Behe and Irreducible Complexity: Failure to Engage the Evidence


Evolution in the body is a documented reality, but how did the process for generating antibodies come about in the first place?

Intelligent Design proponent Michael Behe says scientists don’t have the faintest idea. In his 1996 book Darwin’s Black Box, he laments, “Although great strides have been made in understanding how the immune system works, we remain ignorant of how it came to be” (136). As evidence, he cites two brief articles (in his view, the field’s “best efforts”) before dismissing their conclusions out of hand. He continues, “We can look high or we can look low, in books or in journals, but the result is the same. The scientific literature has no answers to the question of the origin of the immune system” (138).

But is science really silent on this topic? So much blood, sweat, tears, and NIH money have been spent on the study of the immune system that a complete lack of answers about its beginnings would, as Behe suggests, mean doom for evolutionary theory. The truth is, according to Web of Science, even by 1996 hundreds of peer-reviewed papers had been published on the subject, each contributing a tiny piece to the overall puzzle.

Quick review: components of the antibody diversity system

Behe argues that three aspects of the immune system—clonal selection, antibody diversity, and the complement system—are irreducibly complex and pose “massive challenges to a putative step-by-step evolution” (138).

Because we have already examined how antibody diversity is generated, we will limit our discussion to the evidence for how this ingenious system might have arisen. Much more could be said about other aspects of the innate and adaptive immune responses.

Recall that the genome contains several clusters of gene segments (red and green in the figure below), each of which has tens to hundreds of members. In B cells, proteins encoded by two Recombination ActivatingGenes (RAGs; blue) join together one member from each of the clusters by excising the DNA between them. The genome of each B cell is thus irreversibly altered in a unique way, depending on which segments are joined, such that the recombined gene segments code for the antigen-binding site of that B cell’s antibody.

RAG1 and RAG2 can’t bind just anywhere on the DNA; they recognize special sequences called Recombination Signal Sequences (RSSs; orange and yellow). RSSs are found flanking each gene segment, similar to special cues we use in grammar like capitalized words at the start of sentences and punctuation marks at the end. The RAG proteins home in on two randomly-chosen RSSs, bring them physically close to one another, and cleave the double-stranded DNA at both gene-RSS junctions. DNA repair machinery then repairs the break, joining the two gene segments together and the two RSSs together. The closed loop of DNA containing the RSSs gets removed, while the recombined gene is now ready to code for an antibody..

Behe’s mistaken assumptions

Behe argues that a minimally functional antibody diversity system needs three components: the antibody genes themselves, start and stop signals (like RSSs), and machinery to cleave and rejoin the DNA at the signals (like RAG and the DNA repair proteins). He can’t imagine how a multi-component system could have arisen by a gradual process, because each component is dependent on the other two for the whole shebang to work.

From the start, Behe makes the faulty assumption that antibody receptors incapable of recombination would be useless. He writes:

A primitive system with only one or a few antibody molecules would be like the propeller turning at one revolution per day: not sufficient to make a difference… Because the likelihood is so small for the shape of one antibody being complementary to the shape of a threatening bacterium—perhaps one in a hundred thousand or so—an animal that spent energy making five or ten antibody genes would be wasting resources…” (130-1).

What Behe fails to recognize is that many, many receptors in the immune system do their jobs without gene arrangements. These receptors bind to molecules commonly found on the surface of harmful microbes. In fact, some 90% of animal species on the planet don’t even have adaptive immunity, so antibody production by a gene rearrangement mechanism cannot be imperative for life (though humans and other vertebrates are quite dependent on it now). Contrary to Behe’s assumption, the first antibody genes could easily have had useful functions without RSSs and RAGs.

A family of molecules called the Toll-like receptors (TLRs) demonstrates the utility of having an all-purpose microbe detector that does not require millions of randomly-generated variants. TLRs, located on the surface of special immune cells in the blood, recognize bacterial cell walls and virus-specific DNA sequences, causing an all-out attack by the body on the foreign invader. In the process, some of the host tissue gets destroyed, but this collateral damage is a necessary cost to slow down the infection.

