An Explanatory Filter
How It Works
How does one determine something in our universe is the result of design?
How can one distinguish what appears to be designed from what is truly intelligently designed?
What qualifies for this inference of intelligent design?
The effort to visualize Intelligent Design today opens an avenue to understanding our origin in a most remarkable context. While science uncovers immense complexity amid a biological eloquence that works so beautifully, simplistic cause and effect explanations fail to provide answers for anything other than what is confined to the strictly material and natural causes. The Explanatory Filter discerns the difference!
To provide specific examples to have in mind, we refer you to Dr. Behe's book (Darwin's Black Box (bookstore link) that describes the cilium, bacterial flagellum, how blood clots (e.g., the biochemical pathway that makes this work), our immune system, and other key examples that are irreducibly complex. This ties into Dembski's discussion on specificity (different terms but same idea!). So, each of Dr. Behe's examples are the types of subjects we seek to put through the explanatory filter. And below, when we say 'feature,' characteristic,' or 'event,' we are really speaking of these types of examples from biology.
Also, remember when we say intelligent design, we are appealing to explanations that step beyond random process or chance. Evolution theory rests on the latter, intelligent design takes features within nature into another sphere of discussion.
To understand intelligent design requires a tool to demonstrate the presence of design in the world and the universe at large. There is at least one such tool that comes to us through Dr. William Dembski. He has written articles and given presentations on what he calls the Explanatory Filter. An overview of how this filter works is given below.
Simply put, if we think something is the result of a design, then it passes through a filter (a test) whereby the product of design proves to be a result of (1) contingency, (2) complexity, and (3) specification ... these three filter elements are explained further in the next section. If all three components of the filter eliminate the conclusions of 'necessity' and 'chance,' then one is left to conclude design.
The filter essentially posses three questions that pass us from one possible consideration to the next. Answering each question requires critical thinking and some research. But getting answers is the intent of this tool, to provide a mechanism for researching (detecting) the presence of design in nature.
And if there is reasonable evidence for design in nature ... what then? Well, that's the reason for a look at the WindowView! The answer to this question is: If the evidence exists, then there is indeed something very special about life! And as more evidence is gathered to document an increasing number of life's features as evidence for design, then the more one is compelled to define life in terms of intelligent design.
Tweet this page address!
Admittedly, reading some of Dr. Dembski's books, chapters, or articles can leave one's head spinning. To present the explanatory filter in plain language—which is the goal of this feature article—risks losing some academic detail while benefiting a larger audience with a general overview. The filter is a means to detect complexity and design.
... Given something we think might be designed, we submit it to the filter. If it successfully passes all three stages of the filter, then we are warranted asserting it is designed. Roughly speaking the filter asks three questions, and in the following order: Does a law explain it? Does chance explain it? Does design explain it? Dembski (MC) Page 94
So, how can we detect design in our world?
Explanations that find a link to intelligent design appeal to something that exists in nature but that transcends simply a natural occurrence—this goes beyond chance occurrence alone. To determine this special case holds requires a tool, identified here as an 'Explanatory Filter.' This is how one can detect or infer we are looking at an example of something that is intelligently designed. And note that we need not know the identity of the intelligent agent to make the inference.
The following paragraphs include a series of short quotations from Dr. Dembski's article that appeared in the July/August 1999 issue of Touchstone magazine (see reference below).
What Is The Cause?
Humans tend to think of events in terms of cause and effect. Tip one domino and all its neighbors fall down in turn. This relates to our common material experience and what our senses tell us is true. Sometimes a bit of detective work is required to find the first domino, but when we can find the initiator, the key domino, the cause of subsequent events becomes clear! But if it's you doing the tipping, then your decision of which domino and when to tip it is not by chance—rather, its the result of an intelligent agent at work! The cause is directed by that agent.
''Events that happen but do not have to happen are said to be contingent. In our workaday lives we distinguish two types of contingency, one blind, the other directed. A blind contingency lacks a superintending intelligence and is usually characterized by probabilities. Blind contingency is another name for chance. A directed contingency, on the other hand, is the result of a superintending intelligence. Directed contingency is another name for design.'' Touchstone Magazine (see reference below)
For reasons Dembski explains elsewhere—through history, philosophy, etc.—scientists helped humanity to distance the question of design from the landscape of common thought.
''But was science right to repudiate design? My aim in The Design Inference (Cambridge University Press, 1998) is to rehabilitate design. I argue that design is a legitimate and fundamental mode of scientific exploration on par with chance and necessity. ... Design, in the form of Aristotle's formal and final causes, had after all once occupied a perfectly legitimate role within natural philosophy, or what we now call science. With the rise of modern science, however, these causes fell into disrepute.
