oxygen concentration there is a cooperative oxygen binding. This is the condition of the blood in the lungs. As the blood returns to the tissues, the oxygen concentration drops, the release of the first 2 molecules of oxygen results in a switch back to the deoxy form, which then makes the release of the remaining oxygen molecules easier, resulting in the dumping of oxygen to the tissues.

The irreducible complexity argument questions how such a complex molecular machine functioning in such a complex physiology involving the circulatory and respiratory systems could possibly have evolved step by step. I can say very little about the evolution of the circulatory and respiratory systems, however, as a result of protein sequence comparisons and the analysis of the structure of the globin coding regions of the genome, it is possible to construct a very plausible picture of the origin of a complex machine.

Another structural detail must be noted before we proceed with the evolutionary story. The alpha/beta dimer self-assembles as a consequence of greasy patches on the surface of each protein. This principle of assembly is due to the same principle that causes oil drops in water spontaneously coalesce. Interestingly, myoglobin, the oxygen storage protein found in muscle, has a very similar structure to the hemoglobin monomers, but it does not have the surface greasy patches, and so does not form multiple subunit assemblies. The two alpha/beta dimers self assemble by the same principles to form the tetramer.

While no doubt wrong in many of the details, here is a plausible evolutionary scenario that describes the origin of this molecular machine. This description is largely taken from Hemoglobin: Structure, Function, Evolution & Pathology by Richard Dickerson and Irving Geis. We'll start with a monomeric oxygen binding globin such as those found in some insects, annelids, and molluscs and even in plants. [The origin of the first globin-like structure will not concern us, although there is some speculation that it may have arisen from the cytochrome a which binds to the same kind of heme group.] These groups do not have globins that have differentiated oxygen storage functions (myoglobin-like) and oxygen tranport functions (hemoglobin-like). These globins do not have the complex oxygen binding behavior that hemoglobin has but are similar in their oxygen binding properties to myoglobin.

A key first step is a gene duplication event that allows the preservation of the original functional protein, but provides a second copy of the gene that can be altered by mutation, providing a source of new material on which selection can operate. There are multiple versions of globin genes that differ by only a few amino acids in insects and molluscs of the organisms mentioned above. In humans alpha and beta hemoglobin exist in gene clusters containing multiple copies of each type gene. In the alpha cluster there are two identical copies of the alpha gene, two copies of the alpha-like zeta globin (found in fetal hemoglobin), and one alpha-like pseudogene, that appears not to be expressed. In the beta cluster there five different beta-like genes and one beta-like pseudogene. It appears from various lines of argument that these have arisen by gene duplication followed by mutations. Some of the mutated copies appear to be functionless, whereas some of them appear to have new functions, i.e. in fetal hemoglobin with altered oxygen binding. These gene duplications are pre-adaptations, i.e. changes that occur for other reasons, but once they have occurred they provide the necessary conditions for some selectable function.

Sea lamprey globin evidences what might be the next intermediate stage. Sea lampreys have a separate myoglobin for oxygen storage and a hemoglobin-like molecule for oxygen transport. Lamprey hemoglobin is dimeric rather than tetrameric. It does display cooperative oxygen binding, though. Lamprey deoxyglobin forms dimers which dissociate upon oxygen binding. The dimer contacts are in exactly the region of the molecule where one alpha-beta dimer interacts with the other alpha-beta dimer. This is the region that modulates the 15° rotation and the cog-slipping effect that was described above. Murray Goodman and co-workers cite evidence from their sequence comparisons that suggest that mutations accumulated in this region of the molecule at four times the rate for the molecule as a whole during the evolution of this new function. Clearly, cooperativity of oxygen binding is a consequence of dimerization. But dimer formation is the result of greasy patches on the surface of the protein, which could well have arisen by a few amino acid substitutions (or even one as is the case in deoxyhemoglobin S fibers in sickle-cell anemia). Dimer formation would have been a Darwinian pre-adaptation to the evolution of cooperative oxygen binding.

