Skip to content

Revisiting Irreducible Complexity

September 14, 2011
Diagram of the bacterial flagellum

Revisiting Irreducible Complexity

For those of you who are new to the discussion of biological origins: irreducible complexity (IC) is a term coined by Michael Behe, Professor of Biochemistry at Lehigh University, in his 1996 book Darwin’s Black Box. Professor Behe defines IC in this way:

     “By irreducibly complex I mean a single system 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.”

For example, the scissor is an irreducibly complex system, made up of 3 basic components: two distinct cutting blades and a needle that attaches the two ends of the blades together. If just one of any of these parts were missing, the scissor would “effectively cease functioning” as a scissor. Behe argued that irreducibly complex biological systems present a problem to Darwinian evolution because any pre-cursor state is non-functional, and any plausible evolutionary pathway must consist of a number of functional steps, each step building on the last. Natural selection cannot favor non-functional systems. Thus, it seemed to Behe that IC biological systems would have to arise in “one fell swoop” where all the components of the system fall into place by chance, producing a functional system.

Irreducible complexity can further be illustrated by the following:

A + B + C = function X

In this example, the components A, B, and C are all necessary for function X. Loss of any one of those components results in a loss of function X.

However, there is a problem with the view that irreducible complexity is always an obstacle to Darwinian evolution. Nothing in the definition of irreducible complexity suggests that components A, B, and C can only perform function X. For example, component A could perform function Y, and component B could perform function Z, and component C could perform function W. And if components A and B are added together, perhaps we get function F. Natural selection has something to favor now, and function F is spread throughout the population. Next, component C associates with function F, and when this happens, we get function X.

This idea – the idea that components can shift their function – is called cooption or exaptation. So it looks like Behe’s thesis that IC systems cannot plausibly evolve has been refuted. Or has it?

The first point I wish to make here is that IC actually is a useful concept in biology. Even if cooption could possibly explain the evolution of IC systems, IC is still a useful concept in biology because it can help us identify the possible pathways for the evolution of a given biological system. If a system has a large number of components, and that system is IC, then this means that the only plausible evolutionary pathway for the evolution of that system is one consisting of cooption events. In other words, the standard observational evidence for Darwinian evolution – the lengthening of the giraffe’s neck, the Galapagos finches, peppered moths – does not provide any evidence whatsoever that IC systems can evolve. In the example of the lengthening of the giraffe’s neck, we start with a population of short-neck giraffes, which, over time, evolve longer necks. This is evolution along a single route. On the other hand, the evolution of an IC system must consist of cooption events, where components must shift their current function and adopt a new function. In short, while the evolution of the giraffe’s neck involves a direct Darwinian pathway, the evolution of a truly IC biological system can only occur through an indirect Darwinian pathway.

Now, one of the most well-known supposed examples of irreducible complexity is the bacterial flagellum. Some have pointed to the type III secretory system (TTSS), a functional subset of flagellar components – but it has a function other than bacterial motility – and they claim that the TTSS disproves the idea that the flagellum is irreducibly complex. However, this argument does not hold because irreducible complexity, as originally stated by Behe, simply means that removal of a core component of a system results in a loss of the original function. It does not mean that all possible biological function is lost. It merely means that the original function is lost, which means that that system, if it evolved, must have evolved through an indirect, circuitous Darwinian pathway.

The second point to be made is that many (if not most) molecular systems are irreducibly complex to some degree. A two-component molecular machine is irreducibly complex if removal of any of those two parts results in a loss of the original function. But two-component molecular machines can plausibly evolve, so it seems as if irreducibly complex systems can evolve, and also that cooption provides an actual Darwinian mechanism for the evolution of IC systems.

However, I will argue that (a) some irreducibly complex biological functions can plausibly evolve, (b) the higher the level of irreducible complexity in a biological function, the less plausible it is for that function to evolve.

Here, I will emphasize a particular point: that some irreducibly complex structures can plausibly evolve through Darwinian mechanisms. In other words, if we find any ole’ molecular system, remove one of its components, and find that the original function is lost, intelligent design proponents should not immediately conclude that Darwinian evolution cannot plausibly account for the origin of that system. This is because there are literally hundreds of molecular functions out there, and most of them are irreducibly complex to some degree. Yet Darwinian evolution can account for the origin of a large percentage of irreducibly complex molecular machines. The crucial point to digest is that Darwinian evolution can account for the origin of systems with a low level of IC but the higher the level of IC of a given system, the less likely it is to have evolved through a Darwinian mechanism.

 

The level of IC is determined by various factors, but I will discuss that in a later article. For now, we will limit our definition of the level of IC to just one factor: the minimal number of components a given function requires in order for that function to be maintained. If function Y requires a minimum of 2 components in order for that function to be maintained, and if function X requires a minimum of 4 components in order for that function to be maintained, then function X has a higher level of IC. This is a fairly simple concept to grasp.

