From Brain to Machine: How Neuroscience is Reshaping Artificial Intelligence

This piece, by Selina Hui, was published on 04/29/25.

What if the key to building smarter, more adaptable artificial intelligence (AI) lies within our brains? At the crossroads of neuroscience and AI, scientists are uncovering how the brain's remarkable ability to learn and adapt drives breakthroughs in machine learning and reshapes the future of technology. AI systems mimic human intelligence by analyzing large datasets and identifying patterns to improve over time. As Dr. Kenneth Miller from Columbia University explains, "Modern AI systems are based on the idea that computations in brains occur without explicit logic or symbols, through neurons summing inputs from other neurons, and that learning occurs by resetting synaptic strengths between neurons."

The concept of 'machine intelligence' emerged in 1949 when Edmund Callis Berkley published his paper comparing early computers to human brains. This idea gained momentum in the 1950s with Alan Turing's famous Turing Test, which evaluated whether a machine could exhibit behavior similar to humans. In 1958, Frank Rosenblatt introduced the perceptron, a model inspired by biological neurons that used a single layer of artificial neurons to process inputs and make decisions. Although limited by the computing power at the time, this innovation laid the foundation for today's image recognition technology, which uses multi-layered neural networks to analyze data more effectively.

Artificial Neural Networks (ANNs), modelled after the brain's network of interconnected neurons, are central to AI development. Over recent decades, advances in computing technology and access to larger datasets have significantly improved neural networks. However, AI systems still face significant challenges: they require vast amounts of energy to train, have difficulty handling messy or inconsistent data, and often can't adapt well to new situations. In contrast, humans and animals continuously adapt by synthesizing information from varied experiences over time.

The realm of neuroscience offers insights into these challenges. One key concept is synaptic plasticity: the brain's ability to modify (strengthen or weaken) neural connections based on experience. Donald Hebb's principle of "neurons that fire together, wire together" explains how repeated activity strengthens connections between neurons. This concept has inspired AI systems to learn by adjusting the "strength" or weights of virtual connections between artificial neurons. For example, AI can recognize image patterns by reinforcing connections that detect recurring features, like how the brain learns to identify familiar objects. These algorithms allow machines to learn without being explicitly programmed, mirroring how our brains refine skills through practice.

One of the most notable breakthroughs in AI is the Convolutional Neural Network (CNN), used in computer vision and inspired by the brain's layered processing of visual information. For example, when you see a dog, your brain first notices simple features like edges or shapes and then combines them to recognize the full object. Similarly, CNNs start by identifying basic patterns, like lines or textures, and build on these to detect more complex objects. By stacking these layers, CNNs can analyze images with remarkable accuracy, just as our brains interpret visual scenes.

Lastly, reinforcement learning in neuroscience has inspired AI systems to learn complex strategies by mimicking the brain's reward-based learning process. For example, when a child learns to ride a bike, they adjust their actions based on positive outcomes, such as staying balanced, or negative ones, like falling. Similarly, AI uses algorithms that reward successful actions and penalize mistakes to reinforce certain behaviors and improve over time. This method allows AI to master tasks like playing games or controlling robots independently, much like how humans learn through trial and error.

By merging neuroscience and AI, researchers are transforming both fields by creating more intelligent, more adaptable systems while uncovering deeper insights in artificial and biological intelligence. This collaboration holds immense promise, bringing us closer to machines that assist us and illuminate how our minds work.

Sources:

1. https://zuckermaninstitute.columbia.edu/neuroscience-artificial-intelligence-neuroai 

2. https://torstenvolk.medium.com/three-milestones-in-machine-learning-from-1958-1986-and-1989-6bd274eeb909#:~:text=In%201958%2C%20Frank%20Rosenblatt%20created,use%20cases%20around%20image%20recognition. 

3. https://pubs.aip.org/aip/aml/article/2/2/021501/3291446/Brain-inspired-learning-in-artificial-neural 

4. https://openmedscience.com/cognitive-neuroscience-and-ai-unlocking-the-future-of-intelligence/ 

5. https://www.tableau.com/data-insights/ai/history 

6. https://arxiv.org/pdf/2001.07092 

Want to submit a piece? Or trying to write a piece and struggling? Check out the guides here!

Thank you for reading. Reminder: Byte Sized is open to everyone! Feel free to submit your piece. Please read the guides first though.

All submissions to berkan@usc.edu with the header “Byte Sized Submission” in Word Doc format please. Thank you!