Does Magic Work?

Does magic work?  I mean: does it work objectively, not merely psychologically. If you cast a spell for some goal (like health, or love, or money), does casting the spell change the physical universe outside of your body in some way that you intend?

Before trying to answer that question, we might want to draw some distinctions. Consider treatments for diseases.  Before regulatory bodies (like the FDA) approve a treatment, it has to demonstrate safety and effectiveness. It has to work reliably. Here reliability means something like this: a treatment (or a spell) works better than placebo in at least two large well-controlled randomized clinical trials. No spell works in this strong sense of reliable effectiveness.  This strong sense is the technical sense of effectiveness. A treatment has to work at least this well to be included in our medical technologies.

So consider a weaker type of effectiveness: The spell produces some statistically detectable change in at least two large well-controlled randomized clinical trials; but the change is no greater than placebo effects. A spell that works in this sense probably just works via placebo effects. This kind of magical effectiveness is purely subjective or psychological. Magic is treated as a psychological technology, a technology of the self. Still, no spells have ever been shown to work even in this weak technological sense.

There is an even weaker sense of effectiveness: A spell cast for some goal merely increases the probability that the goal will occur. Here we are really comparing two conditional probabilities. The probability of the goal given that you cast the spell is greater than the probability of the goal given that you don’t cast it. Suppose the probability of the goal given no spell is one in one trillion, but the probability of the goal given the spell is two in one trillion.  The spell increased the probability. In the weak sense, the spell works. But it does not work in any detectable way. It doesn’t work in any reliable way. This is a minimal sense of magical effectiveness. If spells work in this minimal sense, then spells are effective, but they are also entirely useless. Spells are not technologies.

Technologies (which are used for purposes) usually involve goals in the future. If you’re sick, your set of possible futures divides into two classes: those in which you continue to be sick, and those in which you get well. These classes can be defined using the statement “I will be healthy”.  In the class of futures where you stay sick, “I will be healthy” is false, while in the class of futures where you get well, “I will be healthy” is true. More generally, any statement P about your future divides your set of possible futures into a class in which it is true and a class in which it is false. A technology for health aims to increase the probability that “I will be healthy” is true.  The goal of any technology is to change the causal structure of the universe so that your actual future is more likely to be in your desired class of possible futures. Technologies work by cause and effect.

But spells are not technologies. They do not cause effects in the external universe. They might cause subjective or psychological effects in your brain and body, but those are not the effects under consideration here. If spells work at all, they work via non-causal links between present events and future events. Non-causal links involve conditional probabilities, and they are common in information theory. Consider two synchronized clocks. An information channel exists between them: information flows from each clock to the other; each clock carries information about the other. But neither clock causes any effect in the other. Flows of information do not require causality.

If magic works at all, it works through informational linkages. It works via correlations that carry information. Older theories of magic, from Pliny and Plotinus through the Middle Ages, talked about correlations involving sympathies and antipathies. So perhaps magic involves correlations. Such correlations do occur in physics, and they have been abused to produce pseudo-scientific theories of magic, specifically, theories that involve quantum entanglements. Magic does not work via quantum entanglements, and magic does not work via quantum consciousness. Magic has nothing to do with quantum mechanics. But quantum mechanical entanglements are not the only kinds of entanglements.

Our universe is filled with signs, signs which carry information about their referents. We are all familiar with the signs in human languages. According to the American philosopher Charles Sanders Peirce, these signs are symbols. Symbols depend on arbitrary social conventions. But Peirce recognized two other types of signs, namely, icons and indexes. An icon is a sign that looks like what it means. Icons are often found on gendered restroom doors: a stylized image of a woman and a stylized image of man. An index is a sign by association. Indexes can be social. Given the social conventions of America today, a tie can be an index of a man, and high heels an index of a woman.

Although icons and indexes can be conventional (that is, social constructions), they can also be objective. Objective indexes often appear in biological mimicry. An orchid is an objective icon of a female wasp.  A walking stick insect is an objective icon of a stick. Sunspots are an objective index of the disalignment of the planets.  The tides are an objective index of the position of the moon. Objective indexes occur in aposematic signaling, that is, warning coloration, in biology. The bright colors of certain insects and frogs are indexes of their toxicity. The high-contrast coloration on skunks and badgers is an index of their danger. The waggle dances of bees are both iconic and indexical.

