New research reveals decision-making pathways in the bee brain, and sheds light on the ability of honeybees to quickly and accurately assess flowers for honey, inspiring autonomous robot designs. The study, led by various academic experts, emphasizes the effectiveness of evolutionarily refined insect brains that could guide future AI development in industries. Credit: Theotime Collin
New study reveals how robots can be designed to think like bees.
Bees excel at balancing reward and risk, quickly determining which flowers can provide sustenance for their colony. A recent study published in the journal eLife Explains how eons of evolution have fine-tuned bees to make quick judgments while minimizing risk.
This research provides insights into the workings of insect minds, the evolution of human cognition, and improved robot design.
This paper presents a model of decision-making in bees and outlines the pathways in their brains that enable rapid decision-making. The study was led by Professor Andrew Baron from Macquarie University in Sydney and Dr Hadi Mapudi, Neville Dearden and Professor James Marshall from the University of Sheffield.
„Decision-making is at the heart of cognition,” Professor Baron says. „It’s the result of an evaluation of possible outcomes, and animal life is full of decisions. A bee has a brain smaller than a sesame seed. Yet she can make decisions faster and more accurately than we can. A robot programmed to do the work of a bee would need the backup of a supercomputer.
„Today’s autonomous robots often work with the support of remote computing,” Professor Baron continues. „Drones are relatively brainless and need to be in wireless communication with the data center. This technology path doesn’t allow a drone to truly explore. tuesday separate – NASAThe amazing rovers on Mars have traveled about 75 kilometers in years of exploration.
Bee. Credit: Theotime Collin
Bees must act quickly and efficiently, finding nectar and delivering it back to the hive while avoiding predators. They have to make decisions. Which flower contains honey? When they fly, they are only vulnerable to aerial attack. When they land to feed, they are vulnerable to spiders and other predators, some of which use camouflage to look like flowers.
„We trained 20 bees to recognize five different colored 'flower discs’. Blue flowers always contain nectar,” says Dr Mapudi. „Green flowers always contain quinine. [tonic water] It tastes bitter to bees. Other colors sometimes contained glucose.
„We then introduced each bee to a 'garden’ where the 'flowers’ contained distilled water. We filmed each bee and then watched over 40 hours of video, tracking the bees’ paths and how long it took them to make decisions.
„If bees were confident that food was available on a flower, they took an average of 0.6 seconds to quickly decide to land on it,” says Dr Mapudi. „If they believe a flower doesn’t have food, they make a quick decision.”
If they were unsure, they took longer—1.4 seconds on average—and time reflected the probability that a flower had food.
The team then built a computer model from first principles aimed at mimicking the bees’ decision-making process. They found that the structure of their computer model resembled the physical structure of a bee brain.
„Our study demonstrates complex autonomous decision-making with minimal neural circuitry,” says Professor Marshall. „Now that we know how bees make such smart decisions, we are studying how they are so fast at collecting and sampling information. We think that bees use their flight movements to improve their visual system when finding the best flowers.
AI researchers can learn a lot from insects and other 'simple’ animals. Millions of years of evolution have led to incredibly efficient brains with very low power requirements. The future of AI in industry will be inspired by biology, says Professor Marshall, who co-founded Opteron, a company that adapts insect brain algorithms to make machines run autonomously, just like nature.
Reference: “How Bees Make Decisions Quickly and Accurately” by Hadi Mapudi, James AR Marshall, Neville Dearden, and Andrew P. Barron, 27 June 2023. eLife.
DOI: 10.7554/eLife.86176