“Man is the Storytelling Animal, and that in stories are his identity, his meaning, and his lifeblood…” - Salman Rushdie, in Luka and the Fire of Life
“Socrates’ speaks” – Louis Joseph Lebrun (1867)
We're told fascinating stories of science about how things work: how dinosaurs roamed the earth millions of years ago, how putting atoms together gives us gases like carbon dioxide or liquids like water, how genes determine certain biological traits… And they're particularly fascinating stories because they're factual – at least for me!
A common criticism that has come up in recent years is that scientific publishing has a bit *too much* storytelling, that the research that gets the most attention and published in the most prestigious journals, may not be based on the quality but rather on how well the authors sell the work.
This made me wonder: is good storytelling in science a bad thing?
In this blog post, I will discuss the idea that the role of storytelling in science goes deeper: rather than to be used as a device to report the discovery, storytelling might be deeply rooted in the scientific process itself, including theory building, creative hypothesis generation, problem solving, and in the discovery itself. That rather than regulating it, we should be encouraging it in the scientific process, but at the same time discourage the bad incentives and practices in science.
Good storytelling and bad incentives in academic publishing
The criticism is that scientific publishing has overemphasized the importance of storytelling. To get the attention of journal editors, the work must be novel, sound exciting to scientists, and make a broad impact on the field. To catch the attention of the popular press, it has to sound exciting to the public, and potentially have an impact on society. This means scientists have to sell their work, hard, even if the findings are not as spectacular as they claim. For those of us who know the pressures to publish, this is old news. Those who publish in big, "high-impact" journals get rewarded with prestige, grant money, and jobs. So the incentives for publishing in these venues are huge. This is not my focus, but it's an important issue to raise as this is typically what comes to mind when people talk about storytelling in science. It relates to the so-called 'replication crisis' in various empirical sciences, as people cherry pick results so they can publish, even when the data don’t support their conclusions.
Among the discussions that were happening (including on Twitter), I saw academics, often senior and high-profile professors, who spoke out with a different view: that good writing is important for science, and that a good narrative is key in a good article or presentation. In one sense this is obvious – if the writing is bad, where the ideas are incomprehensible and no one understands it, that's not great for science. It's only when it goes overboard (which it does), that this is a problem. And the field is working on this.
Whilst pondering this issue, I wondered: why is there such an emphasis on storytelling in science? And are there are positives to the fact that we’re often so obsessed with a good story?
Storytelling in science: more than just a good story?
I wondered if there might be something deeper to why we are obsessed with the idea of storytelling and a good narrative. It led me to think: could storytelling also be a crucial part of the scientific process? Almost all the time, we’re try put the pieces together and try to come up with theories of how things 'work', after which we turn to our instruments to test these theories. Rather than the role of storytelling exclusively as a reporting device, it also seems to be a key part of the scientific process. Maybe story generation is part of the process to produce explanations that allow us to consider the many different ways of how things might work.
So, where does storytelling happen in the scientific process?
To take an example from my own field, we might see an article claim something like: we found that a brain region X is active when people successfully navigate a maze to a goal (e.g., playing a game during a functional MRI scan). But how? What are the psychological and brain mechanisms that support navigation? The raw data (MRI images) can’t tell you anything. Before thinking about how the data is analyzed, we must step back. We might consider more basic questions: How is navigation possible? We probably need some cognitive ingredients: short-term memory, self-localization, long-term memory, a mental map… and there must be a precise way these things are put together. Then you might search for these things in the brain and if they come together to create the behaviour. Maybe you’ll find short-term memory representations of the goal location in some brain region. Maybe there’s a hint of a ‘map’ being formed during learning, and the brain uses that map to navigate to the goal. Maybe various brain regions responsible for short-term memory and goal representations work together.
Even from this simple example you can see you need to work to put things together creatively, to make sense out of it. In a compelling report, the authors often tell you a story, the big picture and all the little stories of how the pieces fit together. We need a theory – a coherent story, of how it's even possible to solve this (navigation) problem. We might even formalize it in a computational model (a set of equations that implements the underlying processes: e.g., a short-term memory store, a map construction process, combines them and outputs the directions). Then we might try to find the pieces (cognitive constructs, or model processes/variables) in the data. Some argue that we should "just look at the data". Though data can provide hints as to what the pieces are, and they can help modify our theory, we still need to generate the story (or theory) to construct a satisfying explanation. When we do science, we want a satisfying explanation. I’d go as far to say we want a convincing story of how the phenomenon at hand arises.
Whilst writing this I realized a lot of it is about theory building, which is a crucial part of science (e.g., van Rooij & Baggio (2021) and Guest & Martin (2021) – who I have been highly influenced by). But here I’m emphasizing the creative, story-generating, putting-the-puzzle-together side of the picture that extends beyond theory building to data analysis and the writing process. As people say, it's difficult to teach creativity (impossible?), and many will agree that scientific discovery requires creativity. But what does that mean? I think it might be about how well we can generate theories, how good we are at telling ourselves a story, changing the plotlines, consider the ways in which the pieces are put together, adding new pieces, and testing them. And formalization can help, as you can build a model that implements your theory, and test what happens when you change the way the pieces are put together.
The key is not to get too stuck in your own theories and ignore *all* data. It can sometimes be valid (like when it's not reliable or you can't be sure; one example is Crick and Watson when working on poor images of DNA. To paraphrase from memory: when data gets in the way of theory). But sometimes it's not – and it's up to us to figure it out. Good methodology and statistics help, but in the end it's up to us. Science is not simply about data analysis, but thinking, theorizing, and creative mental storytelling (possibly with a good dose of daydreaming) – for good science to be possible!
Final thoughts: Destined for the garden of forking paths?
Most likely, our desire for a good story extends way beyond scientific practice. As the quote at the top suggests, storytelling is in our nature – it's how our minds work! We always try to find an explanation for everything, even if we sometimes get it wrong. We like stories. They make things easier to understand, easier to remember, and much more interesting. That's why good books and films must have a good narrative that both engages the audience, and to help us understand what’s going on.
When we ask questions during the scientific process, we also strive for satisfying explanations. And often that means there should be a story to tell, and one that is engaging. In Cognitive Psychology and Neuroscience, we create theories – with mental constructs, or cognitive capacities, algorithms, neural mechanisms, which are aimed at providing good explanations for our behaviour – multi-part stories that help us to comprehend the issue at hand. When done well, theories can let us predict things and make causal statements. And the actual building of the explanation is key.
Thinking about the process and the creative aspect of theory building, where we try to construct a story, a narrative of how things happen, really makes me think that it is a bit like creating a universe in our heads, like how writers imagine their fictional worlds. And we can consider many, many hypothetical universes. So let us embrace our storytelling animal in the scientific process! Let us dream a bit, consider multiple forking paths, and once in a while, it might lead us to a genuine, exciting discovery.
Good science requires good storytelling. We should be firm on good methodology and statistical inference, but the story is equally important. It must flow, make sense, and be a satisfying explanation – from the process of theory generation, to understanding the data through hypothesis testing and data exploration, to the final version of the published article. As of any good story, it should be engaging and helps us understand the world that the writer is trying to portray. Except that instead of a fictional world, it’s the fascinating world we live in.