The next step in hemoglobin evolution is the result of a gene duplication of the ancestral hemoglobin-like gene into the modern alpha and beta globin genes. Again, the original oxygen transporting function could be preserved, while mutations acted upon the second copy of the gene. The very similar but slightly different version of the globin allowed for the formation of the alpha beta dimer which upon interaction with another alpha-beta dimer allowed the preservation of the tetramer structure even upon oxygenation. Again Goodman's group believe that their sequence comparison data suggests that the alpha-beta dimer interface accumulated mutations at nearly twice the rate for the whole molecule during the evolution of this new function. Again, the gene duplication event and the alpha-beta dimer formation are pre-adaptations to the formation of the complex tetramer.

In the 450 million years since the origin of the hemoglobin tetramer there has been ample time to finely tune the primitive transport function. And there does appear to be additional evolution, especially related to the rise of warm-blooded creatures and, as I already mentioned in my discussion of the gene structure of the human alpha and beta gene clusters, the rise of placental mammals and special adaptations utilized in oxygen transport there.

Hemoglobin is a marvelous molecular machine. When we look at it in all its complex features, many of which we haven't even discussed today, we might imagine that it's like a mousetrap: you take one part away and it doesn't work any more. That may, in fact, be true with the modern finely tuned version. But I have shown a plausible evolutionary scenario based on the structural, mechanistic, genetic, and sequence data concerning globins from all parts of the animal kingdom.

The biochemistry: other proposals

I don't have the time, the expertise, or the vast amount of necessary data to make similar proposals for the specific complex systems that have been mentioned. However, I think that it is possible to see the broad outlines of an evolutionary explanation for even these. Here I want to focus on the same concept as I mentioned above, that of pre-adaptation. The irreducible complexity argument crumbles if each of the components of the complex system could arise having been selected for some other useful function. Gene duplication and mutations causing intermolecular interactions and thus assembly might result in a primitive version of the complex machine that now has a selectable function.

First, concerning the cilia. Are there any components of the cilia that function as other parts of the cell that could have been pre-adapted? The answer is an obvious yes. Tubulin and its assembled macrostructure, microtubules, functions in many other ways than in cilia. Microtubules are involved in cell structure and in mediating vesicle movement and chromosome movements during mitosis. In addition, there is a growing number of microtubule associated proteins, in general, and motor proteins proteins in particular. There are cytoplasmic versions of the ciliar dynein and kinesin. It seems to me that a cytoplasmic microtubule with a cytoplasmic motor protein is an incipient cilia. We don't even have to use our evolutionary imagination to come up with that. Nonetheless, I don't have a detailed evolutionary story to tell about the origin the cilia, but it seems to me to be very plausible that the various ciliar proteins could have arisen independently selected for other purposes and that a primitive cilia function could have arisen from a novel assembly of those already existing components.

Second, concerning vision. The argument has been made that the vision system is also an irreducibly complex system. Mike Behe has found no fault in Darwin's lack of concern with the origin of light reception at the detailed cellular and molecular level, but now with our opening of the black box of vision, we have no excuse for not concerning ourselves with those sort of details. Mike claims that upon examination of the open box that we must conclude that evolution of this complex system is impossible. So again we must ask, does the pre-adaptation argument get us anywhere in the discussion of the origin of vision? Again, the answer is an obvious yes. First, if we restrict ourselves to light reception, then I think that it's fair to say that a nerve cell is a pre adaptation to vision. Given a nerve cell, I don't have to explain where all those components come from (at least when explaining vision). Second, transducin, one of the key proteins involved in the light signal transduction from rhodopsin to nerve cell, is a member of the G protein family, a large family involved in all sorts of signal transduction events, including hormone signaling. The main novel feature of transducin is its specificity for rhodopsin. The generic G protein is a pre adaptation for transducin. Finally, rhodopsin, the main light reception protein, is a membrane protein similar in structure to other sensory receptors and to hormone receptors. These other receptors whose physiological effects are also mediated by G proteins may have been pre adaptations for rhodopsin. My answer here may be a form of question-begging, because you can always ask where did these other systems come from, but I think that the functional diversification of similar signal transduction sytems is reminiscent of the hemoglobin tale told above. What's needed is detailed sequence and structure information about all these proteins in a variety of organisms that are representative of the tree of life. Then maybe we can say that we have opened the black box. Until then, I think that given the present data that the evolutionary explanation for complexity is not only plausible but likely.

