Collective Intelligence: How Insect Swarms Respond to External Stimuli

Arvin Anoop | arvin.anoop@yale.edu February 3, 2016

Collective Intelligence: How Insect Swarms Respond to External Stimuli

When people congregate into groups, two things can happen: they can either be highly intelligent like protesters during the Civil Rights movement, or exceptionally stupid like stampeding crowds. Science tells us that group behavior is not random. How individual units come together, what information each unit gains from another, and what actions the units perform as a group are all questions biologists, physicists, and engineers have aimed to answer by modeling groups as a whole. Most scientists have converged on one principle: Complex group behavior is often explained by surprisingly simple individual behavior.

 

In the past, research on group behavior has focused on highly ordered groups such as flocks of skylarks and shoals of golden shiner fish. Now, researchers led by former Yale associate professor of mechanical engineering Nicholas Ouellette are shaking up swarm science. They are examining how a seemingly random insect swarm responds to external perturbation. “From the swarm’s response, we can have a new way of understanding the dynamics of the graphic behavior of the animals,” said lead author Rui Ni, now assistant professor of mechanical engineering at Penn State.

In order to observe the swarm’s response to distress, the scientists first placed a colony of small flies known as midges in a closed cubic enclosure. They waited for them to spontaneously swarm.

Then, they captured movies of the swarms and computationally processed parameters such as velocity and trajectory. The insects were excited by playing the recorded sound of midges. This method was chosen because C. riparius midges are particularly sensitive to acoustic stimuli. For example, males can locate females because the sounds of female midge wing beats have a different frequency than those of males.

 


Midge mating swarms are nearly random when they form spontaneously and not driven by external stimulus but show specific, measurable patterns of motion when perturbed. Image courtesy of Flickr.

When female sounds were played, the insects all landed on the speaker because they thought the speaker was an attractive female. However, male sounds elicited a more fascinating group-level response. The center of mass of the swarm traced out specific elliptical trajectories moving away from the speaker as the volume increased and towards it as the volume decreased. “It’s like you don’t want to get too far from another male but at the same time, you don’t want to get too close. [The male midges] wanted to maintain personal space,” Ni said.

 

By unraveling group-level responses and providing more effective modeling techniques, this research has applications across engineering systems, especially to an exciting new subfield called swarm intelligence. “There are things we need to know before we can put any of this into real world applications, but in the long term, [research like this] can really help to optimize group-level behaviors,” Ni said.

Research about swarms can assist engineers in developing robotics technology. In this way, the present study could have far-reaching applications across medicine, disaster control, and further research. Image courtesy of Wikimedia Commons.

Research about swarms can assist engineers in developing robotics technology. In this way, the present study could have far-reaching applications across medicine, disaster control, and further research. Image courtesy of Wikimedia Commons.

For example, swarms of rescue robots could coordinate with each other and respond to contingencies like natural disasters, or swarms of micro-spacecraft could be launched into an asteroid belt in order to efficiently collect data on rocks of interest.

 

Despite this promise, it is clear that many questions in the field remain unanswered or unexplored. Ni suggests that a next step might be to observe group responses in different animals in different controlled environments perturbed by different signals. Hence, we might have to wait before we can scientifically engineer intelligent group behaviors and prevent detrimental ones.