In recent years, an “Intelligent Design” movement has emerged, claiming that the scientific community has neglected the possibility that the features of living things are the product of intelligent design. Although the intelligent-design hypothesis is sometimes viewed as a Trojan horse for introducing certain theological hypotheses into science, intelligent design can be performed by entities that are both nonhuman and nontheological, and the scientific community has actually dealt with several important examples of these.
I will discuss two main sets of these examples: various animal species and some important proposed evidence of extraterrestrial visitors. In both cases, the verdict has been overwhelmingly negative, with only a few exceptions in the first case. But how this verdict was reached suggests serious difficulties in the recognition of intelligent design–difficulties which the intelligent-design movement seems unwilling to consider.
Various animals construct a variety of structures, raising from simple burrows and nests to much more complicated nests, webs, and dams, and some sorts of animal behavior do seem very intelligent. And some scientists have, in fact, proposed that many animals are capable of performing intelligent design. The leading example is George Romanes, who published a book, “Animal Intelligence,” in 1888, collecting numerous anecdotal examples of that. However, later students of animal behavior have mainly cited Romanes’s work as an example of how not to do animal-behavior research, and they have concluded that most animal species lack any capability to perform intelligent design. In fact, they often call views like Romanes’s “anthropomorphism.” We now turn to some examples.
Members of some species of spider make elaborate orb webs; such webs look as if they had been intelligently designed by their builders. It is easy to show that such a web cannot arise by chance. Also, such a web is not some simple pattern, such as a crosshatched pattern of horizontal and vertical strands. Samuel Zschokke has some good pages on spiderweb construction: the spider first lays down some frame strands, then the radial strands, and finally a spiral “capture” strand.
An open-and-shut case for spiders performing intelligent design? Have spiders successfuly passed through William Dembski’s Explanatory Filter? Not quite. Spiderweb construction is very stereotyped–each species has only one preferred architecture of web. For example, members of some species of spider lay down an additional temporary spiral strand before laying down the capture strand, while members of some others do not. And while most orb webs are vertical, some are horizontal. Many further details can be found in several pages at the International Society of Arachnology site.
Furthermore, spiders can build webs by following some simple algorithms, as described in the work of Thiemo Krink. His work is a good example of artificial-life research; his software creates a simulated spider that looks for nearby web strands, and using what it finds and simple rules to decide where to move next. This spider has four sets of these rules, one for each phase of web building: the frame strands, the radial strands, the temporary spiral strand, and the capture strand. The rules have various associated parameters, which were improved with the help of genetic algorithms, a computerized form of evolution by natural selection. The webs were optimized for maximum prey capture and minimum total strand length, and the resulting webs look much like real spiderwebs.
We now turn to some other famous arthropod architects. The best-known of these is the honeybee, known for building honeycombs with precise hexagonal lattices of cells–in the complete darkness of their hives. How might some bees build such a remarkable shape of comb without having an overall view of them? The answer lies in that remarkable regularity. Each cell has neigbors at the same distances and angular separations, so if a bee can build each new cell at some appropriate position relative to some existing cells, all those cells will have that remarkable regularity of pattern.
Though it is not clear exactly how a bee decides to place each bit of wax for a comb, there has been an abundance of research into modeling other aspects of social-insect behavior, such as foraging, resource management, and deciding which work to perform. And great intelligence on the part of ants, bees, wasps, and termites has generally been considered a totally unnecessary hypothesis; much of their behavior can be explained by their following various simple rules. One interesting book on this subject is “Swarm Intelligence: From Natural to Artificial Systems,” by Eric Bonabeau, Marco Dorigo and Guy Theraulaz, published by the Santa Fe Institute, with a good review at The Journal of Artificial Societies and Social Simulation site. Here is a simple example of what the book discusses. Ants will sort out their larvae and dead ants. But how much intelligence does doing so really require? The authors propose that ants follow these simple rules:
“If I am not carrying anything and I find an object out of place, then grab it.”
