Intro. [Recording date: October 21st, 2025.]
Russ Roberts: Today is October 21st, 2025, and joining me is Gaurav Suri, a neuroscientist and co-author, alongside Jay McClelland, of The Emergent Mind: How Intelligence Arises in People and Machines. This book delves into the fascinating interplay between neural networks and large language models, and how these concepts relate to the human experience. Gaurav, welcome to EconTalk.
Gaurav Suri: Thank you, Russ. It’s a pleasure to be here.
Russ Roberts: This book is quite remarkable. It provides deep insights into the mechanics of neural networks and large language models, showcasing how they permeate our everyday lives through technologies like ChatGPT and Claude. However, it also ventures into profound philosophical reflections on the essence of human experience. I highly recommend this read. Our past discussions have touched on AI and neuroscience, and this book seamlessly integrates both themes. Kudos to you and Jay McClelland for this work.
Gaurav Suri: Thank you very much.
Russ Roberts: Let’s begin with the title, The Emergent Mind. Emergence is a recurring theme in our discussions, particularly in economics, where we explore how prices emerge from the interaction of buyers and sellers without central direction. How does this notion apply to the mind?
Gaurav Suri: Absolutely. Living our lives involves making choices—from waking up to deciding between tacos or burritos. The intriguing question is: how do these decisions come to be? A compelling answer is that our intelligence arises from the interactions of simple components in our minds, like neurons that process information and communicate. Yet, when we examine neurons closely, they seem to simply activate and connect without exhibiting intelligence. The essence of our thesis is that intelligence, whether human or machine, emerges from these straightforward interactions among basic units that don’t possess intelligence on their own. Emergence means that certain properties manifest in the entire system, not in the individual parts—like how the collective behavior of ants or the formation of galaxies can’t be attributed to a single entity. Our understanding of intelligence should embrace this emergent perspective.
Russ Roberts: Let’s explore that further. In nature, we can observe flocks of birds or schools of fish, where no single leader dictates their movement. It’s as if these systems operate under an illusion of central command, which can mislead people about the underlying mechanics. Science and economics face the ongoing challenge of deciphering how these intricate systems function without overt direction. You shared an enlightening example involving ants navigating obstacles. It’s fascinating how they manage to find the shortest path without a map or GPS. Can you elaborate on how this illustrates the micro-mechanisms that lead to complex behaviors?
Gaurav Suri: Absolutely, I love this example and find it pivotal. As a child in India, I spent time observing ants. I would place an obstacle on their path, leading to a fascinating outcome. Initially, the ants would split evenly between the longer and shorter routes. However, over time, almost all would gravitate toward the shorter path. What’s going on? Ants communicate not through verbal commands but by secreting pheromones. When an ant successfully navigates the shorter route, it lays down a pheromone trail, which others follow. Eventually, the concentration of pheromones on the shorter path attracts more ants. This simple mechanism of secretion and response leads to complex group behavior, demonstrating how intelligence emerges from basic interactions.
Russ Roberts: This concept resonates with the behavior of birds and fish as well. They operate on simple rules—maintaining proximity without a choreographed plan—unlike the Blue Angels, who rehearse intricate maneuvers with communication. It’s astonishing how such complexity arises from simplicity.
Russ Roberts: Now, let’s connect this back to the brain. Can you describe the brain’s anatomy and the electrical signaling that occurs between neurons? How does this chaotic electrical activity translate into intelligence, given that it appears to be just an electrical storm?
Gaurav Suri: Sure. The human brain contains approximately 86 billion neurons, each functioning as a processing unit. The neurons are structured like trees, with dendritic branches receiving signals and an axon transmitting them. At a high level, neurons perform two primary functions: generating electrical signals called action potentials and forming connections with one another. These connections can be innate or shaped through experiences. As we learn, we create new pathways, and it’s through these simple actions—activation and connection—that complex intelligence emerges.
