Pan narrans

A few years ago I put up a post on the conjecture that the human ability to throw things far and accurately is one of our distinguishing characteristics, and that developing this ability was likely a key step in our divergence from other lines of the chimp/bonobo family (not my own conjecture, see the post for more details). I think the post holds up fairly well, for whatever that's worth.

Towards the end I mentioned the idea that the advance planning required to throw a projectile at a moving target may have been key to our developing other unique capabilities like speech and "large-scale planning" (I'll call it "complex" planning here since that seems a bit better to me now). My take at the time was that there had to be more to it than that, and that still seems right, but regardless of how they arose, the capacity to make complex plans and describe them in language appears to be another cognitive ability that we have and everyone else seems not to.

Complex planning and the ability to describe it converge, along with a couple of other significant abilities, in one of the most human of all human behaviors: storytelling. I say "converge" because storytelling is not a single ability, much less the basis of other abilities. Rather, it pulls together a number of key abilities that make us who we are.

What makes for a good story? Two key features are

• Multiple things happen, in a given order (for you math majors out there, a partial order, since things can happen at the same time, even if they're generally narrated in some sequential order) ...
• ... but not only that, things happen for a reason. A real story involves cause and effect: "This happened, and because of it, that happened"^*

Cause and effect are the basis of planning: "Do this so that you can do that." To do make complex plans, you need to put together several pairs of causes and effects, with some effects being causes for other effects. This is the basis of plot, one of the key elements of storytelling.

I had originally written "string together a sequence" but then replaced it with "put together several pairs" because cause and effect do not necessarily run in straight lines. An effect can have more than one cause. A cause can have more than one effect. Causes and effect can run in both directions, leading to feedback loops -- to vicious (or virtuous) circles or to longer-term stability. Our brains can conceive of these and describe them. To make up some examples:

• Their success was a combination of skill, preparation and sheer blind luck
• The construction of the new road had far-reaching consequences, with some communities faring better than others
• The new vegetation reduced erosion, retaining more soil, which in turn supported more vegetation
• The shortage led to higher prices, which encouraged more production, eventually eliminating the shortage. Prices fell, leading to more demand, eventually leading to new shortages ...

While I wouldn't call any of those a full-fledged story, it's easy to see how any of them could be made into one.

In the real world, things don't always work out as planned. One way to deal with is to make plans conditional:

• If the weather's good, we'll go through the mountains. If it turns cold, it will be quicker and safer to go around.
• We'll have to wait here until the flood subsides and we can cross the river

Interesting stories involve choices, and many interesting stories revisit the same situation multiple times (the magic number is three, of course).

Both of the sentences above involve cause and effect -- It turned cold, so we went around; The flood subsided, so we could finally cross -- but also something more. They embody the two primitives that, one way or another, underly all models of computing:

• Conditional evaluation: How the result its computed (not just the result itself) depends on the state of the computation. For example, if n is even, the result is n/2, otherwise, it is 3n + 1
• Iteration: The same computation is repeated as the state of the computation evolves, for example: Repeat this computation, at each step using the result as the new n

If n is, say, 8 to begin with, then n becomes 4, then 2, then 1, then 4 (3·1 + 1), then 2, then 1 and so on forever. Other cases are less well-understood.

Given conditional evaluation and iteration, you can compute anything that can be computed (there's a lot more, of course, to what that actually means and implies). In a real sense, modern computers are simply automating something our brains do naturally. Other kinds of brains can handle conditionality and can do the same thing repeatedly, but we appear to be unique in how far we can extend this. To use a fairly shaky metaphor, we have the same kind of CPU but a lot more RAM.

Metaphor, that is, understanding one thing in terms of another, is another building block of storytelling, because it is key to how we understand the world in general. Metaphor is so pervasive that most of the time we don't even notice it until someone uses it deliberately. I've gone into that in several other posts, so I won't go into it here.

Finally, good stories need good characters. That is, they need to show a good understanding of what the people in the story are thinking, what their desires and motivations are, and often what they think about the other characters, including what they think those other characters are thinking. To tell a good story, you need a theory of mind and, for that matter, the ability to convey theories of mind that might be different from yours (the narrator sees one of the characters as a canny operator who's thinking two steps ahead of everyone else, while another of the characters thinks they're a simpleton).