Objective icons and indexes occur throughout our natural universe. They are especially common in biology, and they are effective signals that work reliably. Although biological signs work through physical cause and effect relations, they don’t work because of those relations. They work because those relations carry information. They work because of biological entanglements. An orchid that looks like a female wasp transmits the message: “I am a female wasp”. This message causes a male wasp to try to mate with the orchid, thus pollinating it. The orchids and the wasps are two species that have become entangled over the long course of evolution. Of course, the orchid doesn’t know anything about female wasps. Nevertheless, it has evolved to carry information about female wasps. The orchid species and the wasp species have evolved to become synchronized clocks.

Biological signs can help us understand magic. The orchid casts a spell on the wasp. It is precisely because the orchid is an icon that names the class of female wasps that the orchid gets pollinated. The icon semiotically conditions the wasp. Magic spells often use icons or indexes of their goals. Sympathetic magic works via indexes: when somebody lights a green candle to gain money, the color of the candle is an index. Sometimes spells use symbols. The tables of correspondences in grimoires set up symbolic conventions for magical languages.  And spells are typically procedural combinations of signs. They are recipes or algorithms, which use signs to describe processes.  A spell might depict the process of illness leaving the body or the process of gaining some desired object. The similarity of spells to algorithms has led many writers to point out that spells resemble computer programs. Computers, after all, are purely physical things whose behaviors are entirely defined by signs. Magic spells are signs that objectively name possible futures. A spell for health names the class of futures in which you are healthy.

Peirce used the term “semiotic” to refer to signs. If magic works at all, then it works through semiotic conditioning: signs made in the present condition the probabilities of future events. The semiotic relations between spells and their goals are informational relations; they are semiotic entanglements that supervene on physical relations, but they are not physical relations. They belong to the purely semiotic structure of our universe. If our universe has no semiotic structure, then spells do not work at all. However, it is an astonishing fact about our universe that it does have semiotic structure. Biological evolution discovers many semiotic entanglements, and uses them. Species (like wasps and orchids) become semiotically entangled. Even evolutionary arms races (like between cheetahs and antelope) are based on semiotic entanglements.   The emergence of semiotic entanglements between the earth and moon (the tides), and elsewhere in our solar system, show that semiotic entanglement is far more general than life.

Many writers have observed that our universe is finely-tuned for life. More generally, it is finely-tuned for the evolution of internal complexity. Finely-tuned systems, like eyes and wings, emerge through evolution. This leads to the hypothesis that our complex universe emerged in a long process of cosmological evolution, involving cosmic reproduction and selection.  So, as the result of a long process of cosmological evolution, our universe is finely-tuned for the internal evolution of complexity, including complex life. But life depends on the informational structure of our universe. Life requires the storage and transformation of information (encoded on earth in DNA and RNA).  Therefore, as universes become more finely-tuned for life, or for any kind of emergent complexity, they become more finely-tuned for informational correlations. The physics of a finely-tuned universe supports rich systems of semiotic entanglements. The more a universe is finely-tuned for evolution, the more it supports a richer semiotic structures. Consequently, the more a universe is finely-tuned for evolution, the more effective magic becomes in that universe. This is a fine-tuning argument that magic works. It does not entail that magic works detectably (it does not) nor that it works reliably (it absolutely does not).

Spells name classes of possible futures. Of course, this means that spells can either succeed or fail in the same way that linguistic utterances succeed or fail. A successful spell constructs an iconic and indexical model of the path from the present to the intended class of possible futures. If a spell is unclear, ambiguous, grammatically bungled, or just fails in any way to precisely name the class of intended futures, then it fails entirely.

Spells name classes of possible futures. If my reasoning here is correct, then the very act of naming some class of possible future makes it minimally more likely that the actual future will be a member of that class. Magic works, but magic is entirely useless. Magic is not a technology. Moreover, the way that magic works is entirely rational. Rationality itself depends on the semiotic structure of logic. Magic is rationality. The degree to which magic works in any universe is a measure of the rationality of that universe. Our universe has minimal rationality. Magic is entirely useless here. Spells are not techniques that reliably produce their intended goals. Spells are existence proofs: if you can cast a spell for some desired class of futures, properly using signs to name that class, then there exists at least one path from the present to that class of futures. The existence of that path is sufficient for hope. Magic means that hope is real.


Subscribe to The Spiritual Naturalist Society
Learn about Membership in the Spiritual Naturalist Society

The Spiritual Naturalist Society works to spread awareness of spiritual naturalism as a way of life, develop its thought and practice, and help bring together like-minded practitioners in fellowship.

SNS strives to include diverse voices within the spectrum of naturalistic spirituality. Authors will vary in their opinions, terms, and outlook. The views of no single author therefore necessarily reflect those of all Spiritual Naturalists or of SNS.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.