This so-called innate immune response—the first line of defense—occurs immediately upon infection, while antibodies take several days to produce. Without innate immunity, the animal might die before antibodies even have a chance to work. The fact that virtually all multi-cellular organisms have TLRs indicates how critical they are to survival.

If innate immunity is so effective that 90% of animals live just fine without adaptive immunity, it’s natural to wonder why some animals do have it. A major advantage it provides is an immunological “memory” of past infections, making it easier to fight off similar pathogens in the future. (Vaccines work on this principle—by exposing the body to inactivated or dead viruses, we give B cells a “heads up” so they can make and store antibodies before the real thing hits.) Antibodies also enable targeted killing of the pathogen, preventing further damage to the host by the non-specific innate immune response.

There’s another way to ask why some animals have an adaptive immune response: rather than seek to explain what added function or advantage it serves, we can ask about the mechanism by which it came to be in the first place.

#Why Irreducible Complexity Fails

Intelligent Design proponent Michael Behe popularized the idea that some biological features are irreducibly complex. He defines an irreducibly complex system as one that is “composed of several well-matched, interacting parts that contribute to the basic function, wherein the removal of any one of the parts causes the system to effectively cease functioning” (Darwin’s Black Box, 39).

No problem there. Behe simply builds upon Aristotle, who wrote in his famous work Metaphysics, “The whole is more than the sum of its parts.” Every scientist accepts this principle.

Detecting design through irreducible complexity?

Behe goes further, claiming to have demonstrated that natural processes are insufficient to produce irreducible complexity:

The laws of nature can organize matter…If a biological structure can be explained in terms of those natural laws, then we cannot conclude that it was designed. Throughout this book, however, I have shown why many biochemical systems cannot be built up by natural selection working on mutations: no direct, gradual route exists to these irreducibly complex systems… (203).

If no gradual route exists, so the logic goes, then irreducibly complex systems must have been designed:

For discrete physical systems—If there is not a gradual route to their production—design is evident when a number of separate, interacting components are ordered in such a way as to accomplish a function beyond the interacting components (194).

Legions of scientists have rejected this argument. Why? Is it because they are godless atheists who deny the existence of an intelligent Creator? No. Some of Behe’s strongest critics are deeply committed Christians. These scientists simply see overwhelming evidence that irreducibly complex structures and systems have developed gradually through natural, evolutionary processes. For the believer, these processes are simply God’s chosen means of providentially ordering the world.

Darwin’s Black Box contains an entire chapter about the irreducible complexity of the immune system. Behe argues therein that the system for generating antibody diversity could not have developed by a gradual, stepwise process and therefore must have been designed. Here we examine some evidence to the contrary.

RAG review

Recall from Part I that the antibody production system has three interdependent components, as shown in Figure 1: 1) clusters of gene segments that can be combined in different ways to code for antigen receptors (red and green), 2) start and stop signal sequences between these gene segments, called RSSs (orange and yellow), and 3) machinery to cut and rejoin the DNA at the RSSs (blue). ). (Also necessary are DNA repair proteins, which rejoin the severed DNA strands, but they are not specific to this process and are thus not included in the diagram.)

The RAG genes, which carry the instructions for how to make the RAG proteins, are critical for antibody gene recombination. Surprisingly, these genes are found in jawed vertebrates, but missing in jawless vertebrates and invertebrates. Thus, jawed vertebrates can make a vast array of antibodies, but the other 90% of species on the planet cannot. What could account for this seemingly arbitrary division? We have very good evidence that at least 450 million years ago, an ancestor for all jawed organisms acquired the RAG genes. Where might those genes have come from?