Many interesting and exciting stories hinge on who knows what about whom, and who knows who knows what about whom. Irony, another key element of storytelling, occurs when one person -- in the story or not -- has a different understanding of what's going on from another person. For example, in dramatic irony, the reader/audience knows something that a character doesn't (usually not good news for the character), or the reader only thinks they know what's going on (often good news for the character and, in a different way, for the reader).

This ability to build detailed maps of who wants what and who knows what, including the "meta" level of who knows what about what others know and want, is obviously important for a highly social primate whose survival depends critically on getting along with the rest of the group, which in turn requires navigating its social structures.

Again, other brains are able to do this to at least some extent, but ours seem able to do this much more extensively, particularly at the meta level. More than that, we are able to communicate our understanding of other people: Look out for that person. They'll act like they're your friend, but they're really just trying to take advantage of you. But what they don't know is ...

Stories involve characters, that is, people (usually). Even though they may be fictional, we can identify with them and we care what happens to them. We want the protagonist to attain their goal, so strongly it can lead us to forget whether the goal or the person chasing it are any good. In other words, stories carry an emotional charge, so a good story is fundamentally more memorable than a recitation of facts.

It's striking how many of our characteristic cognitive abilities converge in storytelling. To recap:

• Storytelling is built on an understanding of cause and effect
• This extends to large webs (technically, directed graphs) of cause and effect, which can branch, join and form loops
• Stories involve choices and repetition, the most fundamental elements of computing as we know it
• Stories are built on metaphor, which is key to our ability to learn
• Telling a story with characters requires a detailed understanding of what people want and know, including the meta level of what people know about what other people want and know (or, for that matter, what they want people to know, and so on)
• Irony is built on understanding of that meta level
• Understanding characters requires social understanding
• The ability to communicate complex cognitive structures itself an important cognitive ability
• We care what happens in stories, which makes them memorable

In other words, storytelling is a manifestation of the constellation of cognitive abilities that are a large part of our uniqueness. Moreover, the ability to communicate these things, in detail, to others is itself a key cognitive ability.

I think there's one more thing. Not only are the elements that storytelling is built on key elements of our cognitive abilities, to a large extent we understand the world in terms of stories. That is, not only do we understand the world through cause and effect, choices, repetition, metaphor and theories of mind as capabilities on their own, but to a large extent we understand the world in terms of stories that pull together those elements in a particular way: a protagonist has a goal, encounters obstacles, finds ways to overcome them (or occasionally gets lucky) and (usually) ultimately attains the goal.

Persuasive advertising tells a story (My shirts weren't coming out of the wash looking bright and clean. I started using this detergent (meaning, I bought it). Now my shirts look bright and clean). So does outright propaganda. But even something seemingly factual like an essay on human cognition uses storytelling: I had originally written "string together a sequence" but then replaced it with "put together several pairs" (because the original version didn't fit with my goal of explaining webs of cause and effect).

Given that the ability to communicate complex cognitive structures is important in and of itself, it's not surprising that our minds would favor stories as a means not only of communication, but understanding. 

However, this has its downsides. One common failure mode of our brains is to tell a story where there isn't really one to tell. For example, a financial news item very rarely just says a price went up or down. It almost always gives a reason. Often it will tell of a struggle between bullish factors driving the price up and bearish ones driving it down.

In many cases it's not hard to accept this as a shorthand for something more like "these factors could drive the price up, these could drive it down, both are in operation and we don't have enough information to know in advance what will happen". Often, though, in financial news and elsewhere, the story takes over and continues in episode after episode. Once the narrative is accepted, it's more cognitive work to reconsider whether the narrative actually applies. The consequences of leaving the assumptions unquestioned can range from trivial to tragic.

For better or worse, storytelling isn't just a reflection of individual capabilities like sequencing or theory of mind. It reflects those capabilities being used together in well-established, identifiable patterns. As such, storytelling itself is a fundamental part of our minds, distinct from the many capabilities that enable it.

^* Even the famous six-word story "For sale: baby shoes, never worn" is full of cause and effect.