How Can Evolution Cause Irreducibly Complex Systems?

The Problem

Behe coined the name Irreducibly Complex for systems which would stop working if any of their components were removed. An example he gave is the biochemical system that makes clots in your blood. He said that this was like the common mousetrap, which becomes useless if you remove its spring.

Behe argued that such systems cannot evolve by a series of small modifications, each of which is a slight improvement to some initial system. His proof was that he did not know any plausible scenarios for their evolution.

Scenario #1: Reduction of Function (Improvements Become Necessities)

Let us imagine a species. Each animal has a workable biochemical system that just has component A. Then a mutation in one animal causes component B. B happens to make the system work better. B is optional, but it's valuable to the individual who contains this improved system. Darwinian natural selection will tend to spread the improvement, until eventually every member of the species inherits it. Now every copy of the biochemical system uses both A and B.

So far, the scenario has just added things. Behe assumed that all Darwinian scenarios are additive. But there is no reason why a mutation cannot remove a function. So, let us suppose a second mutation comes along. It removes component A's ability to do the job alone.

This mutation is not a problem to the individual which possesses it. Everyone has B, so the individual has B. The individual functions. The individual's chances of leaving descendants are not changed by the reduction-of-function mutation. To natural selection, the second mutation is neutral and is unselected.

It is statistically possible that eventually, all members of the population will inherit the second mutation. If it happens, this is called neutral drift.

And we're done. A and B are both essential, so the system is now Irreducibly Complex. Apply this same scenario again, and we will have a system in which A, B and C are all essential. Apply it again, and again, as many times as you wish. The system gets more and more complicated, and it probably gets more and more Rube Goldberg. And in fact Behe says, many times, that the Irreducibly Complex systems in the human body are pretty Rube Goldberg. Therefore, the systems that he describes are exactly the kind that this scenario produces.

Example: The Bolas Spider

It is also possible that a reduction-of-function be advantageous, rather than neutral. Here's a non-biochemical example where a reduction-of-function reduced a predator's costs.

The bolas spider mastophora uses an unusual prey-capture system. It spins only a single short strand of web, at the end of which is a drop of glue. Then it swings this strand around in circle underneath it. At the same time, it releases a scent - a pheromone - that attracts male Noctuid moths. A moth flies into the area, hits the glue, and sticks. The spider draws in the trapped moth and dines.

This system is Irreducibly Complex. Without the glue, the moth wouldn't get trapped. Without the web, moths that fly by (as opposed to landing by the spider) wouldn't get caught. Without the pheromone, there wouldn't be enough moths in the area. The web swinging behavior is reducible, though. There is a cave-dwelling fly larva which attracts prey with a light, instead of a smell. It has a single-strand web, but doesn't swing it.

The bolas spider's system could have evolved gradually. Imagine a spider with a normal sticky web. Add the pheromone. Lots of moths get caught, so there's no point in wasting effort on a great big web with lots of glue. The web is gradually simplified all the way down to one strand with one dot of glue. But the pheromone, which was just an improvement, gradually becomes a necessity.

Scenario #2: Loss of Scaffolding

A classical stone arch is irreducibly complex. Without the keystone at the top, it would collapse. Without the other stones, the keystone would fall.

So, how do humans build them? Well, we do it the same way we built the flying buttresses of cathedrals. We use scaffolding. The structure you see today is different from the original structure. Part of the original structure has been removed. A stone arch was not irreducibly complex until its scaffold went away.

Does nature work this way? Yes, over and over.


The first creature to have an air bladder did not depend on it for breathing. That species had gills, but perhaps they gulped a little extra air when they were in foul swampy water. Later there were a whole series of species which gradually depended more and more on their lungs, and less and less on their gills. Eventually, there were species that didn't need gills at all. If the gills (the scaffolding) then happened to vanish, it would be a further improvement for that species.

Another example are intestinal parasites. Genetic analysis recently showed one parasite to be a crustacean. However, it spends its life bathed in pre-digested food, and simply absorbs nutrients through its skin. This parasite is easily explained as just a small crustacean that lost features, such as its digestive tract. It still has some vestiges of a nervous system, which is of no use to it.

A lichen is really a fungus and an alga living inseparably. If either is taken away, the other dies. The system is irreducible. But it's quite easy to see how the symbiosis evolved. There are many places in the world where fungi and algae happily live alone. Lichen's trick is that it can live in inhospitable places. As the first lichen gradually became better at this trick, it gradually lost the need for its scaffolding, the nice hospitable environment.

Humans get sick if there isn't any Vitamin C in their diet. Dogs and cats manufacture Vitamin C inside their bodies. Many years ago, scientists explained this by saying that our ancestors ate a lot of fruit, which contains Vitamin C. One of our ancestors had a mutation which disabled the ability to make Vitamin C. Because of his (or her) diet, that was a neutral mutation, and it didn't harm them. By neutral drift, the mutation just happened to spread to all living humans. More precisely, it spread to all primates who were alive at the time, and all humans are descended from them. Gorillas are primates, so they also have to eat Vitamin C.

We can't make Vitamin C because a certain enzyme is missing. In the last few decades, scientists found the gene which animals use to create that enzyme. We also learned that a gene can be present, but turned off. Sure enough, we found that humans have a gene for that enzyme, but ours is turned off.

Scenario #3: Duplication

See Duplication Mutations.

