Building the greatest artificial intelligence lab on Earth

The Guardian has an article on technologist Ray Kurzeil’s move to Google that also serves to review how the search company is building an artificial intelligence super lab.

Google has gone on an unprecedented shopping spree and is in the throes of assembling what looks like the greatest artificial intelligence laboratory on Earth; a laboratory designed to feast upon a resource of a kind that the world has never seen before: truly massive data. Our data. From the minutiae of our lives.

Google has bought almost every machine-learning and robotics company it can find, or at least, rates. It made headlines two months ago, when it bought Boston Dynamics, the firm that produces spectacular, terrifyingly life-like military robots, for an “undisclosed” but undoubtedly massive sum. It spent $3.2bn (£1.9bn) on smart thermostat maker Nest Labs. And this month, it bought the secretive and cutting-edge British artificial intelligence startup DeepMind for £242m.

And those are just the big deals…

Google has also hired some of the world’s leading artificial intelligence researchers: Geoff Hinton, Demis Hassabis, Andrew Ng and Ray Kurzweil just for starters.

They are all experts in machine learning – which some would say is quite a limited form of AI that doesn’t specifically aim to model itself on human thinking.

But it is clearly the most useful in allowing machines to make conceptual connections from fuzzy data. In particular, a technique called deep learning has proved to be a huge leap forward.

It works best when it has large data sets to work on. Essentially, large data sets make deep learning useful and this is why Google sees its future in AI.
 

Link to Guardian article on Kurzweil and Google engineering.

The Society of Mutual Autopsy

The Society of Mutual Autopsy was an organisation formed in the late 1800s to advance neuroscience by examining dead members’ brains and to promote atheism by breaking sacred taboos.

It included some of the great French intellectuals and radicals of the time and became remarkably fashionable – publishing the results in journals and showing plaster-casts of deceased members brains in world fairs.

In October 1876, twenty Parisian men joined together as the Society of Mutual Autopsy and pledged to dissect one another’s brains in the hopes of advancing science. The society acquired over a hundred members in its first few years, including many notable political figures of the left and far left. While its heyday was unquestionably the last two decades of the century, the society continued to attract members until the First World War. It continued its operations until just before World War II, effectuating many detailed encephalic autopsies, the results of which were periodically published in scientific journals.

The quote is from a fascinating but locked academic article by historian Jennifer Michael Hecht and notes that The Society was partly motivated by self-nominated ‘great minds’ who wanted to better understand how brain structure related to personal characteristics.

It was no backwater project and attracted significant thinkers and scientists. Most notably, Paul Broca dissected brains for the society and had his brain dissected by them, despite apparently never joining officially.

Part of the motivation for the society was that, at the time, most autopsies were carried out on poor people (often grave robbed) and criminals (often executed). The intellectual elite – not without a touch of snobbery – didn’t think this was a good basis on which to understand human nature.

Also, these bodies usually turned up at the dead of night, no questions asked, and no one knew much about the person or their personality.

In response to this, the Society of Mutual Autopsy functioned as a respectable source of body parts and also requested that members write an essay describing their life, character and preferences, so that it could all be related to the shape and size of their brain when autopsied by the other members.

There was also another motive: they were atheists in early secular France and they wanted to demonstrate that they could use their remains for science without consideration of religious dogma.

As with most revolutionary societies, it seems to have fallen apart for the usual reasons: petty disagreements.

One person took exception to a slightly less than flattering analysis of his father’s brain and character traits. Another starting flirting with religion, causing a leading member to storm off in a huff.

In a sense though, the society lives on. You can donate your body to science in many ways after death:

To medical schools to teach students. To forensic science labs to help improve body identification. To brain banks to help cure neurological disorders.

But it’s no longer a revolutionary act. Your dead body will no longer reshape society or fight religion like it did in 1870’s France. The politics are dead. But neither will you gradually fade away into dust and memories.

Jennifer Michael Hecht finishes her article with some insightful words about The Society of Mutual Autopsy which could still apply to modern body donation.

It’s “both mundane – offering eternity in the guise of a brief report and a collection of specimens – and wildly exotic – allowing the individual to climb up onto the altar of science and suggesting that this act might change the world”.
 

Link to locked and buried article on The Society of Mutual Autopsy.

Snow-fuelled neurophilosophy

Pete Mandik is a professor of philosophy and was due to give a class on neurophilosophy before his class got snowed out. Instead of ditching the class he made a fantastic and funny video lecture for his students.

The pipe-chewing Mandik gives a great introduction to this particular philosophical approach to integrating neuroscience and concepts of mind – most associated with the work of Patricia and Paul Churchland.

