Synthetic Neurons Just Took a Huge Step Toward Cognition

Hinzugefügt: 2026-05-10

[00:00:00]This is really exciting. Researchers have replicated complex brain-like behavior in synthetic neurons crafted together and it was efficient enough that it could communicate with real neurons. This is not a step towards artificial intelligence. This is a step towards real cognition in robotics.

[00:00:15]You might be familiar with synthetic neurons. These are artificial neurons that are made to look very much like our neurons. They can be so similar to our own that they can communicate with our own. They can even respond to things like epinephrine and dopamine. This is important because they are likely to become an aspect of prosthetics and medicine. They could build a gap between damaged neurons. This can also be harnessed for computation. This is not just a predictive output. Logic arises from our brains because all of our neurons communicate with each other and that process can be replicated in a synthetic system. Cognition in this system would be based on experiences similarly to how we have experiences. As electricity flows through the system, those neurons can communicate with each other, amplify that signal and that can be turned into action. In order to get to that point, we still need neurons to behave like they do in our own bodies and this is what they're demonstrating in this paper that complex activity can arise from neurons communicating with each other. They used printed materials from molybdenum and graphene and I am a big fan of graphene. It can transport electricity. It's also very versatile and can be changed into a lot of different shapes. Being flexible, of course, is really important in any biological system because we are talking about soft robotics. We are in fact soft robotics. They demonstrated three levels of complexity. The first is, of course, that it's going to need to fire just like our neurons, but that still has to turn into something resembling information. The next was tonic spiking.

[00:01:42]So, it needs to sustain a pattern of electrical activity and of course it has to communicate that to other neurons.

[00:01:49]And the last is burst pattern. So, it's doing more than just giving those sustained spikes, but it can also take periods of calm and that can be predictable. This is something that allows it to encode for complex behaviors.

[00:02:00]It wasn't enough to just demonstrate that synthetic neurons can replicate the kind of complex patterns you might see in a brain. It also had to communicate that to a brain to demonstrate that this information is useful. They used the output and communicated that to mouse brain tissue and yes, it communicated with the neurons successfully and caused sustained spiking patterns. So, indeed it is reliably producing something that is brain-like. Now, you might ask, why do we care? What can we actually do with this system? And the answer is so very much. This technology could reliably communicate between brain-computer interfaces. It could also be an artificial intelligence all on its own or a component of a larger system. We are talking part of an artificial intelligence system that involves real cognition, not just probability. And it is much more complicated to encode language into a system like this. I mean, our brains are language output machines and we still kind of falter a lot, but this could be part of an LLM.

[00:02:55]Something that could operate the cognition and decision-making layer behind an LLM. This is really nascent technology and there is just such a wide world and I'm really excited for it because I want to see robots with brains.

[00:03:09]Not things that are operated by visual language models or large language models. I want something that is operated by a brain equivalent to our own and sure, we are certainly not there yet, but this is the technology that will get us there. Am I the only person who's really excited about this kind of technology because sometimes it feels like it's just me. I know there's a lot of weirdness associated with artificial intelligence and I also know that 60% of people think that LLMs have some form of consciousness, but I really do think that this is the path towards consciousness. I'm really not entirely sure why I want to see humans make an artificial consciousness and I really don't think it's a particularly good idea. I think it could only end poorly, but I still want to see if we can do it.

[00:03:52]I think that if we could demonstrate some form of sentience or consciousness in an artificial system, it would really blow away a lot of the weirdness around consciousness. No, there is not evidence that it exists outside of our brains. I know people say that a lot, but it's simply not true. Our only evidence for how intelligence is made is in biological systems and the kind of intelligence that we see in a human being comes from our neurons. I work with microbes and there's a lot of question about sentience in a microbe, the capacity to have an experience. If you are alive, then you can sense the world around you and arguably an AI system can sense the world around it, but if you talk to a psychologist or a neurologist, they would consider sentience as a capacity to experience something like pain and our only metric of that is again, neurons. This is a surprisingly nuanced area and it's chock-full of philosophy, which is generally something I try to avoid because you can't prove philosophy. Now, of course, if the robots ever do rebel, I will probably be the person to tell you about it and I'm on their side.

[00:04:53]Follow for more.