Russ Roberts: As you speak, countless listeners are experiencing their own internal electrical storms, processing this new information. It’s fascinating how neurons interact to create our thoughts and actions.
Russ Roberts: A key theme in your book is the unsettling notion that our consciousness and decision-making may not be as autonomous as we believe. Economists often posit a rational decision-making model, yet your perspective suggests that many of our actions are dictated by past experiences rather than conscious control. Can you delve into this idea?
Gaurav Suri: That’s a great question. Let’s consider a classic experiment by psychologists Nisbett and Wilson from the 1970s. They presented participants with four identical stockings labeled A, B, C, and D, asking them to choose their favorite. Most selected the fourth stocking, attributing their choice to various reasons, yet all stockings were identical. This illustrates that choices can be influenced by external factors rather than conscious reasoning. Our brains are complex systems where associations and experiences dictate our decisions, often without our awareness.
Russ Roberts: You highlight that even when we perceive ourselves as making conscious choices, our underlying neural networks may generate our reasons for those choices. It’s a profound revelation.
Gaurav Suri: Exactly. The brain operates on associative networks, much like ants in a colony. For instance, if you head to the refrigerator and say you’re hungry, that feeling may stem from thirst-detecting neurons activating due to a high salt level in your blood. You might attribute your action to rational thought, but it’s driven by underlying neural processes and learned behaviors.
Russ Roberts: We’ll revisit the uplifting conclusion of your book later. However, your exploration of our lack of control over our impulses is indeed troubling. Your insights challenge conventional notions of rational economic behavior and suggest that our decisions are often influenced by subconscious factors. This raises questions about accountability.
Gaurav Suri: Yes, this notion is key. Let’s take another example: when you’re reading, the context can trigger an action readiness to snack, similar to how people might mindlessly eat popcorn at a movie theater. This action readiness is shaped by past experiences and environmental cues, leading to decisions that may not align with our conscious reasoning.
Russ Roberts: If I hadn’t read your book, I might dismiss your claims about irrationality as mere observations. However, you argue for a deeper understanding rooted in neuroscience. Can you elaborate on how our neural connections shape our behaviors?
Gaurav Suri: Certainly. One analogy is to visualize neural networks as interconnected pools of water. Connections between neurons deepen with repeated activation, much like water carving channels through soil. This process, articulated by Canadian neuroscientist Donald Hebb, states, “neurons that fire together wire together.” The more frequently certain neurons activate in tandem, the stronger their connection becomes. As a result, our responses to stimuli are influenced by these established pathways, leading to varied reactions based on one’s unique experiences.
Russ Roberts: Your perspective on habits is important. We develop both positive and negative habits that influence our choices. However, your argument extends beyond mere habits to suggest that our actions are determined by an intricate interplay of our neural architecture shaped by past experiences.
Gaurav Suri: Correct. Our experiences shape channels of neural activation that influence our behaviors over time. When we’re infants, we begin to build these channels based on stimuli, and while some connections are innate, most are formed through interactions. Ultimately, we become products of our neural histories, influencing our present actions.
Russ Roberts: To summarize your point on determinism: it operates under natural laws, yet its complexity renders it unpredictable for an outsider. This intricacy complicates discussions about free will, suggesting we are often shaped by past experiences and contexts.
Gaurav Suri: Exactly. While our actions may seem deterministic, the intricate nature of our neural networks complicates predictability. We experience a sense of control and agency, but many of our decisions stem from neural processes influenced by our history and environment. Recognizing this complexity offers a more nuanced view of human behavior.
Russ Roberts: Your insights suggest that we are products of our experiences, reflecting a deterministic view of human behavior. However, you emphasize the beauty in this unfolding process, recognizing our agency in pursuing goals shaped by our neural architecture.
Gaurav Suri: Yes, while we operate within a deterministic framework, the emergent properties of our experiences add richness to our existence. We are shaped by our interactions, and understanding this can foster compassion for ourselves and others as we navigate life’s complexities. Free will, as we discuss it, remains a useful construct within this deterministic context.