“If I am carrying something and I find oneself among similar objects, then drop it.”
Which can successfully accomplish that sorting.
Returning to honeybees, Honey Bee Simulation describes some simulations of how bees keep themselves warm in the wintertime; the bees cluster, with every so often the inner bees moving outward to escape their accumulated body heat.
These are examples of artificial-life (a-life) research; attempts to simulate the features of living things, with all their seeming specified complexity, with simple algorithms. One favorite item of artificial-life research has been flocking; Boids has a nice applet showing a flock of “boids” that use some simple rules to flock: move to where one sees lots of neighbors, but not too close to them. And move parallel to them. The field has become very large; consider all the links at Artificial life links. Its successes suggest that widespread unwillingness to perform the “design inference” is very justified.
Such research may seem to apply only to very tiny-brained creatures; moving closer to our species, beavers seem like they intelligently design their dams. Dembski himself seems to agree, in Conservatives, Darwin & Design: An Exchange. But do they? They build their dams with a much simpler technique: they put sticks and mud wherever they hear rushing water, as demonstrated by playing the sound of rushing water with an underwater speaker. This is an effective method for constructing dams, because an incomplete dam will allow water to make some sound as it flows past–at an appropriate point for adding more dam material. This is also an effective way of deciding which parts of dams need to be repaired, since a broken spot will have water rushing through it.
All these patterns of behavior are called by ethologists “fixed action patterns,” which are genetically programmed responses to certain stimuli. But as my examples show, combinations of these can yield very complicated results that may seem like intelligent design.
But some of these examples require some internal state, as in the case of a spider that moves from phase to phase while building its web. And genetic programming does have limits; programming an appropriate reaction to the enormous variety of possible circumstances can be difficult, and a common alternative is genetic programming that sets up adaptability, including the ability to learn. But does the ability to learn indicate the ability to perform intelligent design?
A simple form of learning is imprinting; in various species of ground-nesting birds, newly-hatched ones will imprint on whatever they see moving near the nest and follow it around. This is not a very intelligent mechanism, as can be seen from what baby birds can imprint on: Learning Who is Your Mother: The Behavior of Imprinting; (neither Konrad Lorenz nor Silvia Helena Cardoso nor a white ball looks much like those birds’ mothers). However, under wild conditions, it is generally successful, because what is nearby is usually their mother.
A more flexible mechanism is conditioned learning, though it also is essentially unintelligent. There are two types: Pavlovian or classical conditioning, and operant or instrumental conditioning. An example of Pavlovian conditioning is how my family’s pet cat would run to her food bowl when she heard the sound of a can being opened. A famous example of operant conditioning was the case of Clever Hans, a seemingly learned horse who had lived in Germany a century ago.
This mathematical horse got famous for being capable of a variety of arithmetic and literate feats, reporting his conclusions by tapping the ground with his hoof. However, the psychologist Oskar Pfungst did a variety of experiments on Clever Hans, demonstrating that that horse was picking up unconsciously-generated cues from his questioners, and working from those when to stop tapping. What Clever Hans’s owner, Wilhelm von Osten, had inadvertently done was some operant conditioning; giving Hans a piece of carrot whenever he had tapped the right number of times. The horse then learned to associate his master’s getting subtly tense with when to continue tapping, and his master’s getting relieved with when to stop. Pfungst even went on to demonstrate that he could do what Clever Hans had done, picking up subtle cues from his human experimental subjects.
The case of Clever Hans is important in another way; it demonstrates an important reason why mainstream biologists have generally been very skeptical of claims of animal intelligence: the “unintelligent” mechanisms involved can sometimes be very subtle.
There are some additional learning mechanisms that are common in the animal kingdom, notably habituation and latent or exploratory learning. Habituation is becoming accustomed to some stimulus, like some background noise, that has no association with any direct effect. Exploratory learning is, as the name suggests, exploration of a new environment without an attempt to utilize its contents. Like imprinting and conditioning, these mechanisms are also “unintelligent.”
