Inside the Attention Layer: How Generative AI Reflects the Brain’s Selective Focus

Technology is pretty great, but it can get overwhelming. We are constantly connected through the internet, and it’s hard to maintain focus with so many distractions: phone notifications and multiple tasks happening simultaneously result in overstimulation and an ineffective use of our time. With so much going on at the same time, all the time, how does our brain decide what to pay attention to, and how can we use AI models to study it?

A good place to start is the brain’s ability for selective attention. It allows us to focus on relevant stimuli while filtering out distractions. This is coordinated by multiple sections of the brain working together, such as:

• The prefrontal cortex: the “project manager”, tells the brain what to focus on and what to ignore

• The parietal region: shines a spotlight on priority information

• Dopamine: a neurotransmitter that ‘rewards’ you for focusing

So when I’m studying, my brain works like an internal editor, constantly switching between focusing on the text and ignoring all other noises happening in the background.

In generative AI models like GPT, there is a direct parallel to the human selective attention process called the “Attention Layer”. It is designed to determine how much weight to assign to each piece of input data, like keywords that help predict what the user would say next. Just as the brain filters sensory input, AI filters data to make sense of the context and meaning of the user’s prompts. For example, if the user prompts with a sentence “The cat sat on the …”, the model focuses on “cat” and “sat” to predict the next likely word, and will provide you with multiple options to fill in that gap. I invite you to give it a try and see all the things AI thinks your cat sat on.

Learning from this parallel, we see how AI models do not see everything at the same time, but instead prioritize effectively depending on the user’s specific needs. This can really help students focus on what is important, especially with subjects that require focus and extensive comprehension. AI tools can mimic selective attention to highlight key concepts, summarize text, and refocus learning tasks for more efficient and effective learning processes.

And these tools already exist! I’ve mentioned in a previous post about AI programs that I personally use to help me with memory and focus, like Anki and NotionAI. There are numerous others to choose from, like Otter.ai, Focus Mode on Grammarly, and Knewton, designed to adjust content difficulty based on signals of user engagement. There are AI note-takers that summarize key ideas and help me prioritize useful information, and generative tools that model human attention to teach me how to extract key information from complex material, saving time when the source material is beyond my level of understanding. These tools have allowed me to see my cognitive blind spots and maneuver accordingly, making my learning journey a lot smoother and more enjoyable.

Adaptive technologies aid my understanding of how focus works, but real attention takes time, intentional practice, and self-awareness. Algorithmic filtering is helpful, but not enough. Our metacognitive abilities–thinking about how we think–are our way of centering ourselves, the same way an AI model checks its own attention matrix and where its focus compass is pointing, using algorithms and code. If AI can model attention mathematically, maybe it can also remind us that focus is a trainable skill, both biologically and digitally.

Although AI mirrors more of the brain’s cognitive processes, it cannot replace human thinking, and this shows just how much we don’t yet understand. We’ve mapped the surface of Mars more clearly and accurately than the folds of our own cortex, and that is an exciting concept for me. Maybe, just maybe, the next step is creating AI models that, beyond words, prioritize real human values, real human meaning.