Complexity theory

Up to this point, I have set forth a scenario for the origin of irreducibly complex molecular machines in terms of a traditional Darwinian concept--pre-adaptation. However, I'm not entirely convinced that this Darwinian idea can do all the necessary work in explaining complexity. Now I will introduce and summarize work, mostly done by theoretical biologists and computer scientists, that I believe is very promising in getting us through some of the most difficult problems in evolutionary theory. I'm going to focus on the the work of Stuart Kauffman in The Origins of Order: Self-Organization and Selection in Evolution. His bold proposals account for the origin of body plans, the origin of diverse protein topologies, the origin of metabolism, the origin of developmental regulatory networks, and the origin of life itself. Note that this is a non-Darwinian explanation; I hope that the organizers will pardon my switch to the other side. However, the explanation is still evolutionary and does not appeal to special divine action.

The origin of life: a new view

Kauffman recognizes the difficulties of many of the origin of life scenarios that have been proposed. He recognizes the difficulty of both the protein-first and the RNA-first view proposals for the origin of life given the reality in modern cells that nucleic acid replication requires proteins and that protein synthesis requires nucleic acids. His proposal is that an emergent property of a sufficiently complex system of catalytic polymers is auto-catalysis. To quote Kauffman's own words

"..this new view, which is based on the discovery of an expected phase transition from a collection of polymers which do not reproduce themselves to a slightly more complex collection of polymers which do jointly catalyze their own reproduction. In this theory of the origin of life, it is not necessary that any molecule reproduce itself. Rather, a collection of molecules has the property that the last step in the formation of each molecule is catalyzed by some molecule in the system. The phase transition occurs when some critical complexity level of molecular diversity is surpassed. At that critical level, the ratio of reactions among the polymers to the number of polymers in the system passes a critical value, and a connected web of catalyzed reactions linking the polymers arises and spans the molecular species in the system. This web constitutes the crystallization of catalytic closure such that the syytem of polymers becomes collectively self reproducing...[this new body of theory] is also robust in leading to a fundamental new conclusion: Molecular systems, in principle, can both reproduce and evolve without having a genome in the fmiliar sense of a template-replicating molecular species."

In essence, Kauffman's view acknowledges irreducible complexity, i.e. the system has to be sufficiently complex in order for auto-catalytic behavior to emerge. An interesting feature of Kauffman's system is that there is no stepwise evolution of this emergent property; it suddenly appears once the polymer complexity has achieved the threshhold level. Thus, the system is complex and whole from the start. So while granting irreducible complexity, Kauffman has explained its origin.

The Origin of a Connnected Metabolism

Kauffman also tries to explain the origin of complex metabolic pathways that in their modern representation appear to be perfectly integrated and designed. The argument is similar to that given in the preceding section. Above, for a given polymer length, the number of possible molecules increased more slowly than the number of reactions by which they could interconvert. This led to the result that at some critical polymer length, with the concommitant complexity, that an autocatalytic system would emerge. Likewise, in a primitive metabolism, where catalytic polymers are interconverting metabolites, as the number of metabolites increases, the number of reactions by which they interconvert also increases. Eventually, as the system becomes sufficiently complex a connected metabolism will spontaneously emerge. In the middle of the discussion of this point, Kauffman makes the following critical observation:

"...we must envision, and need to test mathematically and eventually experimentally, that, under selection, the autocatalytic system will select out peptides, polypeptides, or ribozymes of increasing specificity and higher maximal reaction velocity and will also collect inhibitors of catalysis, thereby trimming away useless metabolic branches to leave a core coupled metabolism which is self-consistent: All needed tranformations have catalysts of ever-increasing specificity. A metabolism using a specific pathway to form a specific predetermined metabolite Z required for present life...has not evolved; rather, one of an enormous number of alternative coupled, consistent metabolisms has been focused on collectively by parallel simultaneous selection on the entire integrated, if initially imprecise, autocatalytic system. After the fact, once a subcritical specific metabolism and the requisite enzymes to catalyze it have been achieved by selective focusing, we stand amazed at its integration and complexity. We are thence led to calculate the probability that a set of polypeptide or RNA catalysts would jointly catalyze just that metabolism... [with a probability of 10<sup>-40,000</sup>]. It is the wrong calculation, the solution to the wrong problem. The details of my argument almost certainly are incorrect, but sure the invitation to consider the hypothesis that life began, not simple but imprecise and complex and then became simpler by selective focusing to more specific catalysts, deserves serious attention."

