Editor's pick: In 1972 this neural network said 'don't know'
Michael Harrison suggests that neural networks should have a “don't know” output (Letters, 30 June). If he visits the mathematics gallery of the Science Museum in London he will find , which I invented in 1972. It used two “classifiers” on electroencephalograph data from patients in a coma after cardiac arrest. These were hardware neural networks built from variable resistors and summing amplifiers. One classifier was trained to recognise patients who would survive and one to recognise those who wouldn't. If both said “yes” or both “no” it reported “don't know”.
I doubt this was the earliest system to do so. I never released details, because I was alarmed at the ethical and legal minefield that could accompany its use. The introduction of sedation to minimise brain injury affected EEG readings and made the Survival Predictor redundant.
Next to it in the gallery is a film showing a cerebral function monitor that I also – although the film credits “a team of medics”.
The complicated beast that is depression
Clare Wilson is not entirely convinced by Edward Bullmore's argument in The Inflamed Mind that mental adversity causes inflammation, which then causes depression (30 June, p 46). Our bodies may be facing many potential hidden stressors, so I wonder why emphasis is placed on psychological stress as the most important driver of the long-term, low-level inflammation often found in depression.
Infection, especially covert infection, is not vanquished yet. It is surely worthy of being included in the unfolding hypothesis.
There are still fascinating questions here about the potential for there to be different types of depression. The very different responses that people have to antidepressants may be another hint at this. Were this to be the case, the whole field would need a radical rethink.
First class post 鈥� 28 July 2018
I remember at school being told people used to use spiderwebs on injuries to help clotting
Eve Coy with new work on artificial skin grown from spider silk that could help heal wounds (21 July, p 17)
A new epoch of evolution or just more of the same? (1)
Catherine Brahic asks whether we will soon use CRISPR to create children that are fitter by design, rather than by evolution, and if so whether we will have transcended the state of living things (30 June, p 36). If this comes about, surely it will be another step change in evolution, comparable to the changes from non-nucleated cells to nuclear single-celled organisms and from them to multicellular organisms and then to the evolution of consciousness.
Not everything we produce can be described as living, but I think CRISPR-aided human design must count, even if it isn't purely Darwinian in origin.
A new epoch of evolution or just more of the same? (2)
Brahic suggests that we may soon no longer fit NASA's criteria for life because we now use CRISPR editing techniques to modify our own genome and hence may no longer be subject to Darwinian evolution. But we have been modifying our genomes for as long as we have existed as a conscious species, simply by choosing with whom to procreate.
Darwinian evolution can't be turned off, by this or by social care, medicine or technology. It is an inevitable logical consequence in any system in which traits that can affect the breeding success of an individual are heritable – regardless of what those traits are or how they arise. It doesn't require this to be by “natural” +selection rather than “artificial”.
The reasons for wearing a tie are rather futile
You report that wearing a tie restricts blood flow to the brain (14 July, p 20). This confirms my long-held view that I don't think better when wearing a tie, so I refuse to wear one. You have also reported online (24 May 2004) that half of doctors' ties carry disease-causing germs.
The best reason I have seen offered for wearing ties is that it distinguishes you from labourers, who don't wear ties, because they might get caught in things. It is an attempt to show you are important, but is necessary only if you have no other useful abilities.
Consider an idealised spherical scarf…
Researcher Frédéric Lechenault concludes that the stretchiness of knitting is due to the way interlocked stitches spread friction through the fabric, based on an assumption that the yarn itself doesn't stretch (30 June, p 19). Perhaps he should watch someone knit a scarf.
Depending on what pattern of stitch a knitter uses, they form yarn into near circles around circular needles and interlock these with the rows below and above. These circles can easily be pulled into an oval in any direction, but these individually “want” to return to the more relaxed circle – as well as putting other sections of the scarf in different tensions. Knitters know that the yarn they use does stretch; a woollen yarn will stretch about 25 per cent, while a cotton one only about 5 per cent.