Transposons on the move

To answer that question, I need to first tell you about a strange and fascinating feature of bacterial DNA. Bacteria have short segments of DNA, called transposons, which can jump from one location to another in the genome (See Figure 2). By itself, DNA can’t do anything—it just provides the information the cell needs to make proteins. Proteins do all the mechanical work in the cell, including interpreting the instructions in DNA. Transposons jump around by carrying instructions like this:


The cell’s protein machinery reads the message and promptly makes the “scissors and glue” proteins. The scissors break the DNA at the “cut here” sequences (blue regions), and the glue pastes them back into the genome somewhere else. Where the transposon gets pasted is indeterminate—it may land within a gene, disrupting its protein-making instructions, or between genes, as shown in Figure 2.

A possible stepwise route to an irreducibly complex system

How does this relate to antibody gene recombination? Well, the RAG genes, which do the cutting to make functional antibody genes, are believed to have come from a transposon. The working model for how jawed vertebrates came to have antibody genes capable of recombination is known as the “transposon hypothesis.”

According to the hypothesis, a RAG-containing transposon invaded the genome of an early jawed vertebrate, probably through a bacterial infection. Such an infection would have occurred around 450 million years ago, shortly after the jawed and jawless vertebrate lineages diverged from their common ancestor.1

Suppose the transposon became inserted within a pre-existing, non-rearranging antibody gene (see Figure 3, step 1). (At this point in history, both vertebrates and invertebrates already had perfectly useful antibody-like receptor proteins—they just didn’t rearrange.) The inserted transposon sequence would have the effect of splitting the gene into two segments.

When the cell began to read the host gene, the bacterial RAG genes were interpreted instead. The resulting RAG proteins snipped out the transposon and pasted it into a new location in the genome (step 2). Over time the RAG genes became immobilized in the genome, never to jump again. The RAG proteins still recognized the RSSs, though, and continued to excise the DNA between them. Now, instead of cutting out the transposon, they cut out the short stretch of DNA separating the two halves of the original antibody gene (step 3). Thus, the gene segments were reunited and voila! The antibody gene could make protein again.

Suppose that in a future generation, the gene segments duplicated. This would create more gene segments, allowing some limited recombination to take place. Suppose that later still, the whole set of gene segments underwent duplication. Successive rounds of duplication and divergence would lead to the diversity of gene segments we see today (step 4).

But is there any evidence it happened this way?

This sequence of events might sound like an evolutionary “just-so” story, but it actually rests on solid evidence. First, the RAG genes themselves bear striking resemblance to known bacterial transposons. Most immunologists think the similarity is too great to have occurred by coincidence. The RAG1 and RAG2 genes are also physically very close to each other in jawed vertebrates. This would be expected if they arrived together from an external source like a transposon. In addition, the signal sequences (RSSs) to which the RAG proteins bind are closely related to those found associated with bacterial transposons.

Perhaps most convincing: the RAG proteins can not only excise DNA, but insert one stretch of DNA into anotherin vitro (in a test tube). The RAG proteins don’t normally do this latter step in cells; there they are only used for DNA cleavage. In bacteria, transposon proteins do both, and in the test tube, the vertebrate RAGs can do both! How could the RAG proteins function just like those from a transposon, unless they had originally been part of one?

Irreducible complexity by a gradual route—God’s route

Behe argues that design can only be detected when no possible stepwise route to an irreducibly complex system exists. Here I have described such a route (and a well-supported one at that) by which the antibody diversity generating system—one of Behe’s best examples of irreducible complexity—could have developed gradually. We’ll examine his response to the data next time.

To me, the fact that the transposon hypothesis has passed though decades of scrutiny and experimental testing leads me not to question God’s design of the adaptive immune system, but to marvel at his ability to create it gradually through a long but ultimately fruitful process. Immunologist Ronald Plasterk reflected poignantly, “We may owe our existence to one transposition event that occurred 450 million years ago.” In my mind, while it may have occurred by entirely natural processes, the evolution of our immune system was no accident. The fact that we can use science to read this small story from God’s book of nature is a special gift indeed.

