The Narrative Fallacy
We love stories and narratives. We are adept at constantly interpreting what happens around us and to us. This is not a new idea. It was explored by many philosophers, writers, and psychologists. But sometimes we interpret too much, our need for a narrative overruns our logic. We can’t help it. The need to interpret and make up stories is too strong. In his book The Black Swan, Nassim Taleb gives us an example. Let’s consider these two statements:
Joey seemed happily married. He killed his wife. Joey seemed happily married. He killed his wife to get her inheritance.
The second statement seems to make more sense and is more memorable, because it ties together apparently disparate facts with a narrative. The narrative gives us a thread to glue together facts that would otherwise be mere cacophony, and we can hold on to this thread to understand a cause and effect, we have a narrative that makes sense instead of the two random facts “happily married” and “killed his wife”. When the two events are presented to us without any glue to link them, they feel unrelated, it is just a jarring and contradictory juxtaposition: why would he kill his wife if he was happily married? That doesn’t make sense! If we remember these facts, their incongruity and the lack of an explanatory narrative will be salient in our memory. We will remember that something is missing, a cause for Joey to kill his wife, whereas the second statement is satisfactory: it tells us what the cause was. We are satisfied that we understand why he killed his wife. But we can’t be satisfied if we don’t have a narrative with causes and effects clearly making sense to us.
Taleb makes the point that sometimes our need for causal narratives is so great that we will make up one when there is none, when random chance was the real “cause” of what happened, and he calls that the “narrative fallacy”. If presented with the first two facts above without being given a reason for Joey’s behavior, we would ask “why?”, and maybe even think “hmmm, he must have wanted her money”, or “in fact, he didn’t really love his wife, the happy marriage was just a front”. We would instinctively supply hypothetical causes to explain the situation, get over the contradiction implied by the raw facts, and maybe even have a “hunch” that one of them must be the case. That hunch may even color how we later act. Sometimes we supply a narrative when there is none. And we prefer simpler narratives.
Let’s look at another example to show how our need for simplified narratives can sometimes lead us to conclude more than is the case. This second example is a famous experiment related by the psychologist and economist Daniel Kahneman in his book: Thinking: Fast, and Slow. In that experiment, college students at several prestigious colleges in the United States were asked to read a story and then choose between two possible conclusions. The story was:
Linda is thirty-one years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in antinuclear demonstrations.
and the two conclusions the students could choose from were:
1. Linda is a bank teller. 2. Linda is a bank teller and is active in the feminist movement.
The study showed that around 85% to 90% of the students found the second conclusion more appealing. In the first conclusion, the fact that Linda became a bank teller is surprising, and nothing in the story prepared us for this new fact. After all, she majored in philosophy, not accounting or any other field related to banking. The path from an education in philosophy to a job as a bank teller is somewhat counter-intuitive, something random and unforeseeable must have happened to Linda, it is not what we would expect, it is surprising and unexplainable. Because the conclusion doesn’t seem to follow particularly well from the story, we are more inclined to reject it. In contrast, in the second conclusion, the story we have just read makes it likely that Linda would be active in the feminist movement. The story primes us to see a reasonable link. It “makes sense”, so we prefer that second conclusion, and even overlook the lack of connection between philosophy and banking, which we couldn’t get over in the first conclusion. The element of chance that must have happened in Linda’s story is downplayed in the second conclusion because we have a narrative.
As an aside about chance, several authors have made the point that we tend to discount chance in our narratives, that we strongly prefer causal links. Uncertainty can be dangerous, or at least unpleasant. We do not like to see our limitations, that we are in fact subject to forces outside our control, that we have less agency in our lives than we think we do. A causal narrative is more reassuring.
The fact that we sometimes go too far in our quest for narratives only highlights that we cannot function without narratives. This could be seen as some sort of cognitive bias, something to be aware of, to be fought and countered, but it is in fact constitutive of who we are. And it is not a bad thing. We need narratives to condense series of events into an average blueprint of what happened, a “model” of the world. Narratives are a type of model that mirrors the world and without which we could not function. We are stories, we model the world in terms of stories, we hold the world as a set of stories in our heads.