Analyzing the brains of London's taxi drivers could be key to improving AI navigation systems and Alzheimer’s research, scientists say. University College London (UCL) researchers have found that London taxi drivers develop a larger posterior hippocampus (the part of the brain responsible for memory and navigation) due to years of training.
Each driver must pass a rigorous exam known as ‘The Knowledge’, memorizing over 26,000 streets to navigate the city efficiently.
Unlike app-based navigation systems, taxi drivers don’t rely on satnavs. Instead, they use years of training and experience to plan the best possible routes in real-time.
Now, researchers believe their unique approach to route planning could inform the development of artificial intelligence, and further research into Alzheimer’s disease.
For Marc Manning, driving a taxi in London is more than just a job; it’s a skill that has taken him years to master. “I love my job. 25 years and I still love it. Very proud of it. Very proud of the fact that, you know, I’m one of a very small number that can actually do The Knowledge.”
He says London’s taxi drivers rely on a deep understanding of the city’s streets, something which no satnav can fully replicate.
Now, scientists are studying how this expertise in route planning could help improve AI.
Hugo Spiers, Professor of Cognitive Neuroscience at UCL says, “In this study, we asked London taxi drivers to plan routes through the city, saying each street out loud as they worked it out. Sometimes they were very quick, sometimes much slower. We used models to understand why—and found that the layout of London’s streets plays a major role. It shows just how efficient these taxi drivers are, and it gives us new insights into human navigation. That knowledge could help us develop better AI systems in the future.”
Unlike satnavs, which calculate multiple routes before choosing one, taxi drivers prioritize the most complex parts of the journey first, and then work around them.
This article was provided by The Associated Press.