''But suppose we lay aside a priori prohibitions against design. In that case, what is wrong with explaining something as designed by an intelligent agent? Certainly there are many everyday occurrences that we explain by appealing to design.''
''When intelligent agents act, they leave behind a characteristic trademark or signature—what I call specified complexity. The complexity-specification criterion detects design by identifying this trademark of designed objects.'' Touchstone Magazine (see reference below)
Designed objects are much like those cellular features considered in the WindowView section above (see Irreducible Complexity).
''Complexity and probability therefore vary inversely: the greater the complexity, the smaller the probability. Thus to determine whether something is sufficiently complex to warrant a design inference is to determine whether it has sufficiently small probability.'' Touchstone Magazine (see reference below)
Okay, so what is Dembski saying? First, the concept of design—even finding examples of it in nature—has met opposition. In a word it's been denied! Denial does not make truth untrue? Denial itself is a barrier to explore what may be true.
Second, if we find something that fits his definition of "specified complexity," then there really isn't much room—low probability is like no room—for this to have come about, to exist, by chance.
To simplify matters further, let's briefly describe Dembski's flowchart entitled ''Explanatory Filter'' (which is discussed in the body of the article below and illustrated in the animated graphic here).
To start, we can ask: Is something designed? Is an intelligent agent behind this?
The explanatory filter looks at contingency as the first step. Something is either here by necessity or possibly design. If not necessity then we move to the next step and ask: Is it complexity?
If not complex (in terms of Dembski's definition), then you are looking at a product or object of chance.
But if not by chance, then is there specificity? Dembski notes that specification is that part of the filter that differentiates a pattern as either a product of chance or those patterns in the presence of complexity that warrant a design inference.
If not chance, then design!
This series of questions, explained best in context—in the Touchstone article or other of Dembski's writings—is a research tool to take out into the world. This is a design detector, a tool for thought, discovery, and revelation. The more we find examples of intelligent design in the world, the more one is challenged to objectively consider the source of design!
What is presented above is a simple 'thumbnail' overview. What follows gives more detail and reveals more about the test that is provided by the explanatory filter.
The Explanatory Filter
To put all this in other words, the filter is being used to eliminate chance as an explanation. Statisticians do this all the time with the various mathematical tools at their disposal. A statistician looks at probabilities that something (i.e., an event) will happen. Some events leave patterns but we must ask if this comes with high or low probability.
Patterns may therefore be divided into two types, those that in the presence of small probabilities warrant the elimination of chance and those that despite the presence of small possibilities do not warrant the elimination of chance. The first type of pattern will be called a specification, the second a the fabrication. Dembski (MC) Page 97
And specification is the 'telltale' indicator that we are looking at design.
Looking a bit further we see that an event may be explained in one of three ways, by: law, chance, or design. If by law the event will happen as long as conditions are right.
To attribute of an event to chance is to say that its occurrence is characterized by some (perhaps not fully specified) probability distribution according to which the event might equally well not have happened.
To attribute an event to design is to say that it cannot plausibly be referred to either law or chance. In characterizing design as the set-theoretic complement of the distinction law-or-chance, one therefore guarantees that these three modes of explanation will be mutually exclusive and exhaustive.
It remains to show that this eliminative approach to design (i.e., as the negation of law and chance) corresponds to design in the ordinary sense (i.e., as the product of intelligence). Dembski (MC) Page 98
So, we want to exhaust potential explanations one by one to end up at the true reason an event results in what we observe. Events characterized by Dembski's series of considerations are either highly probable (HP), of intermediate probability (IP), or are improbable (in some cases exceedingly improbable (SP)). The following three quotations touch on these three levels of probability:
Explanatory priority is a case of Ockham's razor. Accordingly, when any one of the three modes of explanation fails adequately to explain an event, we move to the mode of explanation at the next level of complication. Note that explanations that appeal to a law are simplest, for they admit no contingency, claiming things always happen that way. Explanations that appeal to chance add a level of complication, for they admit contingency but one characterized by probability. Most complicated are those explanations that appeal to design, for they admit contingency but not one characterized by probability. Dembski (MC) Page 100
We think of the laws of nature as giving us an explanation that is quite certain as long as the conditions are right. Chemistry, physics, biology, etc., appeal to these laws and events and results are characterized as highly probable and thus a consistency in events leads to what is HP.
Events of intermediate probability, or what I am calling the IP events, are the events we can regularly expect to occur by chance in the ordinary circumstances of life. Rolling snake eyes with a pair of fair dice constitutes an IP event. Even someone winning a lottery where the probability of winning is as little as one in ten million will constitute an IP event once we factor in all the other people playing the lottery. IP events are sufficiently probable to give us no reason to suspect they were the result of something other than the chance. Dembski (MC) Page 100
But now we turn to the more special case of what is improbable. Albeit, we cannot review Dembski's entire discourse on the topic here, but the following quotations begin to frame the more discriminating questions and considerations that lead us to the final endpoint of the filter and the inference for design.
