Follow Slashdot blog updates by subscribing to our blog RSS feed

 


Forgot your password?
Close
typodupeerror
×
Biotech

Scientists Make Biochem "Brain" From DNA Strands 63

Posted by samzenpus from the big-brain-little-brain dept.
thebchuckster writes "Scientists from the California Institute of Technology have created an artificial neural network (or a "tiny brain," in the words of the lead scientist) from DNA strands that interact with biochemical inputs. The artificial neurons of this network can take incomplete inputs, interact with each other, and come up with a complete conclusion. This is what the human brain does on a much more complex scale. It's also a principle scientists have used for computing and robotics."
This discussion has been archived. No new comments can be posted.

Scientists Make Biochem "Brain" From DNA Strands More

Scientists Make Biochem "Brain" From DNA Strands

Comments Filter:

  • headdesk* (Score:3, Funny)

    by girlintraining ( 1395911 ) on Thursday July 21, 2011 @12:56AM (#36831946)

    The artificial neurons of this network can take incomplete inputs, interact with each other, and come up with a complete conclusion.

    So they've managed to create a republican using only a few brain cells...

    • Re: (Score:1, Offtopic)

      by toastar ( 573882 )

      The artificial neurons of this network can take incomplete inputs, interact with each other, and come up with a complete conclusion.

      So they've managed to create a republican using only a few brain cells...

      Well, it's not that impressive considering republicans can do it without any brain cells

  • I always wondered (Score:4, Interesting)

    by Windwraith ( 932426 ) on Thursday July 21, 2011 @01:12AM (#36831974)

    I always wondered if biomechanical stuff is actually better than "pure" mechanical stuff.
    Aren't the organic components less durable than inorganic ones by definition? If you had a robot (cyborg rather) whose organic brain expires, replacing the organic brain will keep the same functionality? Otherwise, will the metal/plastic parts work perfectly but the machine will remain an empty, useless shell?
    (Will patents and other tricks of "real life science" meddle on this? History dictates they will.)

    I don't know, maybe I am just a "metal purist", but I am not sure about having materials that can rot, into machines that might need to move in too-harsh enviroments or last long. I don't want such components to expire or rot because of one overheating (something a classic CPU can resist fine unless it's fire-inducing hot).

    • Damn, just thought of this after posting.
      How would damage affect the output of such a brain. Will the number 12 still be the number 12 after something external alters the layout of the bio-brain? How will a possible bacterial infection alter the metrics for that CPU, rendering the system defective or useless? (since this is organic, any change might alter the registers or precision, maybe 12-12 won't be 0 as expected). How will it fare when it starts to expire?

      With this I am not saying a scifi scenario of r

      • Re: (Score:2)

        by Plekto ( 1018050 )

        Put someone on enough drugs or in a coma and they forget what 12 is as well.

        • I find your theory to be a deep Wednesday at best.

          • Re: (Score:2)

            by Plekto ( 1018050 )

            Contrary to what Hollywood will have you believe, almost everyone who wakes up from a long-term coma (on the rare instances when they do) has significant brain damage as a result. It's not really that different than altering a mechanical brain would be. Or a really huge blow to the head could do that same thing, of course.
            "12" might not be "12". To assume that the human brain is any more or less robust than an artificial one is kind of silly. If anything, the mechanical one would likely be easier to pr

        • Re: (Score:1)

          by kmoser ( 1469707 )
          "This one derps to eleven."
    • Assuming Turing was correct (a reasonable assumption), no biological brain, or any other type of brain, can do anything a computer can't do. The main difference you'll see is performance differences due to things like parallelism (DNA computing like the guys in this article are doing, for example, can solve a traveling salesman problem surprisingly quickly), or in favor of metal, the fact that signals travel faster through a computer than through neurons. But in terms of fundamental solving power, they are
      • I've long thought this was an implication of Godel's theorems but wasn't aware of which part of Turing's work applies here. Can you expound a bit? Thanks.
        • This conjecture is known as the Church-Turing thesis [wikipedia.org]. Hofstadter's GEB [wikipedia.org] has some discussion of this.
        • The GP is referring to the Physical (or "strong") Church–Turing thesis [wikimedia.org] which says that all physical processes (including, say, any computation done by the human brain) are Turing-computable. I do not know if Turing or Church actually suggested that version or if only later computer scientists came up with it. It cannot actually be proven without a much better understanding of physics, but it is generally believed to be true.
      • Brains likely hold memeories in holographic-like form, and there are probably many (insightful? imaginative? deceptive?) 'errors' in their access pathways. Maybe you could simulate just anything on a Turing Machine, but the whole seems distinctively different.
        • I think the errors in retrieving memories is part of a "working-as-intended" system in the brain that allows older pathways to break down to make way for newer ones. Makes having total recall seem like a defect rather than an advantage.
          • Yes, that too. I'm suggesting that the partial error of being reminded of an analogy, for example, enables an analytical method encoded for one purpose to be applied to a quite different perception (sometimes, of course, with bad results - 'experience' picks the winners).
        • A human can simulate a Turing machine, and (theoretically, no one has proven otherwise) a Turing machine can simulate a brain. You might have to use a paper to augment your memory, since our memories are not optimized to act as an infinite length of tape. This is a different problem, however, than a finite state machine, which can never solve some problems that our brain can solve.

    • Aren't the organic components less durable than inorganic ones by definition?

      I see nothing in the definition of organic (or inorganic) that implies or even suggests a measure of durability.

