Which is a better magic trick: turning a dove into a glass of milk, or a glass of milk into a dove? Turning a rose into a vase, or a vase into a rose?
For most people, the way these transformations go makes a big difference. In each case, they find the transformation from a nonliving object to a living thing more interesting — but why? Is it just more exciting to see a living thing appear than to have it vanish? Or is there something deeper at work?
In a paper forthcoming in the journal Cognition, Tom Griffiths*, a cognitive scientist at University of California, Berkeley, argues that these intuitions about magical transformations reveal our ontological commitments about the world — that is, our deep-seated assumptions about the kinds of things that exist and how they're related. The upshot, the paper suggests, is that we can learn about a fundamental aspect of human cognition through magic.
What do our ontological commitments look like? For many years, cognitive scientists have suggested that some of our ontological beliefs can be represented in the form of a hierarchy: At the broadest level we have objects, which include both living and nonliving things; within the category of living things, we have plants and animals; within the category of animals we have humans and nonhumans, and so on.
Following this hierarchy, the properties that apply at one level also apply to those below it. So, for instance, "objects" have the property of having mass, and so do all living things, all animals and all humans. But properties at a lower level can't necessarily travel "up" the hierarchy. For example, you can say that plants and animals have the property of being alive or being dead, but you can't say that a brick is alive or dead — to do so would be metaphorical, or what's called a "category mistake." Similarly, you can say that a person is honest or dishonest (honesty is a property that can meaningfully be claimed of people), but you can't say that a daffodil (a plant) is honest, or that a brick (a non-living object) is dishonest. Several researchers — most prominently Frank Keil at Yale University — have provided evidence for the psychological reality of an ontological hierarchy along these lines.
To investigate the relationship between people's ontological commitments and their intuitions about magic, Griffiths conducted a series of experiments — partially inspired by studies that Keil performed with children — in which he had adult participants judge which kinds of transformations make for the most interesting magic tricks. In some cases, the transformation moved "up" the hierarchy, like a dove (a living organism) turning into a glass of milk (a non-living object). In other cases, the transformations moved "down" the hierarchy, like a vase (an object) turning into a rose (a living organism). Griffiths found that people in the study judged the latter significantly more interesting than the former, suggesting that it's the addition of ontologically relevant properties (like becoming the sort of thing that is dead or alive, or the sort of thing that can be honest or dishonest) that makes for an interesting magic trick.
Importantly, Griffiths was able to rule out some alternative explanations for why this might be. For example, the effect can't be explained by the fact that having a dove appear is just more interesting than having a glass of milk appear. It seems that the type of transformation involved matters, and not simply the start or end product. As a result, intuitions about magical transformations can tell us something about how people represent the relationships between different ontological categories — their place in the hierarchy. Knowing this, we can use intuitions about which magic tricks are most interesting to map out previously uncharted ontological domains. For example, what are people's ontological commitments when it comes to domains like the paranormal? If Griffiths is right, we can find out by testing people's intuitions about how interesting they find magical transformations involving relevant entities, such as turning a living person into a ghost, versus a ghost into a living person.
But is there really something special about magical transformations?
In two subsequent experiments, Griffiths was able to isolate the key aspect of people's judgments that mapped onto the ontological hierarchy. These studies showed that it isn't making judgments about magic, per se, that reflects ontological commitments, but more generally making judgments about which kinds of transformations are interesting. So, for example, people's judgments about how interesting they found different machines that enacted each transformation closely matched those for how interesting they found the corresponding magic tricks, but judgments about how plausible they found each hypothetical machine had different characteristics. In his paper, Griffiths quotes Arthur C. Clarke: "Any sufficiently advanced technology is indistinguishable from magic." When it comes to reflecting our ontological commitments, it's not about judging magic or machines, but about what compels us to say, "Well, that's interesting."
To me, this is the most intriguing aspect of the findings from Griffiths's paper. It raises important questions about what drives human interest — in other words, about what we decide to pay attention to in a complex world.
The divergence between ratings of plausibility and those of "interestingness" suggests that interest tracks something beyond our assessments of straight probability. Instead, the findings suggest that we're especially compelled by observations that violate our assumptions about the structure of the world. These observations are the ones to attend to, because sometimes they mean we got something wrong. Perhaps only a little wrong (for example, the transformation we thought we observed was merely a trick), perhaps very wrong (for example, we were wrong about the ontological structure of the world). And when we're wrong, there's an opportunity to learn something new. It's a signal to start paying attention.
* Full disclosure: Tom Griffiths is my husband as well as my colleague at University of California, Berkeley. I'm writing about this research because I still think it's fascinating despite having heard about it repeatedly — for years. We were both amused last week when our 4-year-old, wearing a fairy costume and carrying a wand, began "transforming" things around the house and consistently moved down the ontological hierarchy in doing so, as predicted. (She also claimed to freeze us a few times, but for that we blame Disney, not her ontological commitments.)
Tania Lombrozo is a psychology professor at the University of California, Berkeley. She writes about psychology, cognitive science and philosophy, with occasional forays into parenting and veganism. You can keep up with more of what she is thinking on Twitter: @TaniaLombrozo