Why our brain agrees on what we see: New study, at Reichman University, reveals the shared neural structure behind our common perceptions

How is it that we all see the world in a similar way? Imagine sitting with a friend in a café, both of you looking at a phone screen displaying a dog running along the beach. Although each of our  brains is a world unto itself, made up of billions of neurons with completely different connections and unique activity patterns, you would both describe it as: “A dog on the beach.” How can two such different brains lead to the same perception of the world?

A joint research team from Reichman University and the Weizmann Institute of Science investigated how people with different wired brains can still perceive the world in strikingly similar ways. Every image we see and every sound we hear is encoded in the brain through the activation of tiny processing units called neurons - nerve cells that are ten times smaller than a human hair. The human brain contains 85 billion interconnecting neurons that enable us to experience the world, think, and respond to it. The question that has intrigued brain researchers for years is how this encoding is performed, and how it is possible for two people to have completely different neural codes, yet, end-up with  similar perceptions?

The research team, led by Reichman University graduate student Ofer Lipman and supervised by Prof. Rafi Malach and Dr. Shany Grossman from the Weizmann Institute and Prof. Doron Friedman and Prof. Yacov Hel-Or from Reichman University, set out to observe how brain neurons encode information in real time. This is a most challenging task as most brain-imaging methods provide only a low resolution picture, similar to a satellite photo of a city where you can see the highways but not the people on the streets. To overcome this challenge, the researchers drew on a unique source of data: epilepsy patients with electrodes implanted in their brains for medical purposes. While the implants were placed to help doctors locate the epicenter of the patients’ seizures, they also offered the researchers a rare window into the activity of brain neurons  - recorded live and not simulated or inferred - while the patients viewed images.

The researcher team discovered that, just as in artificial neural networks (the technology behind AI), the raw patterns of activity in the human brain differs from person to person. When observing a cat, the neurons that “light up” (are active) in one person’s brain may be different neurons in another person’s brain.  But here is the surprising finding: When the researchers shifted from examining the raw activity of neurons to observing the relationships between the general activity patterns in the neurons – i.e. how strongly the brain responds overall to a cat versus a dog - they discovered a common relational structure across all participants. For example, if one brain’s general activity in response to a cat is more similar to its response to a dog than to, say, an elephant, that same relationship is likely to hold in all other brains. In other words, the actual activity patterns in different brains may not be identical, but the relationship between them is preserved. This relational representation may be the brain’s way of organizing information so that all humans can understand the world in a similar way, even when the underlying neural coding differs.

“This study brings us one step closer to deciphering the brain’s ‘representational code’ - the language in which our brains store and organize Information,” explains Lipman. “This understanding helps advance not only neuroscience, but also AI: insights into how the brain represents information can inspire the design of more efficient and intelligent artificial networks, and vice versa  - artificial networks can generate insights that deepen our knowledge of the brain. This study forms a part of a broad series of works in which researchers compare the representation of information in natural networks (the human brain) with the representation of information in artificial networks (AI). This integration opens the door to a richer understanding of ourselves and the systems we build.”

So, the next time you see a dog running on the beach and you think “a dog,” remember that behind this simple thought lies a vast and complex code that science is only beginning to crack.