Our intuition is that SP events are so improbable that they cannot happen by chance. Yet we cannot deny that exceedingly improbable events (i.e., SP events) happen by chance all the time. To resolve the paradox we need to introduce and extraprobablistic notion, a notion I referred to as a specification.
Specifications are common in statistics, where they are known as rejection regions. Dembski (MC) Page 102
We'll jump over a lot of the discussion on the statistics ... that's not really fair, but again we are just giving a hint of the details you can get by reviewing the published source cited here.
In a nutshell, Dembski leads us through all the various side routes and if these are eliminated, then we end up a the extremely improbable (labeled here as sp/SP; i.e., specified small probability). We aren't working through a specific example, but if the filter brings us to this point, then the feature (e.g., a particular biochemical pathway) or function (e.g., say a certain cell structure) should be looked at as a product of design. Chance is thrown completely out the window!
There is, however, an important difference between the logic of the explanatory filter and the logic of statistical hypothesis testing: by ending up at the terminal node of the filter of labeled "design," one sweeps the field clear of all relevant chance explanations. This contrasts with statistical hypothesis testing, where eliminating one chance hypothesis opens the door to others. The SP/SP events of the explanatory filter exclude a chance decisively, whereas the fence that fall within the rejection regions of statistics indicate that some probability distribution other than the one originally suspected may be operating. Dembski (MC) Page 102
So, you are wondering what kinds of present day, routine, processes are already employed by people to evaluate events (E) in terms of design.
In fact, humans are in the business of 'filtering' evidence for intelligent causes—take a look here ... we do this all the time!
To put it another way, the design theorist is in the business of categorically eliminating chance in accounting for E whereas the statistician is in the business of ruling out individual probability distributions that might account for E. Dembski (MC) Page 103
... the crucial question now becomes whether E was specified. Is E an sp/SP event or merely an SP event?
... passage through the flow chart to the terminal node labeled design encapsulates:
- how copyright and patent offices identify the theft of intellectual property,
- how insurance companies prevent themselves from being cheated,
- how detectives employee circumstantial evidence to incriminate a guilty person,
- how forensic scientists are able to rely heavily to place individuals at the scene of a crime,
- how skeptics debunk the claims of parapsychologists,
- how scientists identify cases of data falsification,
- how the Search for Extraterrestrial Intelligence (SETI) program seeks to identify the presence of extraterrestrial life, and
- how statisticians and computer scientists distinguish random from non-random strings of digits. Dembski (MC) Page 104
The Filter works when we are fully aware of all possible implications of the evidence and how we conduct our research. So, one is cautioned to carefully look to make correct and not false conclusions.
The filter is a criterion for distinguishing intelligent from unintelligent causes. Here I am using the word criterion in its strict etymological sense as a method for deciding or judging a question. Dembski (MC) Page 104
Any medical test is a criterion.
All criteria, not just medical tests, the face the problem of false positives and false negatives.
The explanatory filter is a criterion for detecting design. Is it a reliable criterion? The target group for the explanatory filter comprises all things intelligently caused. ... The things we are trying to explain have causal stories. In some of those causal stories intelligent causation is indispensable whereas in others it is dispensable. ... When the explanatory filter fails to assign something to the target group, can we be confident that no intelligent cause underlies it? If not, we have a problem with false negatives. Dembski (MC) Page 105
Now we are in an area where our intelligence is put to the test. Simply put ... to NOT be fooled. To research carefully, test and double check, and then to discern what the evidence reveals.
This problem of false negatives, however, is endemic to detecting intelligent causes.
One difficulty is that intelligent causes can mimic law and chance, thereby rendering their actions indistinguishable from those of unintelligent causes.
It takes an intelligent cause to know an intelligent cause. But if we do not know enough, we will miss it.
Intelligent causes can do things that unintelligent causes cannot and can make their actions evident. When for whatever reason an intelligent cause fails to make its actions evidence, we may miss it. But when an intelligent cause succeeds in making its actions evident, we take notice. This is why false negatives do not invalidate the explanatory filter. Dembski (MC) Page 106
We are now in the end game, at the bottom of the flow chart ... we've noted false negatives may throw us off course ...
And this brings us to the problem of false positives.
... I want to argue that the explanatory filter successfully avoids of false positives. Thus whenever the explanatory filter attributes design, it does so correctly. Let us now see why this is the case. I offer two arguments. The first is a straightforward inductive argument: in every instance when the explanatory filter attributes design and where the underlying causal story is known, it turns out design is present; therefore design actually is present whenever the explanatory filter attributes design.
The naturalist is likely to object at this point, claiming that the only things we can know to be designed are artifacts manufactured by intelligent beings who are in turn the product of blind evolutionary processes.
It is circular reasoning to invoke naturalism to underwrite an evolutionary account of intelligence and then in turn to employ this account of intelligence to insulate naturalism from critique. Naturalism is a metaphysical position, not a scientific theory based on evidence. Any account of intelligence it entails is therefore suspect and needs to be subjected to independent checks. The explanatory filter provides such a check. Dembski (MC) Page 107
Again, naturalism is that thinking that confines all causes to within the material world and thus within limits that we attribute to the common experience in daily life.
Let's take note, also, that some low probability events may not qualify as design. How low a probability? Interestingly, for example, any event with a probability approaching 10-50 may not be the result of design according to Borel (see reference below). Any smaller a probability (for example: 10-55 or 10-150 or lower would be improbable. Look at the numbers computed for probabilities in our feature article on chance and you'll see how improbable life is based on this benchmark for probability!)
Borel's universal probability: bound = 10-50
I challenge anyone to exhibit a specified event of probability less than Borel's universal probability bound for which intelligent causation can be convincingly ruled out. Dembski (MC) Page 108
There has been some thinking on this and we see limits and specificity as a key filtering element. The inference for design is not arbitrary.
My second argument showing that the explanatory filter is a reliable criterion for detecting design considers the nature of intelligent causation and specifically what it is about intelligent causes that makes them detectable. Even though induction confirms that the explanatory filter is a reliable criterion for detecting design, induction does not explain why the filter works. Dembski (MC) Page 109
Interestingly, even with the material framework in which we live, design has to be delectable for us to even have this discussion. That design is delectable puts the discussion on the playing field with all other research disciplines—even after so many scientists have tried to discredit design as a possibility or reality!
What is it about eliminating law and chance that purchases design? ... That design and nothing else could remain once the filter eliminates law and chance is not immediately apparent. And yet and I shall argue that this is precisely the case.
The principal characteristic of intelligent causation is choice.
Intelligent causation always entails discrimination, choosing certain things, ruling out others. Dembski (MC) Page 109
Although small probabilities figure prominently in the explanatory filter, their role in this general scheme of recognizing intelligent causation is not immediately obvious. In the scheme a choice is made among several competing possibilities, the rest are ruled out, and the possibility chosen is specified. Where in this scheme are the small possibilities? Dembski (MC) Page 111
Again, we've hit the highlights here. Dembski gives further examples to bring the quotations above to fullness.
Remember the example of tipping over the dominoes at the start of this article? There may well be design and choices in how many and the dominoes are to be set up. There may be design in that one key domino will play an important role in setting off a wonderful chain reaction of tipped dominoes. And another aspect is the choice as to which domino to tip and when.
My second argument ... The explanatory filter is a reliable criterion for detecting design because it coincides with how we recognize intelligent causation generally. In general, to recognize intelligent causation we must observe a choice among competing possibilities, note which possibilities were not chosen and then to be able to specify the possibility that was chosen. Dembski (MC) Page 111
Finally, in expressing what the term intelligent design means, let's take a look at the words that make the term itself:
As a post script, I call the reader's attention to the etymology of the word intelligent. The word intelligent derives from two Latin words, the preposition inter, meaning "between," and the verb lego, meaning "to choose or select." And thus according to its etymology, intelligence consists in choosing between. It follows that the etymology of the word intelligent parallels the formal analysis of intelligent causation inherent in the explanatory filter. Intelligent design is therefore a thoroughly apt phrase, signifying that design is inferred precisely between an intelligent cause has done what only an intelligent cause can do, to wit, make a choice.Dembski (MC) Page 112
In the final analysis, the explanatory filter functions well by identifying what we need to be looking for when researching and detecting design. This is a tool that can be used in future research.
Touchstone Magazine - A Journal of Mere Christianity. Special double edition July/August 1999 on Intelligent Design. To obtain back issues or to subscribe: Publishing Management Associates, 129 Phelps Ave., Suite 312, Rockford, IL 61108. (815) 398-8569. (Note: this back issue may be sold out, contact the publisher for availability; click here to return to text above)
Quotations from "Mere Creation" (MC) edited by William A. Dembski are used by permission of InterVarsity Press, P.O. Box 1400, Downers Grove, IL 60515. www.ivpress.com All rights reserved. No portion of this material may be used without permission from InterVarsity Press.
Writer / Editor: Dr. T. Peterson, Director, WindowView.org