As we have seen, most seeming examples of intelligent design in the animal kingdom are bogus; the animals have unintelligent mechanisms for producing those “designs.” But by biological continuity, the species closest to our species ought to have at least some ability to perform intelligent design; can that be observed or at least reasonably inferred?
Early in the 20th century, the psychologist Wolfgang Kohler followed that line of reasoning, concluding that chimpanzees are the best choice of species to experiment on, since they are the anatomically closest to our species (later work would demonstrate that they are also the genetically closest). He placed chimps in enclosures with out-of-reach bananas and a variety of items that could be used to reach those bananas. The chimps would at first jump to try to get those bananas, and when that failed, they would pause for a while and then try some other solution, like stacking crates or using long poles, to get at those bananas. For some nice pictures, see Kohler’s Research on the Mentality of Apes.
This phenomenon, called “insight learning,” is difficult to reduce to any of the simpler known behavior mechanisms. In fact, it looks as if the chimps were experimenting in their minds with ways to reach the bananas. Furthermore, some of the techniques, like stacking crates, are techniques with which their ancestors had had little experience in their wild habitat, further demonstrating that those techniques were learned. One counterargument is that the chimps successfully use these items only after having acquired experience with them, as some experiments show. However, that is consistent with the hypothesis of purely-mental experimentation, and it is consistent with the working of much human creativity, if not most of it.
Insight learning is rare in the animal kingdom, though some possible examples of it have been observed in several species, notably pigeons and ravens. It is also relatively difficult to recognize experimentally.
And is insight learning really an example of intelligent design? I believe that it can reasonably called that. A chimp who imagines how to stack some crates to reach a banana could reasonably be described as intelligently designing that crate stack. In fact, the pause before stacking could be when the chimp is doing the designing.
If that is a reasonable interpretation, then our species is not alone in having the capability of performing intelligent design. Even so, that capability is a very rare one, and it is significant that the nonhuman species that it is best-developed in is the closest species to our species.
Having handled the question of what other intelligent designers inhabit our planet, we now turn to the question of intelligent designers inhabiting other planets. Although the Moon and the stars have been celestial Rorschach tests for millennia, it was with the invention of the telescope that it became very apparent that the other planets were essentially Earthlike objects–something especially apparent for the Moon.
Not long after Galileo made his remarkable discoveries with his recently-invented telescope, his good friend Johannes Kepler wrote a book about traveling to the Moon, “Somnium” (“Dream”), published posthumously in 1634. This book described the Moon as being abundantly inhabited, and the Moon’s craters as built by the Moon’s inhabitants. Their circular shape seemed much too regular compared to typical mountain-range arrangements, which led Kepler to infer that they had been intelligently designed.
However, Kepler’s successors have developed other hypotheses, of which the most successful has been giant meteorite strikes. The Moon’s craters resemble scaled-up versions of laboratory-produced impact craters and bomb craters, and their continuous size distribution and random scatter are consistent with an essentially random population of impactors.
And it is very likely that the only life there has ever been on the Moon has been visitors, because the Moon’s crust is very short on elements needed for Earthlike biochemistry, such as hydrogen, and because the Moon is too small to hold even an approximately Earthlike atmosphere.
We now advance in time to the late nineteenth century, when astronomer Giovanni Schiaparelli was observing Mars. He reported that there were “channels” there, but he remained noncommittal on their origin, preferring a natural-phenomenon hypothesis, though thinking that being built by Martians was not impossible. But he had written in Italian, using the word “canali,” which got mistranslated as “canals.” The astronomer Percival Lowell, who also reported seeing them, expanded on that “description,” describing in detail how they had been built by Martian civil engineers trying to irrigate Mars’s surface.
Other astronomers, notably Eugene Michel Antoniadi, failed to see them, even when observing under very good conditions resulting from a very-still atmosphere. Antoniadi would see some canal-like features, but they would be irregular stretches of detail very unlike the network of straight-line features that Schiaperelli and Lowell claimed to have seen.