The lecture is called ‘Two Flavors of Neurophilosophy’ and comes in three parts.

If this is what happens when it snows in New Jersey, let it snow.
 

Link to part one.
Link to part two.
Link to part three.

2004-02-14 Spike activity

Quick links from the past week in mind and brain news:

Cocaine use increases stroke risk in young people reports Science News. Risk of being a giant knob-end already well established.

The New York Times has an interesting piece on how musical hallucinations are giving researchers clues about the workings of the brain.

For the first time, a baby is born to a brain-dead woman kept viable on life-support to be able to give birth. Reported by the Otago Daily Times.

Brain Watch has an excellent explainer on brain death for those wanting some background.

Focussed ultrasound to stimulate the brain. The mighty Neuroskeptic has a look at this new neurostimulatory technique.

New Scientist advises us to fall for a robot to fend off heartache and explores the robot relationship subculture.

A dozen of the craziest romance-related studies ever featured on Seriously Science. Sex apparently burns 3.6 calories a minute. A minute? I barely make 30 seconds.

NHS Choices takes a level-headed look at the ‘male and female brains are different sizes’ story which has gone all shades of wibble-wibble-daft in the media.

The origins of the F-word. A brilliant post from the historians of language at So Long As It’s Words… traces it’s history. Also features John Le Fucker from 1286.

Respect is a medicine

Aeon magazine has an excellent article on how social interactions among medical team members affect clinical outcomes, patient well-being and the number of medical errors that occur.

It’s probably worth saying that the vast majority of doctors and warm and respectful people but it remains one of the last professions where teaching though humiliation is given a place to survive.

The article in Aeon looks at research on teamwork, communication style and respect and finds out that this ‘treat ’em mean, keep ’em keen’ attitude actually leads to higher rates of medical errors.

…many in medicine actively protect the culture of disrespect because they hold a fundamentally flawed idea: that harshness creates competence. That fear is good for doctors-in-training and, by extension, good for patients. That public shaming holds us to higher standards. Efforts to change the current climate are shot down as medicine going ‘soft’. A medical school friend told me about a chief resident who publicly yelled at a new intern for suggesting a surgical problem could be treated with drugs. The resident then justified his tirade with: ‘Yeah, yeah, I know I was harsh. But she’s gotta learn.’

Arguments such as these run counter to all the data we have on patient outcomes. Brutality doesn’t make better doctors; it just makes crankier doctors. And shame doesn’t foster improvement; it fosters more mistakes and more near-misses. We know now that clinicians working in a culture of blame and punishment report their errors less often, pointing to fear of repercussion. Meanwhile, when blame is abolished, reporting of all types of errors increases.

This, incidentally, tends to impact on certain students and trainees more than others. I still meet medical students who want to train as psychiatrists but have to suffer being humiliated in front of their peers by senior doctors when the inevitable ‘what speciality are you interested in’ question comes up.

The Aeon article is a brilliant analysis of the dynamics and interactions in medical teams and why respectful communication and a supportive teaching style is actually better medicine in terms of medical outcomes.
 

Link to Aeon magazine on interactions in medicine.

A reality of dreams

The journal Sleep has an interesting study on how people with narcolepsy can experience sometimes striking confusions between what they’ve dreamed and what’s actually happened.

Narcolepsy is a disorder of the immune system where it inappropriately attacks parts of the brain involved in sleep regulation.

The result is that affected people are not able to properly regulate sleep cycles meaning they can fall asleep unexpectedly, sometimes multiple times, during the day.

One effect of this is that the boundary between dreaming and everyday life can become a little bit blurred and a new study by sleep psychologist Erin Wamsley aimed to see how often this occurs and what happens when it does.

Some of the reports of are quite spectacular:

One man, after dreaming that a young girl had drowned in a nearby lake, asked his wife to turn on the local news in full expectation that the event would be covered. Another patient experienced sexual dreams of being unfaithful to her husband. She believed this had actually happened and felt guilty about it until she chanced to meet the ‘lover’ from her dreams and realized they had not seen each other in years, and had not been romantically involved.

Several patients dreamed that their parents, children, or pets had died, believing that this was true (one patient even made a phone call about funeral arrangements) until shocked with evidence to the contrary, when the presumed deceased suddenly reappeared. Although not all examples were this dramatic, such extreme scenarios were not uncommon.

This sometimes happens in people without narcolepsy but the difference in how often it occurs is really quite striking: 83% of patients with narcolepsy reported they had confused dreams with reality, but this only happened in 15% of the healthy controls they interviewed.

