Behaving in Geometric Collectives
- neuroversecc
- Oct 9
- 5 min read
When a locust swarm sweeps across fields, it looks like a single coordinated organism. But each insect is following its own set of rules – sensing, moving and responding to its environment. From the collective actions of each individual insect, order emerges.
By studying the collective dynamics of animals, we can better understand how intelligence itself is built from ... small parts that together become more than the sum of their actions
In episode 97 of Neuroverse, Dr Iain Couzin joins Clara & Carolina to explore how collective intelligence arises in organisms such as locusts. While locusts are solitary creatures, they undergo dramatic physiological and behavioral transformations when population density rises. They change colors, metabolism, and enter a social phase that facilitates large scale coordination.
Couzin’s research employs advanced tracking systems and computational modelling, revealing a key principle: complex collective outcomes do not require centralized controls. Individual actions, structured by simple interaction rules, can scale into group intelligence. This principle extends beyond locusts. It resonates across biological systems – from cells to neural circuits in the brain, to the behavior of humans within societies. By studying swarms, scientists gain insights into how distributed systems organize themselves and solve complex problems without a central authority. This understanding helps establish a foundation for understanding the universality of collective computation.
The Geometry of Information Flow
Observing patterns of decision making in swarms and flocks of birds show how the geometry of connections – who interacts with whom, and how – shapes the flow of information in groups.
Locust swarms operate with a feedforward geometry: motion is across one direction, triggered by localized threats. While this ensures speed, it sacrifices flexibility. Locusts may be able to change direction, but are less capable of integrating conflicting signals or correcting errors once momentum is built. The human nervous system illustrates a similar principle. Reflex arcs bypass higher-order cortical processing to prioritize rapid action, allowing an organism to respond instantly to danger. While reflexes provide critical speed, they cannot weigh complex evidence, integrate context, or adapt flexibly to novel situations.
By contrast, bird flocks or fish schools often operate with a recurrent network geometry. In these systems, information circulates among multiple neighbors before a collective response is enacted. This enables the integration of diverse signals, allowing the group to balance competing inputs and correct errors dynamically. The result is not speed, but robustness – where no single noisy neuron or an outlier can tip the balance on its own.
These examples highlight a trade-off present in all collective systems: feedforward architectures optimize for rapid action, whereas recurrent systems optimize for accuracy and robustness. Both strategies recur across scales—from insect swarms to neural networks—suggesting that collective intelligence, regardless of substrate, relies on similar organizational principles.
Seeing, Simulating & Replicating Nature
How can we go about exploring what guides collective behaviors? A deep understanding of collective behaviors requires us to not only observe them, but also simulate and replicate them so that we can more directly investigate factors that may be involved, without disrupting nature.
Historically, animal behavior was studied merely as fleeting spectacles, which can be too fast and complex for the human eye to make sense of. Today, technologies such as drones, high-speed imaging, and LED-based tracking allow scientists to monitor thousands of organisms simultaneously. This detailed data captures how individuals interact, revealing the rules that govern collective behavior.
Simulation allows researchers to test hypotheses derived from observation. Computational models formalize how local rules (attraction, repulsion, and alignment) scale into emergent group dynamics. Mathematical frameworks, such as the Fokker–Planck equation, describe probabilistic distributions of movement and behavior over time, enabling scientists to predict how variability and randomness influence collective outcomes.
Finally, scientists move to replication, by translating biological insights into engineered systems. Swarm robots use the principles derived from insect and animal behavior to design autonomous robots that organize information, adapt to their environment, and complete coordinated tasks. This process helps form a comprehensive framework to understand decision making processes in both natural and artificial systems.
Noise and Error as a Strategy
While models and robots reveal the rules of coordination, nature reminds us that order is never perfect. Biological systems are riddled with randomness and noise, but this variability is not always a weakness. Noise acts as a built-in strategy, allowing collectives to remain flexible, adaptable, and resilient in the face of uncertainty.
In ant colonies, for example, most individuals follow chemical trails to reach food sources efficiently. However, some ants deviate, wandering into unexplored areas. These deviations, though seemingly inefficient, allow the colony to discover new resources, balancing exploitation of known paths with exploration of new opportunities.
The same principles observed in locust swarms also operate at the cellular and neural level, illustrating the universality of noise in collective systems. At the cellular level, gene expression (the process by which DNA is converted into proteins) is not deterministic in nature. It generates variability among otherwise identical cells, fostering resilience and adaptability within tissues. Similarly in neural circuits, noise prevents rigid behavior, enabling the system to remain receptive to new information and to explore alternative responses during decision making processes.
From Swarms to Societies
The lesson is that order in complex systems does not emerge from command, but from interaction
Humans have the potential to pool knowledge with great effectiveness, but are also prone to failures. This contrast has piqued the interests of both social scientists and neuroscientists for centuries. For instance, Condorcet’s jury theorem provides a glimpse into what should be possible: If each member makes an independent judgement with a better than random accuracy, then the chance that the pooled judgement would be right, is more than likely. When scaled up, this would demonstrate how crowd decision making is accurate. However, independence, a critical assumption of this theory, is fragile. Human decisions are rarely made in isolation. It is subject to influence, social conformity, and a plethora of cognitive biases.
This highlights deeper parallels linking swarms, brains, and societies. The lesson is that order in complex systems does not emerge from command, but from interaction. Recognizing this universality allows us to see that the difference is not in kind, but in scale. By studying the collective dynamics of insects, animals, and artificial systems, we can better understand how intelligence itself is built from the bottom up and out of small parts that together become more than the sum of their actions.
Listen to the episode here
This article was written by Purnima BR and edited by Clara Lenherr
Resources
Dussutour, A., Beekman, M., Nicolis, S. C., & Meyer, B. (2009). Noise improves collective decision-making by ants in dynamic environments. Proceedings of the Royal Society B: Biological Sciences, 276(1677), 4353–4361. https://doi.org/10.1098/rspb.2009.1235
Sayin, S., Couzin-Fuchs, E., Petelski, I., Yannick Günzel, Salahshour, M., Lee, C.-Y., Graving, J. M., Li, L., Deussen, O., Sword, G. A., & Couzin, I. D. (2025). The behavioral mechanisms governing collective motion in swarming locusts. Science, 387(6737), 995–1000. https://doi.org/10.1126/science.adq7832
Miller, N., Garnier, S., Hartnett, A. T., & Couzin, I. D. (2013). Both information and social cohesion determine collective decisions in animal groups. Proceedings of the National Academy of Sciences, 110(13), 5263–5268. https://doi.org/10.1073/pnas.1217513110
Seeme, F., Green, D., & Kopp, C. (2025). Ignorance of the crowd: dysfunctional thinking in social networks. Frontiers in Communication, 10. https://doi.org/10.3389/fcomm.2025.1547489
Berlinger, F., Gauci, M., & Nagpal, R. (2021). Implicit coordination for 3D underwater collective behaviors in a fish-inspired robot swarm. Science Robotics, 6(50). https://doi.org/10.1126/scirobotics.abd8668
Bahrami, B., Olsen, K., Bang, D., Roepstorff, A., Rees, G., & Frith, C. (2012). What failure in collective decision-making tells us about metacognition. Philosophical Transactions of the Royal Society B: Biological Sciences, 367(1594), 1350–1365. https://doi.org/10.1098/rstb.2011.0420
Comments