    • Show me an articulated robot that moves around as long as humans do without totally falling apart every 3 days/months/years and we can start this discussion. Even without any moving parts: Show me *any* kind of artificial (non-organic) device that can process information for 70 years without any interference from outside (like my brain can do), and we can start to think about how organic components are less durable than inorganic ones. Until then, my imaginated cyborg is going to have his inorganic mechanic

      • You might consider thinking more precisely about the degree to which systems are closed, the characteristics of activity, and their scales.

        A brain that only interacts with itself for 70 years will have some interesting dynamics.

        Our solar system has been reliably running calculations and resisted assaults on it for several billion years longer than the class of processes commonly called "organic".

      • Re: (Score:2)

        by Thiez ( 1281866 )

        > Show me an articulated robot that moves around as long as humans do without totally falling apart every 3 days/months/years and we can start this discussion.

        I think most planes move around at least *as far* as humans do in a lifetime without falling apart. Doesn't that count? The have embraced the "Live fast, die young" philosophy :p

    • Re: (Score:1)

      by ipwndk ( 1898300 )

      I think they are less durable, but they have potentially higher computing power in their connectivity; within connectivism that is.

      Each neuron can independently perform its calculations, even though the transfer of information is quite slow. And we can clearly see it works; animals and humans have very powerful computing capabilities. So I expect organic neural networks to be more powerful, at least short term. In the future we might surpass their capability with computer processors, but it's a different ty

    • Even metal and plastic parts are biologically vulnerable. (Look at footage of the Titanic.) I imagine it would be rather more of an issue if these materials had been prevalent in nature, but they may become a more popular food source pretty soon.

      You might not want parts to wear out, but what if it is virtually free to grow a new biological part, and costs 10k for the inorganic equivalent? Biological parts are also pretty good about replacing themselves ("healing"), which means they will be fine on their

    • Organic Technology is some awesome stuff and definitely where it's all at. What we need first is proper neural interfaces with existing technology and we should be sharing this as a species in our universities. Frankly, we need to get over the concept of "war" and except the fact we have differences and in fact rejoice because of them. I don't want to lecture on the logic of that particular subject at the moment. But with those "war" resources turned to science, we can be out gallivanting about the galaxy o

      • Re: (Score:2)

        by Thiez ( 1281866 )

        > Imagine authoring our DNA so that it absorbs metals, converts them into a biological alloy that then becomes our skeletal and muscular system?

        Are you high?

        > We need this. The entire species is located (as far as we know) on just ONE planet at the moment. We are fucking around with all of our eggs in one basket; one planet killer event and we are all GONE. That and we have an entire universe to fuck with, and we all fight about retarded shit here, makes me want to "Three Stooges Slap" the entire huma

        • Nah, someone at Cal Tech decided to setup a Slashdot account for the Biochem Brain.

        • So, wait. Worrying about human capacity and wanting to go to the stars is "being high?" Wanting to get beyond our limits and our short lifespans is "being high?"

          Pardon me, but we need more people like the GP and less dismissive asses like you.

    • Non-biological machines may be more durable but require far more energy and resources to operate, repair, and reproduce, so there is a trade off. Remember any molecule that can absorb energy and resist damage will also require more energy to manipulate thus making repair and reproduction more energy intensive. I would hazard a guess that is why life developed and evolved from such molecules, because they were easier to alter.
    • biomechanical stuff can potentially regenerate

    • Re: (Score:2)

      by mcgrew ( 92797 ) *

      I always wondered if biomechanical stuff is actually better than "pure" mechanical stuff.

      Mechanical stuff breaks down, too, and in my experience breaks down a lot more often. Right now the sound out RCA jacks on my nine year old TV have quit; I'm still trying to decide whether or not to open the damned thing (It's a Sony, for all I know they have C-4 wired in there after what they did to my computer with XCP).

      Meanwhile, I'm 59 and have been a cyborg for five years now. I have a device in my left eye that ma

    • Re: (Score:1)

      by IDK ( 1033430 )
      Natural selection has selected for easy reproduction using few materials, that is the same as cheap.
    • Do you mean we have to, like, imbue computers with a sense of mortality so they are pushed to achieve more in a limited lifetime ?

  • take incomplete inputs, interact with each other, and come up with a complete conclusion.

    That could describe nearly anything in computing. From the article, it looks like they're doing a form of DNA computing. It's not clear to me from the article what their innovation was. It seems like it's the same stuff that's been going on in DNA computing since the 90s (but please someone correct me if I'm wrong).

    • Re: (Score:2)

      by xkuehn ( 2202854 )

      Since this is Slashdot, I didn't RTFA but I'm answering anyway. :)

      Neural networks are a computing model just like Lambda Calculus, Turing Machines, the Von Neumann model, etc. They are all equivalent by the Church-Turing Thesis.

      Neural networks, also known as parallel distributed computing or a bunch of other names, is a biologically inspired parallel computing model with a very vague definition. Now there are some authors who use narrow definitions, but under most just about any form of processing with mult

  • A neural network is not a "tiny brain", it's an idea that was inspired by how the brain works, but it's nowhere near a "tiny brain". I didn't RTFA but this sounds like something like a hopfield network, i.e. a neural network that can retrieve something stored in its "memory" from inputs that share many of the characteristics of that memory, much like the human memory.

  • The Mind thinks, the Brain only connects those thoughts to the body, acting like a switchboard. The Brain can be used for rudimentary thoughts, but little beyond that. Sooner or later these 'scientists' must realise that: this conclusion is inevitable from basic considerations of the maths involved and basic natural assumptions drawn from everyday experience.
  • I for one welcome our new tiny brain overlords. Oh wait! we already have those.

Slashdot Top Deals

Statistics are no substitute for judgement. -- Henry Clay

Close