Computational models of brain function oversimplify human cognition

Our brains are far more intricate than the simplistic comparison to computers has led us to believe, and a closer look at human evolution reveals just why we're struggling with the impact of AI on our lives.

The idea that our minds work like computers dates back to Alan Turing's concept, which views thought as a three-step process: input from the world, computation, and output. This model has been productive in driving technological advancements, from adding machines to artificial neural networks.

However, this simplistic view of the mind is misleading, failing to account for the complexities of human cognition. Our brains are not just processing units, but dynamic systems shaped by millions of years of evolution. The computational approach focuses on understanding the end product of our minds, attempting to reverse-engineer their function.

As a result, AI developers often overlook the long-term consequences of their creations on human cognitive systems. They focus on creating products that appeal to us in the moment, without considering the potential harm they may cause to our mental and emotional well-being.

The development of AI has led to a proliferation of complex technologies that are increasingly integrated into our daily lives. But this integration comes with risks, as we're beginning to see in the impact of AI on various aspects of society.

The integration of artificial intelligence into various aspects of our lives is being driven by computational models that view the brain as an information processor. However, this approach has its limitations and risks. Some researchers are taking a different approach, one that draws from the long arc of evolutionary history to build a biological model of brain and nervous system development.

Neuroscientist Paul Cisek at the University of Montreal has been working on such a model for millions of years. He believes that our brains should be understood as feedback-control systems rather than information processors. This perspective is more in line with how our bodies function, as they don't just receive input but also take action to adjust what we receive.

This approach is exemplified by the process of catching a fly ball in baseball. According to Cisek's model, the outfielder doesn't need to perform complex mental calculations to estimate the ball's velocity and trajectory. Instead, the body takes action to adjust the stimulus it receives, making adjustments based on available options. This view is more true to our experiences as humans.

The computational model of brain function separates out mental processes from physical movement, requiring complex calculations that don't reflect how we actually perceive and respond to our environment. By contrast, Cisek's feedback-control system approach maps neatly onto the biological architecture of brains as they've evolved over time.

Throughout evolution, new behaviors have emerged in response to environmental possibilities. When dinosaurs died off, for instance, nocturnal creatures could move about during the day with less risk of being eaten, giving rise to new capacities. This process has continued from ancient fish to modern humans, with each stage building upon previous ones.

This model highlights how our brains are more dynamic and responsive than the computational model suggests. By understanding brain function as a feedback-control system, we may gain insights into how AI can be developed in a way that better reflects human cognition and behavior.

The computational model's limitations become apparent when examining how our brains developed in response to environmental pressures. A new perspective on brain function as a feedback-control system suggests that our minds evolved not just for computation, but also for interacting with and adapting to the world around us.

This shift in understanding highlights the importance of considering evolutionary history in understanding human cognition. For example, as vertebrate animals gained mobility, it became advantageous to develop specialized systems for exploration, such as using landmarks or navigating at night, leading to the emergence of the hippocampus and episodic memory.

The computational model fails to account for this process, which is deeply rooted in observable neuroscientific structures. Moreover, it obscures the fact that much of brain function involves controlling an organism's interactions within its environment, rather than simply processing information.

To better understand the relationship between our brains and behavior, researchers suggest a paradigm shift away from algorithmic input-outputs towards more dynamic feedback systems. This would involve recognizing that human development has been shaped not only by physical world feedback but also by profoundly social dispositions.

Human society's most significant "feedback loops" arise from its deeply ingrained social nature, which is often overlooked in the computational model and even in AI design. For thousands of years, humans have developed institutions and norms for learning from and communicating with one another, creating complex networks of interaction and influence.

The rapid development of Big Tech companies has disrupted these social feedback loops, dismantling many of the norms and institutions that underpin human communication and learning. This raises important questions about how AI can be designed to better reflect human cognition and behavior, rather than perpetuating a flawed computational model.

Our ancestors' ability to imitate each other's gestures and body movements marked a significant milestone in human evolution, allowing us to pass on successful practices and coordinate complex activities through shared behaviors. This development of mimicry laid the foundation for the emergence of human culture, which has been shaped by our capacity for imitation.

The process of imitating sounds eventually gave rise to oral language, enabling humans to communicate their thoughts to one another. However, it's essential to note that language is not synonymous with thought; rather, it provides a means to express and share ideas. This distinction highlights the complex relationship between human cognition and communication.

As our ancestors transitioned from being nomadic hunter-gatherers to settling in non-transient communities, they developed more sophisticated agricultural practices, allowing them to cultivate food and reduce their reliance on migration for sustenance. This shift had a profound impact on human society, enabling the development of complex societies and cultures.

The invention of written marks to represent spoken languages and abstract ideas was another crucial step in human evolution, facilitating the transmission of knowledge across generations. This breakthrough enabled the emergence of formal education as a means of ensuring that knowledge is shared within a human group, further solidifying the connection between culture and cognition.

Each of these developments, from mimicry to written language, can be seen as an extension of our biological evolution, allowing us to exert greater control over our environment. However, this process also carries risks, as evidenced by unintended consequences that arise from cultural changes.

The example of human diets serves as a stark reminder of the potential pitfalls of cultural evolution. In our hunter-gatherer days, fatty foods were scarce but valuable for survival, leading to an evolutionary adaptation that inclined us towards seeking and storing these nutrients.

The abundance of fatty foods in our modern diets has led to a mismatch between our nutritional needs and consumption habits.

This phenomenon is not unique to physical health; our brains are also vulnerable to overconsumption of certain types of information, particularly when it comes to artificial intelligence.

Just as excessive fat can clog our arteries, AI can overwhelm our cognitive capacities by providing an endless supply of pre-digested knowledge that we no longer need to process in our own minds.

The result is a kind of mental obesity, where our brains become accustomed to relying on external sources for information rather than developing their own internal capabilities.

This can have far-reaching consequences for our ability to navigate the world and make informed decisions.

As more and more people rely on AI-powered tools like ChatGPT, we may find that our minds are no longer equipped to handle the demands of a rapidly changing world.

The development and widespread adoption of AI-powered tools like ChatGPT have led to predictions about its advantages being largely dismissed as minor drawbacks comparable to earlier forms of automation.

This is particularly evident in education, where students are increasingly relying on these tools to avoid the effortful thinking required for building durable knowledge.

A recent study from China found that thousands of students stopped doing their homework once they started using AI-powered tools, highlighting the negative impact of these technologies within educational settings.