#Failure to Engage the Evidence

Many non-scientists are unaware of the importance or extent of “the scientific literature.” In the geeky world of science, success doesn’t depend on social prowess or earning power (thankfully!), but on one’s publication record. It’s hard to get published—to do so means making a significant contribution to the field, as judged by a panel of anonymous reviewers. Most people have nightmares about being chased or drowning. Scientists have nightmares about their papers getting rejected.

The main database for accessing the biomedical research papers, called PubMed, contains over 19 million articles. If you search PubMed for irreducible complexity, you’ll get 11 hits, none of which represents a peer-reviewed piece of work showing evidence of an Intelligent Designer. Most of the hits are commentaries on the Intelligent Design (ID) movement itself or attempts to “reduce” irreducible complexity. A couple of them are totally irrelevant.

Could this mysterious absence from the literature be a result of a conspiracy by Big Science to deliberately marginalize the work of people like Michael Behe? Though ID proponents often try to make this case (just watch Ben Stein’s Expelled), I think it’s very, very unlikely. But let’s give them the benefit of the doubt. If ID theory is scientific in nature, we can expect its leaders to engage scrupulously—as all scientists must—with the literature.

Unfortunately, Michael Behe does not have a great track record in this department. Consider his 1996 book Darwin’s Black Box. When reading the chapter on the immune system, for example, I kept expecting him to interact with the evidence for the gradual development of the antibody recombination system. The transposon hypothesis, which we examined in detail in Part 2, is a well-tested model for how this system could have originated. I couldn’t wait to see what Behe would say about it! His response was more than a little disappointing. He does refer to the model, at least indirectly, but only by discounting a two-page commentary—the purported “best efforts” in the field—as mere speculation:

[The authors] make a valiant stab at accounting for the components [of the antibody diversity generation system], but in the end, it is a hop in the box with Calvin and Hobbes. The authors speculate that a gene from a bacterium might have luckily been transferred to an animal. Luckily, the protein coded by the gene could itself rearrange genes; and luckily, in the animal’s DNA there were signals that were near antibody genes; and so on. In the final analysis the authors identify key problems with gradualistic evolution of the immune system, but their proffered solutions are really just a disguised shrug of their shoulders (137).

Seriously, a hop in the box with Calvin and Hobbes? If you search Pubmed for RAG transposon, you’ll get 18 hits, and they’re all relevant to the evolution of the immune system. More papers deal directly with the data for the transposon hypothesis than for the much broader topic of irreducible complexity! The top two hits are recent reviews, each of which cites dozens of papers. And the first paper ever to propose the transposon hypothesis (Sakano et al. Nature 1979, which doesn’t even appear in this narrow search) has been cited hundreds of times in peer-reviewed publications.

Questions certainly remain about the details of the transposon model, but Behe makes it sound like a pie in the sky idea with no supporting data. He writes, “We can look high or we can look low, in books or in journals, but the result is the same. The scientific literature has no answers to the question of the origin of the immune system” (138). This is patently false. Behe may not like the answers, but he misleads the public when he suggests they don’t exist.

Nearly 15 years have passed since Darwin’s Black Box first came out. During that time, scientists have collected some of the most compelling evidence for the transposon hypothesis (see Part 2). What has Behe said about this new data?

In 2005, he testified in a now-famous Dover, Pennsylvania, courtroom in favor of teaching ID alongside evolution in public schools. Behe maintained his previous position: “The scientific literature has no detailed testable answers on how the immune system could have arisen by random mutation and natural selection.”

