Constructing artificial neurons with functional parameters comprehensively matching biological values | Nature Communications
www.nature.com
The efficient signal processing in biosystems is largely attributed to the powerful constituent unit of a neuron, which encodes and decodes spatiotemporal information using spiking action potentials of ultralow amplitude and energy. Constructing devices that can emulate neuronal functions is thus considered a promising step toward advancing neuromorphic electronics and enhancing signal flow in bioelectronic interfaces. However, existent artificial neurons often have functional parameters that are distinctly mismatched with their biological counterparts, including signal amplitude and energy levels that are typically an order of magnitude larger. Here, we demonstrate artificial neurons that not only closely emulate biological neurons in functions but also match their parameters in key aspects such as signal amplitude, spiking energy, temporal features, and frequency response. Moreover, these artificial neurons can be modulated by extracellular chemical species in a manner consistent with neuromodulation in biological neurons. We further show that an artificial neuron can connect to a biological cell to process cellular signals in real-time and interpret cell states. These results advance the potential for constructing bio-emulated electronics to improve bioelectronic interface and neuromorphic integration. Artificial neurons typically have functional parameters that are mismatched with their biological counterpart. Here, the authors develop artificial neurons that emulate biological neurons in functions and match their parameters such as signal amplitude, spiking energy, temporal features, and frequency response.
  • Warl0k3@lemmy.worldEnglish
    2·
    2 months ago

    IDK, I haven’t had time to really sit down and read the full paper, but from a skim it seems like if this is true:

    [taken from abstract] Moreover, these artificial neurons can be modulated by extracellular chemical species in a manner consistent with neuromodulation in biological neurons. We further show that an artificial neuron can connect to a biological cell to process cellular signals in real-time and interpret cell states. These results advance the potential for constructing bio-emulated electronics to improve bioelectronic interface and neuromorphic integration.

    Then this is a pretty huge breakthrough. It seems like they’ve divised a pretty robust framework for this, which is one of the big hurdles in Brain/Machine interfaces (Or HNI if we wanna be trendy). If this can be incorporated into devices viable for longterm implantation, which it probably can be though that’s by no means a trivial bioengineering task, it’ll absolutely pave the way for true scifi-style neural interfaces. Huge step over shit like a Utah array if we can tap digitally into interphase chemical/electrical conversion steps using this tech. Long way to go before that point, obviously, but this is a big advancement nontheless.

    IDK man I’m just hyped, though yeah you’re 100% right that this is absolutely in no way related to simulating a brain (I mean, I guess you could but eugh that would be horrible get working). God science reporting is a travesty. I just wanna make cool robot limbs for people, is that too much to ask?? Can’t we be happy with that, and not make every paper about computational neurology somehow the “advent of AGI” or whatever? ugh.