In Isaac Asimov's Foundation series, social science has advanced to become a powerful force. "Psychohistory" is a new field where social psychology is combined with historical statistics to be able to accurately predict the path of human societies: their political uprisings, economic cycles, and cultural movements. Psychohistory became a powerful tool to be developed in secret, much like nuclear physics of the last century. The ruling class would utilize psychohistory to manipulate the masses: anticipating social unrest and preempting political revolutions.
Obviously this is all science fiction. But it made me wonder; why hasn't social science developed more? The hard sciences have developed with incredible speed and refinement, from early theorists like Newton and Darwin to modern technologies like satellites and gene therapy. Moreover, there is a throughline from theory to application to sweeping socioeconomic change. Physics’ gas laws became practically applied in the steam engine and railroads, which in turn revolutionized commerce and war. Social science doesn’t seem to have that same dramatic evolution and impact. Perhaps psychology has improved mental health, and sociology has improved our ability to integrate diverse peoples, but that pales in comparison to stuff like plastics and nukes.
Back in college, I believed a social science revolution was coming. I believed that some basic building blocks, akin to Newtonian physics, would be established in neuroscience or microeconomics could provide a rigorous foundation and seed an explosion of progress in the social sciences. Heady ideas for a heady time. Let me recount my theory and reexamine it today.
---
Scientific development works best when the foundations are stable, ensuring that anything built on top is also stable. "Stand on the shoulders of giants" and all that. You don't want to stand on quicksand, but social science shifts so surreptitiously that you might as well be. In Psychology, each new iteration of the DSM (Diagnostic and Statistical Manual of Mental Disorders) uproots big chunks of the previous version. Macroeconomics models are always running into unforeseen conditions. Anthropologists are self-doubting navel-gazers, always asking "how are our own cultural conditions affecting our analysis of other cultures?". Young historians chase clout by undermining old historians. This whole ecosystem feels mercurial and ephemeral.
What social science needs is a solid foundation to build on: something akin to Newtonian physics or Origin of Species that offers a stable starting point. It may not get every detail right, but the core paradigm persists. Moreover, this foundation exists at a fundamental level: the theory of the atom, the quantification of energy and inertia, the guiding force of evolution. Researchers are empowered to investigate all the nooks and crannies, to refine and reinforce the core paradigm, rather than trying to flip it on its head.
Where to start? The natural barrier between hard and soft sciences seems to separate the brain and body. There's something about the human brain that is difficult for models of forces or atoms or cell signalling to capture. Let's lay down some new roots. How do humans think and make decisions? If we can understand that, perhaps we can extrapolate out to society, culture, and economics. Two approaches seem promising.
First, the brain itself. Psychology has always had the disadvantageous position of analyzing the brain from the outside, like an alchemist speculating on elementals. Without the ability to examine the underlying mechanics, we can only prod and observe a black box. Neuroscience promised an inside-first approach, measuring neurons firing on a cellular level rather than subjective descriptions of human behavior. Perhaps if we can unpack the complex machinery of the brain, we could someday determine which cell receptors triggered which neurons to fire and cause which psychological phenomenon.
Second, the behavior itself. We can circumvent the bizarre black box of the brain by using behavior as our basic building block. Microeconomic game theory studies decision-making under simple and constrained conditions. You have two options: A or B. I have two options: X or Y. Our outcomes depend on what we choose. What do you do? This basic structure spawns many simple games; prisoner's dilemma [link], hare-stag hunt [link], matching pennies [link], etc. These simple games can be recombined to construct more complex behavior, and the hope is to be able to understand complex decisions from simple building blocks.
For example, you have to make strategic decisions when you order at a restaurant, and when you go on a romantic date. The two scenarios are combined when ordering at a restaurant while on a romantic date. Perhaps if we understood your behavior in each individual scenario, we could combine them like building blocks to understand the combined scenario rather than having to study it specifically.
My dream as an aspiring microeconomist was to spearhead the social science revolution on the horizon.
---
It's been over a decade and neither prediction seems to have made any progress. Neuroscience seems to struggle to actually understand what's going on inside the brain. Their statistical methods seem suspect (see: dead fish paper). Even if we could see every single neuron firing, that would still not be sufficient. The artificial neural networks used in LLMs are fully observable, yet they are still infamously inscrutable black boxes that their creators don't fully understand.
The economic approach also has stalled. Behavioral economics has fallen since its heyday. Many of its purported experimental effects seemed promising, but failed to create any meaningful impact when scaled up (see: The Voltage Effect). Psychology studies are embarrassingly unreproducible, indicating incompetence or fraud. Game theoretic models work well when games are played once or infinitely, but struggle in more realistic scenarios when interactions are repeated a few dozen times.
In Asmiov's imagination, psychohistory utilizes large statistical aggregates to create predictable behavior. However, many complex systems are chaotic, and small changes created by statistical variance can cascade into large differences in outcomes. Like in meteorology, general climate patterns can be studied, but pinpoint predictions and manipulations are pure fantasy.
Ultimately, I think the intractable complexity is rooted in the brain. How we respond to stimuli and make decisions still feels very fuzzy, and none of our current approaches seem to understand that well. Without an understanding of the atoms of society, it seems unlikely that we'll make much progress on larger complex systems.