Recall that earlier I argued that “the higher the level of irreducible complexity in a biological function, the less plausible it is for that function to evolve.” Here’s the rationale behind that train of thought:

If function Y requires a minimum of 2 components in order to function, then this means that in order for this function to evolve 2 different parts – each performing a different function – must be coopted to form function Y. However, if function X requires a minimum of 4 components in order to maintain its function, then this means that, for this function to evolve, 4 different parts (each with different functions) must be gradually coopted to eventually form function X. Obviously, the evolution of function Y is much more plausible than the evolution of function X. While function Y requires that just two proteins have complementary structures in order to form the two-part IC systems, function X requires that four proteins have just the right structures so that they complement each other and are coopted to form a novel IC system.

Ultimately, the cooption scenario is based on chance. Natural selection does play a role, but the primary mechanism is chance. This is because in the cooption scenario: (a) chance alone determines that proteins currently performing vastly different functions will have just the right shape so that they can fit snuggly together with other proteins (that also just happened to have just the right shape) and form a novel IC structure; (b) it is primarily chance that determines that these proteins will happen to associate with each other to form the novel IC structure. Natural selection only kicks in once these proteins have associated with each other to produce new function.

Thus, when the cooption scenario is invoked to explain how systems with a high level of IC might evolve, it should be used with caution.

Conclusion

Irreducible complexity is a real biotic phenomenon. Darwinian evolution can account for some irreducibly complex structures, but the higher the level of IC of a biological system, the less plausible it is for that system to have evolved through Darwinian mechanisms.

The main points to digest are as follows:

  • Traditional evidence for Darwinian evolution (Galapagos finches, instances of speciation, etc.) does not constitute evidence that IC systems can plausibly evolve.
  • The more and more components a given biological function requires, the less likely it is for such components to have just the right shapes to fit snuggly with other components to form a novel biological function.
  • IC does not mean that a system is “too complex to have evolved.” This is a point that needs emphasizing, since many Darwinians either are ignorant of the true definition of IC or deliberately distort it.
  • Non-teleological cooption does not automatically account for all IC systems, as the foundation of the cooption explanation is rooted in chance.

In future articles, I will be writing on how we can measure the level of IC of a system, and how we can use the concept of IC to guide research in biology.

Advertisements
4 Comments leave one →
  1. Monsters From The ID permalink
    November 26, 2011 9:19 am

    As you said, a two part IC system is far more likely to be able to evolve by cooption, than a four part IC system, which is far more likely to evolve than a twenty part IC system, and so on…

    At a certain threshold (different for each IC system) would not there come a point at which it would be considered too improbable for an IC system with that many parts to be able to evolve, given the limitations of the time available, population size, and how often variations can take place?

    I should say YES. It’s true for the same reason that it’s far more probable for a throw of two dice to come up all ones, than it is for a throw of 20 dice to come up all ones. If anyone is instinctively answering NO to that question, then they ought to have some fairly good evidence to back that claim up.

    For very simple IC systems consisting of 1 to 3 parts, for which the probabilities of evolution by cooption are likely high enough to be considered feasible, then one could perhaps adjust the terminology from “irreducibly complex system” to “irreducibly simple system”, if not for the simple reason that a two or three part system wouldn’t normally be considered to be very complex. Regardless of the complexity of an irreducible biological system, the term “complex” exists as an adjective within the term “irreducibly complex system” whether or not the irreducible system is complex, or simple.

    What is at issue are two separate things – (1) whether a system has the property of irreducibility and (2) whether that system is complex, or simple. A system can be complex but not irreducible, simple but not irreducible, irreducibly complex, or irreducibly simple. Where the threshold of simple versus complex resides is arbitrary, but I would argue that since this topic has special relevance in the overarching subject of origins, then perhaps simple and complex thresholds could be set at the point at which the origin can be explained naturally, say as the result of evolution by cooption, or as the result of something else, say intelligent design. We could consider that each threshold for each irreducibly complex (or simple) system, resides at “The Edge of Evolution” as Michael Behe might say.

    If one believes that there is a threshold (different for each particular IC system) beyond which the evolution by cooption of such a system is “too unlikely”, then on the other hand, one should also be prepared to accept also that below such a threshold, then the evolution via cooption for such an irreducibly complex (or irreducibly simple (IS) system ought to be considered to be “likely enough”.

    It is the job of science to determine quantitatively (for particular IC systems) where the thresholds are likely to be.

    When such calculations are eventually performed, (and some fairly crude estimates have been made in a few cases) I say that it will become more apparent why it is that some in the science community advocate for the recent concept of “the landscape” or “multiverse”. It is because the old idea that “time becomes the hero of the plot” is now no longer sufficent to explain arbitrarily improbable events, since the amount of time now known to be available (billions of years at a maximum) is not enough to explain the evolution of the things that need to be explained. Some scientists seem to be realizing that this might be a problem, and are now moving on to the idea that “the vast numbers of alternate universes available, becomes the hero of the plot…”

    The problem being that such justifications are as lacking in scientific evidence as any other religious ideas are as well. They are philosophical assumptions, no longer a part of verifiable science.