A way to build electric passenger aircraft (1)
Peter Wilson is enthusiastic about electric aircraft (30 June, p 24). But one problem they face is that an empty battery weighs just as much as a full one, whereas a liquid-fuelled plane gets lighter and more efficient as its tanks empty. So can we jettison empty batteries and recover them? I recall the aerotow technique used to launch gliders.
Connect two aircraft by a tow cable that is also an electric cable: one aircraft is the passenger carrier and the other a tug dedicated to battery carrying. Flying such an aerotow manually isn't difficult, so I am sure a safe automatic system could be developed. The two would climb to cruising altitude without depleting the passenger carrier battery. The tug could be optimised for climbing, and the carrier for high-altitude cruising.
A way to build electric passenger aircraft (2)
What Wilson fails to mention is that the heavier the plane, the more energy it takes to get from A to B because more lift is needed and there is more drag. About half the weight of a liquid-fuelled long-range flight at take-off is fuel.
We need instead to develop liquid fuel that doesn't come from petroleum or natural gas.
Free Schrödinger's cat while calling a friend (1)
Your encouragement in one issue to think about both consciousness (30 June, p 30) and Schrödinger's cat (p 34) raises interesting questions. When Schrödinger originally formulated his conundrum, he postulated that a single conscious entity making an observation would determine the fate of his poor entrapped moggy.
Subsequently the conundrum was embellished by considering when the wave function actually collapses from the point of view of an additional observer, remote from the original observer. The cat can be considered still to be both alive and dead to this second observer in her reference frame even if the original observer has made an observation.
If the first observer were to telephone the remote observer to tell her that the cat were alive, or dead, some consider that the second observer's consciousness means that hearing the sad or happy news would fix the cat's mortality only in her reference frame. But what happens if the first observer lies? Would the cat be alive to an observer in one frame of reference but dead to an observer in a second?
If the orthodox Copenhagen interpretation is correct, then it is the consciousness of the remote observer that determines the cat's mortality in her reference frame. I think the many-worlds interpretation cannot assist here. Neither quantum bayesianism nor objective collapse theories care about remote observers, and so commend themselves to me.
In any event I suggest that Schrödinger's cat is set free and replaced with a fragile glass replica that can be smashed or not smashed by the hammer.
Free Schrödinger's cat while calling a friend (2)
Schrödinger's cat is one of the great non-questions. Anyone who has ever owned (or rather been staff for) a cat will know that it is conscious and self-aware, so would be perfectly capable of collapsing the wave function itself. Maybe it would be worth discussing “Schrödinger's bacterium”?
Wall posters of pressing science issues please
Roger Taylor mentions that to battle the absence of reason in government, there is a need for engaging the public (Letters, 23 June). I suggest posters in issues of 快猫短视频, detailing – with diagrams, pictures and graphs – the most pressing issue reported that week. Readers could display these publicly.
The benefits of plastic wrappers in a soggy land
Rob Bayly asks why you don't post 快猫短视频 in a paper pack and you say you are reviewing this (Letters, 23 June). I live in the tropical far north of Queensland in Australia, and look forward to seeing my magazine each week in the mailbox at the end of our drive.
During the wet season, any magazine wrapped only in a paper envelope will be ruined by the time I get home from work.
So please can you provide paper wrappers only for those who have their magazine delivered through a hole in their house door, and stick to waterproof wrapping for those of us who live down under.
For the record 鈥� 28 July 2018
• When 900,000-year-old skulls were found at Yunxian in China, Homo sapiens was generally thought to have evolved in Africa at some time between 100,000 and 200,000 years ago (7 July, p 28).
• The genome-edited soybeans that are being grown commercially in the US this year were created using a genome-editing technique called TALEN (7 July, p 22).
• Ring the changes: chlorobenzene is composed of six carbon atoms in a hexagon, plus one chlorine atom and five hydrogen atoms (14 July, p 9).