Our brains are still smarter than AI! Despite the incredible advancements in artificial intelligence, there's one area where the human brain still reigns supreme: the ability to learn and adapt across different tasks. A recent study sheds light on how we achieve this remarkable feat.
Instead of studying humans directly, researchers from Princeton University turned to our close biological cousins: rhesus macaques. These monkeys were tasked with identifying shapes and colors on a screen, responding by looking in specific directions. During these tasks, brain scans were used to monitor the activity and search for overlapping patterns within the monkeys' brains.
These scans revealed something fascinating: the monkey brains utilized different groups of neurons – what the researchers cleverly called 'cognitive Legos' – across various tasks. These existing blocks can be repurposed and recombined for new tasks, demonstrating a level of neural flexibility that even the most advanced AI models struggle to match.
"State-of-the-art AI models can reach human, or even super-human, performance on individual tasks," explains neuroscientist Tim Buschman from Princeton University. "But they struggle to learn and perform many different tasks." The brain's ability to reuse components of cognition in many different tasks is what gives it the edge. By snapping together these 'cognitive Legos', the brain can build new tasks.
In the video, the animals had to differentiate between shapes and colors in three separate but related tasks, requiring them to constantly learn and apply what they knew from one task to the next.
The cognitive Lego blocks identified by the researchers were primarily concentrated in the prefrontal cortex, a brain region associated with higher-level thinking, such as problem-solving, planning, and decision-making. This area seems to play a crucial role in cognitive flexibility.
Interestingly, the researchers also found that when certain cognitive blocks weren't needed, the brain reduced their activity. This suggests that the brain can essentially 'file away' neural Legos that aren't immediately necessary, allowing it to focus more efficiently on the task at hand.
Buschman explains, "I think about a cognitive block like a function in a computer program. One set of neurons might discriminate color, and its output can be mapped onto another function that drives an action. That organization allows the brain to perform a task by sequentially performing each component of that task."
This explains how monkeys, and potentially humans, can adapt to new challenges and tasks by leveraging existing knowledge. This is something that AI, in its current form, has difficulty with.
But here's where it gets controversial... The researchers suggest their findings could help train AIs to become more adaptable. Their work could also be beneficial in developing treatments for neurological and psychiatric disorders where individuals struggle to apply skills in new situations.
For now, these cognitive Legos demonstrate how our brains are more flexible and adaptable than AI models. AI models often exhibit 'catastrophic forgetting,' a weakness where neural networks can't learn consecutive tasks without forgetting the previous ones.
While multitasking isn't always beneficial for our brains, applying what we know from one task to another can be a valuable shortcut. The researchers conclude that if the brain can reuse representations and computations across tasks, this could allow us to rapidly adapt to changes in the environment, either through reward feedback or by recalling information from long-term memory.
The research was published in Nature.
What do you think? Do you believe that the human brain's adaptability will always outsmart AI? Share your thoughts in the comments below!