As a result, many astronomers became skeptical of the existence of the canals, though this question was only settled when spacecraft were sent to Mars over half a century later. The enormous quantity of pictures returned to Earth by these spacecraft contained not a trace of the Schiaparelli-Lowell canals. They were false perceptions, pure and simple.
However, these pictures would soon offer new opportunities for intelligent-design speculation. In 1976, the Viking spacecraft were sent into orbit around Mars, and their orbiters took a large volume of pictures as part of the preparation for the Viking Landers’ landings. Among them was a picture that looked much like a human face. “This is the guy who built the canals of Mars!” announced team member Harold Masursky (my memory of a Science News article or similar publication). He thought that that was very funny, and the Viking team showed pictures of a smiley face in a crater and a picture of Kermit the Frog in a lava flow. See: the Tampa Bay Skeptics “Face” on Mars and More “Faces” on Mars pages.
The Viking team considered these identifications to be funny false perceptions that did not look very much different from other Martian surface features; they never thought that these features were real artifacts.
However, Richard Hoagland and certain others have taken this identification very seriously, making the “design inference” in regard to the Mars Face and such additional features as the nearby “Cydonia City” of pyramids.
Spacecraft have been sent to several other planets and natural satellites, returning similar large amounts of pictures, but for whatever curious reason, these other objects have aroused much less interest from searchers for Mars-face-like features.
Looking beyond the Solar System, we consider the question of SETI, the Search for Extraterrestrial Intelligence. Cocconi and Morrison had shown over 40 years ago that the most energetically efficient way to communicate across interstellar distances was by radio, and since then, there have been several efforts to search for extraterrestrial radio broadcasts. But late in 1967, there was an accidental success that was followed by some other successes. Or was there?
As related in Little Green Men, White Dwarfs or Pulsars?, Jocelyn Bell Burnell, an astronomy graduate student, was working on a project to measure interstellar scintillation of radio sources when she noticed an odd new source that pulsed every 1.337 seconds. She told her thesis adviser Tony Hewish, and they then checked out the possibility that it was some Earthly radio broadcast. Though it looked artificial to them, it behaved in a very odd way for such a broadcast, being fixed in position relative to the stars. It also had a pulse dispersion that suggested that the signal had traveled across some length of interstellar space.
They concluded that it could be some interstellar radio broadcast, and tested the hypothesis that the source was on a planet orbiting some other star by looking for a telltale changing time delay as the signal crossed different amounts of the planet’s orbit over time. However, the only such changing time delay that they noticed was due to the Earth’s motion around the Sun.
They quickly discovered three similar sources, which they named LGM-1 through LGM-4, meaning “Little Green Men,” after a common stereotype of extraterrestrial visitors. However, they eventually settled on the name “pulsar,” short for “pulsating star.”
Astrophysicists did not automatically accept the extraterrestrial-broadcast hypothesis, so they checked out other hypotheses. Gravity is an important contributor to the structure of any object more massive than Jupiter, and Kepler’s Third Law relates the characteristic gravitational timescale of an object to its density. Some pulsars were pulsing too rapidly to be white dwarfs, but they were well within the margin of a long-hypothesized object: the neutron star. But the periods were too slow to be gravitational (about a millisecond), so they must be due to rotation. Thomas Gold published this hypothesis in 1967, and the later observation of pulsars’ periods increasing was perfectly consistent with it–pulsars are spinning down.
Exactly how a pulsar shines is still an unsolved problem, but all the theoretical work on that problem has ignored the LGM hypothesis (intelligent design), focusing on features of pulsar magnetospheres that may be responsible. And there are some promising theoretical leads that are being worked on, like a pulsar’s spinning magnetic field producing an electric field that causes electron-positron pair production in its magnetosphere.