The judge, John E. Jones, wrote in his decision that Behe “was presented with 58 peer-reviewed publications, nine books, and several immunology textbook chapters about the evolution of the immune system; however, he simply insisted that this was still not sufficient evidence of evolution…” Jones ultimately ruled against teaching ID in classrooms, in part because of the impossibly high burden of proof Behe demanded. In 2006, the 10th anniversary edition of Darwin’s Black Box appeared, featuring a new afterword. While Behe cleared up a few apparent misconceptions about irreducible complexity, he stood by the original text as a whole:

Despite the enormous progress of biochemistry in the intervening years, despite hundreds of probing commentaries…,despite implacable opposition from some scientists at the highest levels, the book’s argument for design stands. Other than updating the list of my children in the Acknowledgements…there is very little of the original text I would change if I wrote it today (255).

Specifically about the immune system, Behe mentioned one new paper from 2005, but only to reemphasize that nothing in his thinking had changed:

Whatever interesting things [the 2005 paper] speculated about gradual evolution, however, it had nothing to say about Darwinian evolution. In fact neither Darwin’s name nor any derivative word appeared in the paper. Nor did the phrase “natural selection” appear; “selection” is used once.” “Mutation” appears twice, but the envisioned mutations are not specified (269).

This sort of defense is peculiar. Behe counts words when he should be grappling with the data. Interestingly, he implicitly admits that a gradual evolutionary process may be sufficient to explain the origin of the irreducibly complex immune system, but he contrasts such a process with Darwinian evolution, by which he means mutation and natural selection.

Recall the quote I referenced in my last post, in which Behe lays out the criteria for detecting design:

The laws of nature can organize matter…The most relevant laws are those of biological reproduction, mutation, and natural selection. If a biological structure can be explained in terms of those natural laws, then we cannot conclude that it was designed. Throughout this book, however, I have shown why many biochemical systems cannot be built up by natural selection working on mutations: no direct, gradual route exists to these irreducibly complex systems… (203).

The transposon hypothesis outlines a possible gradual route to an irreducibly complex system, and it relies entirely on the laws of reproduction, mutation, and natural selection. Nevertheless, Behe remains unconvinced. In his mind, to explain the origin of an irreducibly complex system entirely in terms of natural laws, one would have to show that the steps were in fact unguided.

Such a requirement can be seen more clearly in his 2009 letter to the editor at Science, written in response to a news focus article about the origins of the immune system. Science chose not to publish the letter, so Behe posted it on the ID blog Uncommon Descent. He wrote:

Darwin’s chief contribution was not the simple idea of common descent, but the hypothesis that evolution is driven completely by ateleological mechanisms, prominently including random variation and natural selection. Intelligent design has no proper argument with the bare idea of common descent; rather, it disputes the sufficiency of ateleological mechanisms to explain all facets of biology…

Many scientists agree with Behe that evolution may have been guided in some mysterious way by a Mind. But whether or not the methods of science could ever rigorously detect teleology—mindful purpose—by studying the physical world is hotly debated. Most working scientists I know do not believe science is equipped for such a task.

Questions about teleology are fascinating, but today I’m concerned with just one thing: whether or not Behe has sufficiently engaged with the scientific literature. If I had read his book without any prior knowledge of immunology (or the other topics he covers), I’m pretty sure I would be left with a deep distrust of scientists and the scientific process. I think that’s a pity.

Notes & References

Table of Contents

Kathryn Applegate
About the Author

Kathryn Applegate

Kathryn Applegate is Resources Editor at BioLogos. While working on her PhD in computational cell biology at The Scripps Research Institute (La Jolla, CA), she became passionate about building bridges between the church and the scientific community. In 2010, she joined the BioLogos staff where she has the privilege of writing, speaking, and working with a wide variety of scholars and educators to develop new science and faith resources. Kathryn co-edited with Jim Stump How I Changed My Mind About Evolution (InterVarsity Press, 2016). As Project Leader for BioLogos INTEGRATE, Kathryn is excited about the potential for BioLogos to equip parents and teachers to raise up the next generation of Christian students who approach science with wonder, curiosity, perseverance, and wisdom. Kathryn and her family enjoy exploring the beaches and state parks of Michigan and are helping to plant a new church in the Presbyterian Church of America (PCA) denomination.