As I've reflected on Kauffman's approach, the following analogy has come to mind. It's not perfect, but it gets to the gist of the matter. Imagine a pile of rocks of lots of different sizes, say formed by some avalanche. Then by wind, water or other kinds of weathering, all the loose rocks are removed. What is left standing is the tightly knit complex structure where every piece is essential to the structural integrity of the whole. It might even appear as if all the rocks were carefully placed to give the final sturdy structure, i.e. that the whole structure was designed.

The application to a molecular machine like a cilium might go as follows: First, a collection of macromolecules is present that perhaps carry out other functions; some of these are incipient components of the molecular machine. Some of these molecules assemble in a fashion not determined by the final goal, but due to simple interactions such as those described in the hemoglobin dimerization step described before. You might imagine this to be our pile of rocks. With a given level of complexity a new primitive function might emerge from this assembly. Once the new function has emerged, selection operates to remove the loose and unnecessary pieces and to finely tune the interactions. As a result of this selective focusing, the system moves in a direction where each component plays an essential role and where each component is structurally and functionally interconnected with all the others; in other words, the system has become irreducibly complex.

This way of thinking makes an analytical solution to the complexity problem nearly impossible. In other words, since all that we are left with is the end result of the process, we may never be able to reconstruct the details of the original complex, but imprecise system. However, a synthetic approach is possible. We may not be able to produce the exact metabolism or auto-catalysis found in modern cells, but Kauffman's theory does allow experimental verification of some connected metabolism or some autocatalytic system that may emerge from a sufficiently complex collection of catalytic polymers.

Premature appeal to special divine activity to explain the world around us damages the Christian theistic apologetic

The "intelligent design" approach as manifested by the irreducible complexity argument is motivated in part by apologetic concerns. We reject the anti-theistic viewpoints that dominate much of modern science. We also believe, as the scriptures teach, that "the heavens declare the glory of God" and that God's "invisible attributes, His eternal power and divine nature, have been clearly seen, being understood though what has been made, so that they are without excuse." Thus, phenomena that cannot be explained in terms of natural causes or which have such a low probability of occuring by chance are taken to be evidences of special divine operation.

As I have already stated, I have no in principle objection to this kind of conclusion. However, I think that the scriptures have in mind a much broader application. The theistic apologetic ought to claim all of reality as evidence of God's eternal power and divine nature. Every fact of creation drips with the evidence of God as the creator. Every time we think or speak about a fact of creation, it is either acknowledging God as the creator or denying him. It is the unbelieving heart and the depraved mind that suppresses this truth. According to Romans 1, these things are evident, both in creation and in the human heart; we don't need some irreducible complexity argument from molecular biology or some probability calculation to see these things.

Narrowly defining our theistic evidences leads to another even more serious problem for theistic apologetics. In the progress of science once unexplainable phenomena sometimes are explained by natural causes. Putting so much weight on God's role in explaining things that we can't explain via natural causes, given the state of present-day science, leads to a minimizing of God's role if, and when, those things are explained. There are even theists who argue that if you can explain the origin and evolution of life by natural causes then God's explanatory role is nil and he is a superfluous addition to our explanation. In my opinion the decreasing impact of theism in the scientific marketplace of ideas stems more from a sometimes implicit, but often explicit belief that God's role is diminished as scientific arguments explain more and more of the world around us. This has been the problem with the past 250 years of science as theists and atheists alike have bought the argument.

The resurgence of the intelligent design argument may give a temporary respite to the eroding influence of theism in the sciences, but the gains will be short-lived. Although many in the design crowd are already cheering the demise of evolutionary theory, I think that there have been spectacular gains in nearly every area of biology and key new developments in the areas of complexity theory, developmental biology, and paleontology. These design arguments will give the general Christian public much ammunition to fight their misguided battles against evolutionary biology. Real gains in the fight against an atheistic naturalistic worldview will come only when we see that the battle is not concerning the details of some theory in biology, but is concerning the deeply rooted anti-Christian religious convictions that take the glorious truths of God's creation and twist them into an anti-Christian apologetic.

Christians need to see the revelation of our Creator Lord in every square inch of reality; we must counter unbelievers' denial of that revelation with the Biblical response that their denial is rooted in their suppression of deeply-rooted enmity with God. This is the basis for a truly theistic science; a science that sees the glory of God's creative and providential activity in every detail.


Send comments to