Having examined all these successes, I turn to an intelligent-design-related endeavor that has had much less success: SETI, the Search for Extraterrestrial Intelligence. However, the interesting feature here is the justification for the search strategies most commonly used. For some details on some typical SETI strategies, see SETI Institute FAQ
and SETI Science & Education Links.
One very common strategy is to search for narrowband signals, those with a bandwidth of 1 Hz out of ~1 GHz (1 part per billion). This has two justifications: narrowband signals are more prominent against the radio background than the same emission power of broadband ones, and no known astrophysical phenomenon produces such narrowband signals. I will discuss these criteria in more detail.
Radio-frequency receivers always have some frequency resolution, and the broader their frequency resolution, the more radio background they will see (it’s very broadband). A receiver with 10-Hz resolution will thus see 10 times as much noise as one with 1-Hz resolution. And if one’s receiver has a resolution of 1 Hz, a signal with 1-Hz bandwidth will be ten times more prominent than a 10-Hz-bandwidth signal with the same emitted power. So to get more detectability for the megawatt of transmitter energy consumption, it is desirable to make the signal as narrowband as is reasonably feasible.
There are also no known natural astrophysical processes that produce signals with a bandwidth less than about 300 Hz; this bandwidth is a result of thermal broadening of radio-frequency spectral lines. Some of the molecules are moving toward us, and some are moving away relative to their average motion. Temperatures in the interstellar medium are never lower than about a few degrees K; for ~ 1-GHz frequencies and typical molecules’ weights, one quickly derives that limit.
One of the radio-frequency spectral lines is the hyperfine transition of neutral hydrogen atoms at 1420 MHz; searches have often been done near that frequency, because aside from interstellar hydrogen clouds, the interstellar background is relatively weak near that frequency, and also because that frequency forms a prominent spectral landmark.
These reasons are very distinct from Dembski’s Explanatory Filter, which focuses on alleged unexplainability as a natural phenomenon; they are an attempt to predict what an extraterrestrial broadcaster is likely to do, using the fact that they live in the same kind of Universe that we do. A “design inference” would be made by asking “would a designer have created that signal?” rather than “what else could it be?”
What we can learn from all these examples? As we have seen, the large majority of purported examples of intelligent design have turned out to be much more plausibly explained by other hypotheses, including false perceptions on the part of some observers. This alone ought to suggest that detection of intelligent design is much more difficult than it might seem as first, and thus that Dembski’s claim to have solved that problem is very overoptimistic.
It is sometimes argued that scientists work with “materialistic presuppositions” that keep them from considering intelligent design. However, I have discussed here several counterexamples: cases where intelligent design was seriously considered. But it was rejected in most cases because there are much more plausible alternative hypotheses. If spiders, ants, bees, beavers, horses, chimps, etc. are much smarter than they let themselves appear, then they must be hiding it very well, even to the extent of causing much trouble for themselves. Why would they let themselves get killed, stolen from, enslaved, and otherwise mistreated just so they can seem stupid to us? And why would the Moon’s inhabitants be very careful to build a large number of structures that closely resemble what one would expect giant meteorite strikes to produce?
A parallel case is vitalism, the theory that living things have some special “vital force” or something similar that distinguishes them from nonliving entities. Though common in past centuries, this view has suffered numerous setbacks over the last few centuries, to the point that it is now totally discredited. As an example, “organic compounds” had gotten their name from the belief that only living things could make them, but that was descredited by numerous experiments starting with Friedrich Woehler’s 1828 synthesis of urea from inorganic compounds. Thus, the discrediting of vitalism is not due to “mechanistic presuppositions,” but instead to the success of mechanist, nonvitalist explanations.
Another is in the theory of mental functions, where “mentalistic” theories have often been rejected in favorite of physicalist ones for similar reasons.
In conclusion, the mainstream of the scientific community has no fundamental, a priori bias against considering intelligent design. However, many seeming cases of intelligent design have turned out to be something other than that, which suggests good reason to be skeptical of intelligent-design claims.