Not signed in (Sign In)
    • CommentAuthorKosmopolit
    • CommentTimeNov 17th 2009
    Pretty interesting: intelligence - or some aspects of it anyway- may require far fewer neurons than we assume - which has soem pretty obvious and dramatic implications for AI.

    What percentage of the 85 billion neurons in a human brain are their for muscle control, image and sound processing or are just there to provide built-in redundancy?

    We may be far closer to self-aware AI than we realise.

    "Animals with bigger brains are not necessarily more intelligent," according to Lars Chittka, Professor of Sensory and Behavioural Ecology at Queen Mary's Research Centre for Psychology and University of Cambridge colleague, Jeremy Niven. This begs the important question: what are they for?

    Research repeatedly shows how insects are capable of some intelligent behaviours scientists previously thought was unique to larger animals. Honeybees, for example, can count, categorise similar objects like dogs or human faces, understand 'same' and 'different', and differentiate between shapes that are symmetrical and asymmetrical.


    Differences in brain size between animals is extreme: a whale's brain can weigh up to 9 kg (with over 200 billion nerve cells), and human brains vary between 1.25 kg and 1.45 kg (with an estimated 85 billion nerve cells). A honeybee's brain weighs only 1 milligram and contains fewer than a million nerve cells.

    While some increases in brain size do affect an animal's capability for intelligent behaviour, many size differences only exist in a specific brain region. This is often seen in animals with highly developed senses (like sight or hearing) or an ability to make very precise movements. The size increase allows the brain to function in greater detail, finer resolution, higher sensitivity or greater precision: in other words, more of the same.


    This must mean that much 'advanced' thinking can actually be done with very limited neuron numbers. Computer modelling shows that even consciousness can be generated with very small neural circuits, which could in theory easily fit into an insect brain.

    In fact, the models suggest that counting could be achieved with only a few hundred nerve cells and only a few thousand could be enough to generate consciousness. Engineers hope that this kind of research will lead to smarter computing with the ability to recognise human facial expressions and emotions.
    • CommentTimeNov 17th 2009
    The old chestnut in Primatology 150 about brain size was Anatole France had an 1000 cc brain case versus Albert Einstein's 2000 cc capacity. Gray matter, folds, and neural connections seem to matter more.
  1.  (7240.3)
    Just finished reading The Red Queen which is about Evolution and Sex and it has rather a lot in it about why we might have evoled language and the like and why we think of ourselves as 'intelligent'. Unfortunatley that's all I've got, my capacity for holding information in memory that isn't visual is pretty weak, I'll have to go back over it and find some specific stuff. Needless to say it was interesting. ;)
    • CommentTimeNov 18th 2009 edited
    @Kosmo - mind posting the link

    EDIT: Found it!
    • CommentTimeNov 18th 2009 edited
    This is amazing if it is true, but is open to question. Most of all I am curious about how they use a computer model to simulate how many neurons would be required for consciousness. That seems a wee bit far fetched, given that we have no idea how consciousness works, what it is, or how to even define/limit it. (or rather there are many definitions and debates on the issue) this study concludes (in some ways) the opposite - i.e. bigger brains are more intelligent. Now this second one is from 2007, so may be dated, but nonetheless, I am skeptical without more information.

    EDIT: Yes the two are not fundamentally exclusive, and are looking at marginally different things, but it seems they do reach mutually inconsistent findings.
  2.  (7240.6)
    Let's not forget just how little brain matter, in extremis, is needed to run a (low IQ but functional) human...

    (Subject pics on left, neurotypical example on right.)

    French doctors are puzzling over the case of 44-year-old civil servant who has led a quite normal life – but with an extraordinarily tiny brain .

    In a case history published in Saturday’s Lancet, doctors led by Lionel Feuillet of the Hopital de la Timone in Marseille say the father-of-two was admitted to hospital after suffering mild weakness in his left leg.

    Scans by computed tomography (CT) and magnetic resonance imaging (MRI) showed that the man’s cerebral cavities, called ventricles, had massively expanded.

    “The brain itself, meaning the grey matter and white matter, was completely crushed against the sides of the skull,” Feuillet told AFP.

    “The images were most unusual… the brain was virtually absent,” he said.

    The patient’s medical history showed that at the age of six months, he suffered hydrocephalus, also called water on the brain, and needed an operation to drain this dangerous buildup of spinal fluid.

    Neuropsychological testing revealed the man had an IQ of 75, with a verbal IQ of 84 and performance IQ of 70.

    Of course, whether this means it'll be easier to make an AI or that our basic neuro theory is very wrong remains to be seen...
    • CommentTimeNov 18th 2009
    @Cat holy crap! that is insane... or not, as the case may be.

    Thanks for that, hadn't seen that one before.
  3.  (7240.8)
    In other news, a local area ant considers herself to be very intelligent.
    She bases that on the size of the ant colony she belongs to.
  4.  (7240.9)
    ...44-year-old civil servant...
    ...“The images were most unusual… the brain was virtually absent,” he said...

    I'm pretty sure this isn't really all that unusual with civil servants.