Researchers focused on understanding belief dynamics often use analogies to shed light on the intricate cognitive and social systems that explain our beliefs and how they evolve over time. For example, ideas can spread like a virus, ‘infecting’ individuals as they transition from person to person. We can be attracted to others who share similar viewpoints, much like magnets. Societal beliefs can undergo gradual changes until they reach a breaking point, propelling society into a new phase. A recent article delves into the advantages and drawbacks of various analogies frequently used to model these belief dynamics.
Researchers focused on understanding belief dynamics often use analogies to shed light on the intricate cognitive and social systems that explain our beliefs and how they evolve over time. For example, ideas can spread like a virus, ‘infecting’ individuals as they transition from person to person. We can be attracted to others who share similar viewpoints, much like magnets. Societal beliefs can undergo gradual changes until they reach a breaking point, propelling society into a new phase.
A recent study published in Trends in Cognitive Sciences by SFI Professor Mirta Galesic and External Professor Henrik Olsson, both affiliated with the Complexity Science Hub, examines both the benefits and risks associated with the frequent analogies utilized to model belief dynamics.
At SFI, it’s quite common for researchers from one discipline to borrow analogies from other fields. For instance, insights from physics have been applied to understand economic processes, while concepts from ecology have helped clarify scientific collaboration. Over the last century, computers were analogized to human cognition, but now we often use human thought to explain the functioning of large language models. “Analogies can be helpful, but they all have their limits. The key is to know when an analogy has been stretched too far,” Galesic points out.
One well-known analogy in belief dynamics is the Susceptible-Infected-Recovered (SIR) model, initially created in epidemiology. This model illustrates how a singular contagion spreads through a population and can be adapted to more complex situations, such as the idea that holding one belief might increase the likelihood of adopting another, similar to how a cold may heighten the risk of developing pneumonia.
Although analogies can offer valuable “conceptual mileage” by revealing properties that might have gone unnoticed, they carry “conceptual baggage” that might lead to flawed conclusions. Embracing an analogy and its corresponding model without acknowledging its limitations can result in poor decision-making or ineffective responses.
A significant limitation of the SIR model is that the way beliefs spread can differ markedly from virus transmission. Merely being exposed doesn’t guarantee that an idea will resonate. In the case of beliefs, repetition might be ineffective, especially if the ideas are significantly at odds with someone’s current beliefs. Moreover, ideas tend to propagate more readily among individuals who share similar beliefs and characteristics.
The authors also investigate the strengths and weaknesses of additional analogies in belief dynamics, such as ferromagnetism, thresholds, forces, evolution, weighted additive models, and Bayesian learning. While each analogy and its corresponding models offer valuable concepts and methods, none stand alone as wholly sufficient.
“We must scrutinize analogies seriously—determining what they can offer, what limitations they have, and what insights we can draw from them—to develop models that accurately predict and explain real-world belief dynamics,” Olsson states.
Rather than relying solely on a single analogy, it may be more beneficial to integrate insights from various sources, while being cognizant of their respective limitations. “Ultimately, what truly matters is the outcome that aids in explaining the natural phenomena in question,” Olsson emphasizes.
“We provide some guidelines for using analogies to create models of belief dynamics. First, identify the analogies, then translate them into quantitative models. It’s equally crucial to perform empirical testing and comparisons to assess the practicality and relevance of the models inspired by specific analogies,” Galesic adds.