    Since the typical objection made against any form of belief in intelligent design, or some form of creationism, is that such arguments must be excluded in a scientific context because they are said to be “outside of verifiable science” or “religious” in nature, I wonder how such arguments will continue to be waged when it is realized that naturalists also need to make use of concepts that seem to be (for the time being at least) also beyond the reach of verifiable science?

    What will happen is likely what usually happens. People will decide to believe in the concept of origins that seems more personally appealing to them.

  2. November 27, 2011 4:29 pm

    Hey there,

    I think one of the biggest problems to non-teleological evolution of IC molecular machines is that the cooption scenario is essentially rooted in chance (scaffolding and Elimination of Functional Redundancy can explain the origin of some IC systems, but as Mike Gene explains, they fail as a general solution to the origin of IC molecular machines whose functions require the interactions of multiple components). Natural selection only kicks in when two protein components have fortuitously associated with each other resulting in a novel, advantageous function. When one considers just how many different protein combinations there are – the vast majority being functionless – the problem is difficult indeed.

    My own personal favorite response to those who invoke multiverses to explain the origin of IC molecular machines is that those who oppose universal common ancestry can turn this around and say, for example, that given enough time, the patterns of a nested hierarchy in DNA/protein sequences will appear, and so this pattern cannot be used in support of common ancestry under the multiverse concept. Also, if we detected a metallic triangle floating in space, we would conclude it was the work of an intelligence or intelligences. Nevertheless, the multiverse concept would say that given enough time such a structure could appear through chance, thus it is not evidence of intelligence.

  3. November 30, 2011 9:11 am

    The discussion about probabilities assumes totally independent components coming together. There could be a number of sub-parts, also evolved that have specific functions, that are coopted.

    If anything, though there are elements of chance, it’s the significance of natural selection that means that what was chance at one level (time) becomes a certainty at the next level. Once a two-part element is selected for, then it exists, and chance only plays a part to the extent that any other part might fit. It is not as if all parts must be evaluated as a combinotorial probability. The very act of natural selection is biasing combinations already.

    Suppose A + B + C = X would have been a great biological function, but natural selection had selected against B. Then X would not occur. But A + D + C = Y might subsequently occur. With hindsight IDers would then be telling us how unlikely Y was because of the IC of A + D + C. The probabilist arguments about levels of IC are nonsense.

    Note also the double requirement inherent in this piece:

    “…that proteins currently performing vastly different functions will have just the right shape so that they can fit snuggly together with other proteins (that also just happened to have just the right shape)”

    The “that also just happened to have just the right shape” is implying that there is an extra degree of improbability here. What you are basically saying is:

    “A happens to fit B, and, extra improbably, B happens to fit A.” No. A and B happen to fit together is sufficient to describe the connection. What would it mean to say that A fits B, but B doesn’t fit A?

    There’s also the presumption of teleology creeping in too:

    “…function X requires that four proteins have just the right structures so that they complement each other and are coopted to form a novel IC system.”

    Function X doesn’t require anything at the outset. It is because your four proteins have certain structures that X appears at all. X isn’t waiting to be invented by the unlikely appearance of four particular structures, and nor is there some ‘design’ plan for X.

    A big lottery win (UK) is about 14-million to one. When someone wins you say “Hey! It must have been ‘designed’ that this number appear, rather than any other, because it is so unlikely that it can’t be down to luck.” Within a space where certain events *can* happen, that they actually do should be no surprise. You are neglecting all the other possibilities that could also have happened, but didn’t. You are not considering the full space of possibilities. You are using hindsight to spot what you think is design.

    The collosal variability in biological constructs available to evolution to act on is virtually immeasurable, and practically so. That the evolution of complex systems appear to be incredible to some of us is a measure of our incredulity rather than an explication of what can evolve.

    These arguments are replayed, and refuted, so often. Try this:
    http://scienceblogs.com/evolutionblog/2011/11/twenty_years_after_darwin_on_t.php

  4. jottain permalink
    October 23, 2012 10:51 pm

    “A big lottery win (UK) is about 14-million to one. When someone wins you say “Hey! It must have been ‘designed’ that this number appear, rather than any other, because it is so unlikely that it can’t be down to luck.””

    The results in lottery are somehow determined in the rules of the game. For example if the number “765” would appear, it really were very unlikely thing, because according to rules of the game it is not even allowed number. But what is probability for number 765 to appear in lottery? It is impossible to calculate beforehand. According to known rules it is 0. But in fact, there is a small possibility that someone will some day do a lottery ball that is against the rules, and it will be part of the game, and it will appear. In the game of life the situation may be similar. You can see that many things have somehow happen, but you don’t know what exactly happened, and you cannot measure their probabilities.

    “Within a space where certain events *can* happen, that they actually do should be no surprise.”

    Yes. And design is among those possibilities that can happen. Thus it should not be surprise for anyone, if it actually happens…

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: