I was at an ethics lecture when the professor stated that philosophy studies what science does not. This was when I became aware of the philosophical version of Stephen Jay Gould’s concept of nonoverlapping magisteria (Gould 1997). This was a little surprising because the lecture was on the ethics of technology, but it was not as surprising as being introduced to Massimo Pigliucci’s promotion of the philosophical magisterium at CSICon 2018 during his presentation on “The Variety of Scientisms and the Limits of Science.” Is it true that there are questions the scientific method cannot address?

The scientific method studies objective physical reality. It uses reason and evidence (logic and measurement) to do so. We can only experience the physical world. Anything physical can be measured. Logic can be applied to any measurement. This is a logical refutation of a philosophical magisterium. But people keep presenting questions they claim cannot be analyzed via the scientific method, so it behooves us to engage in deeper discussion.

To review and illustrate the scientific method, consider plumbing. Hypothesis: changing a washer fixes a leak. Experimental design and execution: change the washer. Data collection method: put hand under faucet. Analysis method: check if hand is wet. Peer review is conducted by the customer. Because plumbing can be mapped onto the scientific method, it is difficult to understand why Pigliucci thinks it is not science. Perhaps the confusion is that it is practical science rather than theoretical science aimed at advancing human knowledge. Perhaps another confusion is that practical science is not formal science (and we sometimes say that science was not done prior to science’s formalization in the 1600s). But a magisterial discussion is about foundations, not formalizations. Smashing particles into each other is an advanced version of a baby pressing against something to test resistance. Instead of denying plumbing is science, perhaps Pigliucci could use it to illustrate that the scientific method is not mysterious.

A magisterial discussion is also about “in theory” rather than “in practice.” The difference is illustrated by Bertrand Russell’s teapot (Russell 1952). In theory, we could scour space for its existence, but it is neither practical nor of value to do so. A variant of “in practice” is “state-of-the-art.” Science will never analyze every question, but that doesn’t mean the scientific method is incapable of analyzing any specific question.

Pigliucci inferred, as others have, that science isn’t really done using the scientific method because scientists do not perform the scientific method on a daily basis. This ignores the exploratory and preparatory parts of science. Exploring possible hypotheses and formalizing them is possibly the most time consuming part of science. Developing instrumentation can take a lot of effort. Formal hypothesis testing is only part of the scientific method.

Science assumes an objective reality—that something exists other than, and independent of, you. The other option is to assume solipsism, which ends the conversation because it takes two to have a conversation. We all see the same moon. A quantum system still exists when it’s in an unmeasured quantum state. If we are a simulation, there is still a simulator (Moskowitz 2016).

Because we only observe reality through our senses (possibly technologically enhanced), every individual develops a subjective model—a representation—of reality.¹ Groups develop models of reality, such as the body of knowledge produced by the scientific method, through consensus. Neither of these models is reality itself, nor do they represent reality perfectly.

The distinction between objective reality and models of reality is a key concept that is illustrated beautifully by René Magritte’s La Trahison des image (The Treachery of Images). It is not a pipe. It is an image of a pipe. Subjective models are the reason for confirmation bias. The scientific method’s objective is to refine the models to be as close to objective realty as possible. Einstein’s model is more refined than Newton’s.

Nothing nonphysical affects the physical world. Anything that affects the physical world must do so in a measurable way, even if it is just to nudge an electron. In a sense, we do not measure magnetism; we measure its effect on electrons. This is what is meant by a physical force. Assume God talks to me and I tell people about it. Because speaking is a physical act, somewhere along the way to my larynx, God must have made a physical change in my body. In that case, by definition, God would be a physical force. If there is something outside our physical reality, there is zero possibility of knowing anything about it. Thus, it is true that questions about nonphysical things cannot be analyzed by the scientific method—or any other methodology. To discuss such questions is purely entertainment. (Although such questions can be analyzed by science to understand the people asking those questions.)

Science is dependent on measurability. When I say “metrology” (the science of measurement), most people think “meteorology.” Few people know the formal definition of “a measure.” The point is that few people are experts in measurement. Beauty is traditionally thought to be unmeasurable. Yet Netflix’s and Amazon’s recommendation algorithms predictively measure a form of beauty in the eye of each beholder. (When you rate a movie 4 out of 5 stars, you’re saying it is 4/5ths beautiful.) Consciousness is also usually considered unmeasurable. Yet Zap and Zip (Koch 2017) is a first stab at measuring it. Any aspect of physical reality is measurable. As such, any statement about physical reality is testable.

Science depends on precise definitions. An example of the importance of this is God. Because God is so ill-defined, it appears impossible to disprove its existence. However, there are plenty of characteristics that have been used to define God. These have been used to disprove the existence of many types of gods (Martin and Monnier 2003). For example, if being merciful excludes condoning torture, God cannot be merciful and at the same time condone it. Any noun or verb must have physical characteristics to be meaningful. (Though imaginary, unicorns invoke a physical image.) Similarly, meaningful abstract concepts are abstracted from instances of physical reality. Otherwise, terms and phrases are no more than vibrations in the air or ink on paper. In practice, there are meaningful terms that don’t lead to testability because they are only partially defined with physical characteristics. Part of exploratory science is to refine such terms to testability.

Ideally, experiments are repeatable. When I kick the universe in the same way, it kicks back in the same way—and experiments should not contradict each other. This is based on the assumption (for which there is much evidence) that objective reality is consistent. (Even believers in magic tend to assume casting the same spell produces the same result.) However, the scientific method acknowledges human imperfection. When experiments contradict each other, science requires reevaluation and the development of new experiments. It is self-correcting.

In providing a list of questions that he thinks cannot be analyzed by science, Pigliucci (2018) engages in a state-of-the-art fallacy (a form of argument from ignorance). Just because science has not analyzed a question does not mean it cannot. This is a common tactic of proponents of the religious magisterium and apparently of the philosophical version as well. This tactic is also a game of Whac-a-Mole. Well, here are a few whacks.

Contrary to popular belief, social sciences are, in essence, physical sciences, because they measure functions of physical brains. They predictably measure individual and group preferences in ways similar to recommendation algorithms. The Reproducibility Project (Open Science Collaboration 2015) evidences social sciences’ use of peer review, self-correction, and reproducibility. People bemoan that these sciences do not make “precise” measurements; their results are just statistical distributions. Well, different people answer questions of the type “From 1 to 10” differently, demonstrating that distributions are the proper presentation of the results. Besides, the physical sciences are also based on statistical distributions. A fundamental principle of metrology is that every measurement includes an error. We cannot state that “a ruler is a foot long.” The best we can state is that “a ruler is a foot long plus or minus an error.” The uncertainty principle formalizes this for simultaneous measurement of position and velocity. A celebrity example is the existence of the Higgs boson. Nobody has a Higgs boson in their hands. The conclusion is a statistical inference from complex measurements, every one of which has a margin of error (Yirka 2012). The difference between the two types of science is that the “soft” sciences are much harder than the “hard” sciences. We’ve discovered the Higgs boson but haven’t solved world hunger.

Pigliucci (2018) asks, “Is modus ponens valid?” The answer is: it’s your choice. Although not usually called such, modus ponens is an axiom; it is assumed to be valid. It doesn’t have to be, but logics without it tend not to be very useful. For example, I would be surprised if there is a proof of “1 +1 = 2” that doesn’t use it, let alone something such as the theory of evolution. But useful axioms can also be treacherous. Consider the axiom of choice (AC), which, informally, states that given any collection of bins, each containing at least one object, it is possible to make a selection of exactly one object from each bin. At first glance this seems innocuous, yet it leads to the Banach-Tarski paradox (Banach and Tarski 1924). Informally, this paradox states that, given an object of a certain volume, it can be deconstructed into parts that can be reconstructed into two objects, each of which has the same volume as the original object—a physical (vs. logical) absurdity. Ultimately, mathematics, including modus ponens, is a tool. In what sense is a hammer “valid”?

Pigliucci (2018) also asks, “What is the ontological status of numbers?” Or, more prosaically, “Do numbers exist?” It is silly to imagine holding a one in your hand the way you do an apple. Being a concept, a one is a model. Being a thing, an apple is part of physical reality. To conflate models with reality is a category error. We can, however, point to “one apple” and “one other apple” and thus point to “two apples.” This illustrates that numbers and arithmetic “exist” in the physical world through instantiation. We can also point to physical representations of numbers, including ink on paper and electrochemical constructs in brains, but those aren’t actual numbers. In response to the classic question: the instantiations of mathematics are discovered; its representations (the models) are invented.

In providing a state-of-the-art argument about abortion, Pigliucci (2018) joins many others in claiming science cannot be applied to morality. Morality is a human concept about volitional acts and social interactions. Rocks, the universe, and amoebae don’t contribute to this concept. Let us follow a common scientific method approach of first looking at specific examples when analyzing difficult questions. There is a great deal of evidence, such as large infrastructures and legal toxicant levels, that providing clean drinking water is considered good. The approach then asks if there are underlying principles. Among others, conveying the idea of clean water requires communication, which is inherently a cooperative action. Further, providing clean water requires cooperation to develop filtration and delivery systems. This leads to a hypothesis that cooperation is a fundamental principle of morality. The approach then asks if there is further evidence. A set of data to look at is the billions of years of experiments conducted by evolution—what are the successes and failures? More than one person (Shermer 2004; Tomasello 2018) has noted that the evolutionary growth of society—from individual to family to tribe, etc.—requires cooperation. Humans cannot survive as individuals. Even though he presents it as subservient to the selfish gene, Richard Dawkins (1976) provides much evidence that “Selection has favored genes that cooperate with others.” Studies of the iterated prisoner’s dilemma also support cooperation as good. (In contrast, using the single iteration version to evaluate human morality is inappropriate. We are social creatures with memory—there is always potential for a next iteration. Further, it is a self-fulfilling prophesy, because the conclusion depends on whether the objective is personal or group wealth.)

Consider generalizing this. Assume you have a moral maxim. Where did this come from? Didn’t it come from observations of physical reality? Wasn’t it from experiments or people’s thoughts or actions? Even hallucinogenic visions manifest in physical brains. In other words, wasn’t exploratory science used? Now, call that maxim a hypothesis, and experiments can be devised to test it.

The science of morality is in its infancy. As with all fields of study, it will never analyze all questions. But when multiple lines of inquiry—both evidence- and logic-based—converge on a common conclusion, it starts looking like a scientific theory.

I assume philosophers are interested in using any gleaned wisdom to change people’s actions (if only their own). Such actions, being physical, are in the scientific method magisterium. Suggesting that philosophical hypotheses are not informed by science ignores not only exploratory science but also the feedback of experiments to refine hypotheses. Using the phrase “clean water” is not good enough. If acceptable—measurable—levels of toxicants cannot be defined, then all water is “clean,” and the phrase is meaningless.

Philosophy may not be dead as some people claim. But the philosophical magisterium is a nail in its coffin. Why should people be interested in something (other than for entertainment purposes) that does not affect physical reality? Philosophy, like morality, is about human concepts. The millennia of philosophical records are thus evidence that can be used, in the same way social surveys can be used, as a basis of exploratory science. Philosophy has played—and can potentially continue to play—this role. But philosophical consensus is not enough. Philosophers need to acknowledge that any epistemic statement must be supported by more than opinion. For example, the opinion (widely accepted for millennia) that women need to be controlled by men to curb their inherent sinfulness does not hold up against evidence.

This brings us to epistemologies. Assuming there are questions that science cannot analyze, what method does philosophy offer to analyze these questions? Pigliucci (2009) offers “logic or more broadly reason-based arguments.” So, which logic should we use: to AC or not to AC? There are much subtler disconnects between reason-based arguments and reality than the result of the Banach-Tarski theorem. How do you identify them—especially in the presence of confirmation bias? The product of any epistemological investigation is a sequence of words presented as a statement about reality. But words are treacherous. This is why epistemologies must include testing statements against reality. Logics and their conclusions are tools. You don’t choose a hammer because it is derived logically. You choose it because it reliably drives real nails into real wood.

It is absurd to suggest that science doesn’t use logic or reason-based arguments (e.g., the first step in hypothesis testing is checking logical consistency), so the issue is strictly about using evidence. The problem with claiming that the use of evidence is not necessary is that it opens the door to ignoring evidence. Any statement about reality is testable because all aspects of reality are measurable. The only acceptable time for not using evidence is when it is unavailable due to practical limitations. Saying otherwise is saying that using evidence is a choice. Who gets to choose? People who claim that vaccines cause autism?

Pigliucci suggests we use logic and also asks if logic is valid. Besides being a little contradictory, the latter of these raises the specter of a common antiscience meme: science doesn’t “prove” anything. Well, no other epistemological method does either. The difference from some epistemologies is that science acknowledges this. What the scientific method does is provide evidence that is consistent across time, space, and culture. This has led to an overwhelming demonstration of its usefulness—from plumbing to understanding evolution. Using science based on its usefulness is sometimes dismissed as just pragmatism. But why use an epistemology that isn’t useful?

Some philosophers are embracing science. Why aren’t all philosophers? Science is at the point where it can provide—and is providing—objective advances on philosophical questions. There are always practical and state-of-the-art limitations; any decision is based on limited knowledge and evidence. But why isn’t there a form of rejoicing about progress being made on classic philosophical problems? Are philosophers more interested in hearing themselves talk than making measureable changes in the world? People will continue to apply science to philosophy regardless of proponents of a philosophical magisterium. Rather than claim that people cannot do what they are already doing, philosophers could help develop hypotheses and evaluate experimental methods and results.

The claim that some portion of philosophy is outside of science is not supported by analysis. Nothing nonphysical can be studied via the human body. Everything physical is measurable. Science can analyze anything measurable. Further, philosophical activities can be mapped onto part of the scientific process.

The methods used to support the philosophical magisterium are the same types of methods used to promote alternative medicine and pseudoscience: Whac-a-Mole arguments, claiming immeasurability, inaccurately characterizing science, ignoring the treachery of subjective models, and saying that using evidence is a choice (to name a few). These legitimize statements such as: “Science just can’t explain [insert pet idea here].” The philosophical magisterium needs to be refuted not just because it is false but also because it supports the process by which people die from quackery.

Note

1. The modelist theory requires only three assumptions: cogito, ergo sum; something exists besides self (objective reality); and that sensory perceptions interpret, rather than capture, reality.

References

Banach, Stefan, and Alfred Tarski. 1924. Sur la décomposition des ensembles de points en parties respectivement congruentes. Fundamenta Mathematicae 6: 244–277. Available online at http://matwbn.icm.edu.pl/ksiazki/fm/fm6/fm6127.pdf.

Dawkins, Richard. 1976. The Selfish Gene. Oxford, United Kingdom: Oxford University Press.

Gould, Stephen. 1997. Nonoverlapping magisteria. Natural History. Available online at https://web.archive.org/web/20190403152432/http://www.stephenjaygould.org/library/gould_noma.html.

Koch, Christof. 2017. How to make a consciousness meter. Scientific American. November: 28–33. Available online at https://www.alleninstitute.org/media/filer_public/3e/7a/3e7aabb0-5da7-4915-b4b6-2aa896c8faee/2017_11_howtomakeaconsciousnessmeter.pdf.

Martin, Michael, and Ricki Monnier (eds). 2003. The Impossibility of God. Amherst, NY: Prometheus Books.

Moskowitz, Clara. 2016. Are we living in a computer simulation? Scientific American Physics. Available online at https://www.scientificamerican.com/article/are-we-living-in-a-computer-simulation/.

Open Science Collaboration. 2015. Estimating the reproducibility of psychological science. Science 349(6251), aac4716. Doi: 10.1126/science.aac4716. Available online at http://science.sciencemag.org/content/349/6251/aac4716.full?ijkey=1xgFoCnpLswpk&keytype=ref&siteid=sci.

Pigliucci, Massimo. 2009. On the scope of skeptical inquiry. Available online at http://rationallyspeaking.blogspot.com/2009/10/on-scope-of-skeptical-inquiry.html.

———. 2018. The problem with scientism. Blog of the APA. Available online at https://blog.apaonline.org/2018/01/25/the-problem-with-scientism/.

Russell, Bertrand. 1952. Is there a god? Unpublished. Excerpt available online at http://russellsteapot.net/.

Shermer, Michael. 2004. The Science of Good and Evil. New York, NY: Henry Holt and Company.

Tomasello, Michael. 2018. How we learned to put our fate in one another’s hands: The origins of morality. Scientific American. September: 70–75.

Yirka, Bob. 2012. CERN teams post Higgs boson papers—one ups its sigma level of certainty. Phys.org. Available online at https://phys.org/news/2012-08-cern-teams-higgs-boson-papers.html.