Behe's Mousetrap Example

In his book, Behe argues that a mousetrap is irreducibly complex because it consists of five parts: a hammer, a spring, a catch, a holding bar, and a platform. He claims "if any of the parts are missing the trap does not function."

But if you press the holding bar into the board, you don't need the catch. If you mount the parts on the floor, you don't need the board. If you mount the bar vertically, you don't need a spring, since gravity will make the bar fall on the mouse. If you bend the "tail" of the spring slightly, you can fasten the holding bar under the tail, and not need the catch. (In his talks, Behe shows a large trap to audiences. Jeffrey Shallit made this last modification to Behe's trap, and showed it to him.)

In short, four of the five physical components can be eliminated. They are only there to make the trap more efficient. For nice clear diagrams, check John McDonald's web page. He ends with a clever mousetrap that is just a spring. And have a look at this nice diagram.

Creationist Don Stoner argued:

Behe's mousetrap challenge involved illustrating how a "primitive" mousetrap, involving a box propped up with a stick, could be improved by a succession of slight modifications until the modern trap was obtained. Each modification must be slight and each must represent an improvement over the previous design. If this can be shown, then the modern mousetrap does not demonstrate irreducible complexity in the sense Behe intends.

Starting with the box and stick, the trap could be improved, one small step at a time, by adding a tiny wooden base and enlarging it until it closed off the entire area under the box. Each increase in the size of the base would make it more difficult for the mouse to squeeze out under the edge or to burrow out.

The first tiny piece of wood could even be an unnecessary piece broken loose from the box itself. The box could be given this "extra" wood in small increments with each adding to the weight of the box - making it harder for the mouse to lift the box. When the first small piece is broken free from the box, it can do more good keeping the mouse from burrowing out than the small amount of incremental weight would have done.

Once we have increased the base until it covers the bottom of the box, we can further improve the trap by anchoring the box so the mouse can't slide it off the base. This anchoring can be accomplished by raising little ridges on the base on each side of the edge of the box which touches that base when the trap is armed. The ridges can be increased, improving their effectiveness, and even pinched together into indentations made into the box until a hinge is created - each step improving the anchoring of the box.

Next we can start to add a spring, one little bit at a time, starting from a mere whicker and building until the spring holds the box firmly down. Each incremental amount of spring strength does two separate things: 1) it holds the other edge of the box down more firmly and 2) makes the box drop more quickly, giving the animal less time to escape before the trap closes.

Once the box is spring loaded (instead of gravity) we can make the stick longer, tipping the box up higher, making the entrance to the trap larger and less threatening in appearance.

With the longer travel, the box comes down harder, striking more like a hammer. It becomes possible to pinch or crush the mouse and it is no longer necessary that the mouse be entirely within the area of the box for the trap to work. It also gives the mouse no time to work out an escape. The trap can be further improved by changing the box so it hits the mouse an increasingly larger percentage of the time. At some point it always hits the mouse.

Once we have a hammer trap instead of a box trap, the hammer can be changed until it has less air drag, making it close more quickly. The rectangular frame is all that is really necessary; as the mass of the box is decreased and the spring increased, the hammer works more quickly.

Now we can make the stick longer still until the frame is pushed back much farther, giving the hammer a longer and more powerful swing. At some point it becomes helpful to secure the top end of the stick to the hammer by producing a hinge as we did before. This reduces false triggers from the wrong end of the stick causing it to come loose. Connecting bait to the bottom end of the stick and gradually improving the connection until we have a trigger also improves the trap's effectiveness.

At this point we have a trap so much like the modern mousetrap that the final changes might seem scarcely necessary. Still, there appears to be some gain to be had from moving the hinge on the top end of the stick from the hammer to the base. If the hammer is pushed all the way back, this is a very short distance (although it is a quantum leap). This reduces the force the stick applies to the trigger and makes the trap spring more easily. Some designs of triggers will fail when this change is made, but others will not.

At this point we have produced the modern mousetrap from the box and stick by a sequence of very small changes, each change improving the effectiveness of the trap.

The final step appears the most "unlikely" (from a naturalistic position) and the most creative step in the process; still it is a very small physical change. If this step requires too large a conceptual change, the previous design, excluding this step, is still very much like a modern mousetrap.

I conclude that Behe's 5-part mousetrap is not "irreducible", and that it could be arrived at by stepwise "improvements" of simpler designs.

The point here is not mousetraps. The point is that Behe's style of argument is not as strong as it might seem to the casual bystander.

Should Such Systems Be Common?

If a system becomes Irreducibly Complex, then it becomes constrained. It is less free to evolve, because many of the possible changes will cause malfunction. Malfunction means that the individual containing the changed system dies, or at any rate is at some disadvantage. So, changes are a bit more likely to be weeded out of the gene pool by natural selection.

This means that Irreducibly Complex systems are on the average more resistant to evolution. This increased stability should make them more common, because ones that arise would stick around.


Behe said that certain biochemical systems could not be the result of evolution, and his proof was that he couldn't imagine the necessary steps.

However, evolutionary biologists solved this problem more than half a century ago. I have presented their understanding as three simple scenarios, and I gave real world examples of each one. Each example shows that step-by-step evolution can arrive at an "Irreducibly Complex" system. The scenarios explain why this can happen easily.

Last modified: 29 June 2000

Up to the "Darwin's Black Box" page.

Back to the Creation/Evolution page.

Email a comment.

Search this web site