Beyond Top-Down & Bottom-Up: A New Theory for Modeling Complex Systems in Flux
(Published December 6, 2024)
We live in a world governed by complex systems – from the intricate dance of ecosystems and the spread of disease to the fluctuations of global economies and the fundamental laws of physics. These systems, found across disciplines like immunology, ecology, economics, and thermodynamics, share a common challenge: they are notoriously tough to predict and model. for decades, scientists have relied on either “bottom-up” or “top-down” approaches. But what happens when a system is disrupted? A wildfire reshaping a forest, a pandemic upending society, or a financial crisis rocking the markets - these are scenarios where traditional models fall short. Now, a groundbreaking new theory is emerging, offering a more holistic and accurate way to understand and predict the behaviour of complex systems undergoing change.
The Limitations of Traditional Modeling
Conventional modeling strategies typically operate in one direction. Bottom-up models start with the individual components of a system – the individual trees in a forest, the individual peopel in a population, the individual transactions in an economy – and attempt to predict the overall system behavior based on their interactions. Top-down models, conversely, begin with the system-level properties and attempt to explain the behavior of the individual components.
While both approaches have their strengths, they struggle to capture the crucial feedback loops inherent in disturbed systems. As John Harte,SFI External Professor at UC Berkeley,explains,”These unidirectional models can’t capture the interactions between the small-scale behaviors and the system-level properties.”
Such as, a standard epidemiological model (like the Susceptible-Infected-Recovered or SIR model) can estimate the probability of infection based on proximity. However, it fails to account for how rising case numbers might change individual behavior - prompting people to wear masks, practice social distancing, or get vaccinated – which then reduces infection rates. This interplay between micro-level actions and macro-level outcomes is precisely what traditional models miss.
A Hybrid Approach: Bridging the Gap Between Scales
Professor Harte and his collaborators have developed a novel hybrid method that elegantly integrates both bottom-up and top-down causation into a single, unified theory. Their work, recently published in Proceedings of the National Academy of Sciences (PNAS), outlines this approach and demonstrates its potential submission across diverse fields.
“Over the past 14 years, we’ve shown the power of a top-down approach in ecology, accurately predicting patterns like species diversity and abundance,” says Harte. “But we discovered that when a system is heavily disturbed, this approach breaks down. We needed a theory that could describe both the system-level dynamics and the probability distributions of the systemS components when the system is in flux.”
this new theory, initially presented in 2021 in Ecology Letters with their “DynaMETE” paper, leverages the principles of Maximum Entropy (MaxEnt) combined with mechanistic understanding. The team successfully tested DynaMETE against data from a disturbed forest in Panama, demonstrating its ability to accurately predict changes in species distribution. The current PNAS publication expands on this work, generalizing the model for broader application.
What Does This Mean? Predictive Power in a Changing World
the implications of this hybrid theory are meaningful. It allows researchers to calculate previously inaccessible parameters, specifically:
* Predicting System Trajectory: How will the overall system evolve over time in response to disturbance?
* Understanding Individual Responses: How will the probability distribution of individual components within the system change?
“This model allows us to calculate things that haven’t been calculable before,” Harte emphasizes. “In these bi-level systems, when there’s both top-down and bottom-up influence, how do you calculate, when the system is disturbed, how the system and the individuals will respond over time? there was not an adequate theory before.”
The potential applications are vast. Consider:
* economics: Modeling how consumer confidence (a system-level property) influences individual spending decisions, and how those decisions, in turn, impact economic growth.
* Pandemics: Predicting the spread of disease while accounting for behavioral changes in response to infection rates.
* Climate Change: Understanding how shifts in global temperatures affect species distribution and ecosystem dynamics.
* Thermodynamics: Addressing a long-standing challenge in nonequilibrium thermodynamics - predicting the probability distribution of molecular kinetic energies, a crucial factor in combustion processes. Harte proposes testing the theory in a combustion tank, a controlled laboratory setting.
The Road Ahead: Testing and Refinement
While promising, this theory is not without its challenges. H