In terms of how often it happened, 95% of narcolepsy patients said it happened at least once a month and two thirds said it happened once a week. For people without the disorder, only 5% reported it had happened more than once in their life.

Although a small study, it suggests that the lives of people with narcolepsy can be surprisingly interwoven with their dreams to the point where it can at times it can be difficult to distinguish which is which.

If you want to read the study in full, there’s a pdf at the link below.
 

Link to locked study at Sleep journal (via @Neuro_Skeptic)
pdf of full text.

2004-02-07 Spike activity

Quick links from the past week in mind and brain news:

Science News has an extended piece on progress with the still-not-entirely-clear-what’s-going-on billion dollar BRAIN initiative.

There might be a little synesthesia in each of us. Nautilus looks at how our senses combine and cross.

The LA Times reports that boxing and ultimate fighting promoters are donating to a neuroscience study on the long-term effects of being repeatedly punched in the head.

The False Memory Archive. An interesting project covered by an article in The Independent.

The West Briton reports on a Cornish drug dealer who told police he didn’t know how heroin had become taped to his testicles. God bless the Westcountry.

There are ways to prevent loved ones from becoming victims of an overdose. Here are three. Important piece from Time.

The Guardian reports that the UK Government has privatised the ‘nudge unit’. Presumably by making it the default option and waiting to see if anyone opts out.

A new cognitive science news website The Psych Report launches and looks very impressive.

Discover Magazine reports that the oldest human footprints found outside of Africa have been found in Norfolk. The ‘out of Norfolk’ hypothesis soon to be published.

Heroin, addiction and free will

The death of Phillip Seymour Hoffman has sparked some strong and seemingly contradictory responses. What these reactions show is that many people find it hard to think of addiction as being anything except either a choice or a loss of free will.

The fact that addiction could involve an active choice to take drugs but still be utterly irresistible seems difficult for most people to fathom.

Let’s take some reactions from the media. Over at Time, David Sheff wrote that “it wasn’t Hoffman’s fault that he relapsed. It was the fault of a disease”. On the other hand, at Deadspin, Tim Grierson wrote that the drug taking was “thoughtless and irresponsible, leaving behind three children and a partner”.

So does addiction trap people within its claws or do drug users die from their own actions? It’s worth noting that this is a politicised debate. Those who favour a focus on social factors prefer prefer the ‘trap’ idea, those who prefer to emphasise individual responsibility like the ‘your own actions’ approach.

Those who want to tread the middle ground or aim to be diplomatic suggest it’s ‘half and half’ – but actually it’s both at the same time, and these are not, as most people believe, contradictory explanations.

To start, it’s worth thinking about how heroin has its effect at all. Heroin is metabolised to morphine which then binds to opioid receptors in the brain. It seems to be the effects in the nucleus accumbens and limbic system which are associated with the pleasure and reward associated with the drug.

But in terms of motivating actions, it is a remarkably non-specific drug and it doesn’t directly cause specific behaviours.

In fact, there is no drug that makes you hassle people in Soho for a score. There’s no drug that manipulates the neural pathways to make you take the last 40 quid out of your account to buy a bag of gear. No chemical exists that compels your hands to prepare a needle and shoot up.

You are not forced to inject heroin by your brain or by the drug. You do not become an H-zombie or a mindless smack-taking robot. You remain in control of your actions.

But that does not mean that it’s a simple ‘choice’ to do something different, as if it was like choosing one brand of soft drink over another, or like deciding between going to the cinema or staying at home.

Addiction has a massive effect on people’s choices but not so much by altering the control of actions but by changing the value and consequences of those actions.

If that’s not clear, try thinking of it like this. You probably have full mechanical control over your speech: you can talk when you want and you can stay silent when you want. Most people would say you have free will to speak or to not speak.

But try not speaking for a month and see what the consequences are. Strained relationship? Lost job maybe? Friends who ditch you? You are free to choose your actions but you are not free to choose your outcomes.

For heroin addicts, the situation is similar. As well as the pleasurable effects of taking it, not taking heroin has strong, negative and painful effects.

This is usually thought of as the effects of physical withdrawal but these are not the whole story. These are certainly important, but withdrawing from junk is like suffering a bad case of flu. Hardly something that would prevent most people from saving their lives from falling apart.

For many addicts, the physical withdrawal is painful, but it’s the emotional effects of not taking drugs that are worse.

Most smack addicts have a frightening pre-drug history of trauma, anxiety and mood disorders. Drugs can be a way of coping with those emotional problems in the short-term.

Unfortunately, in the longer-term, persistent drug use maintains the conditions that keep the problems going. Even for those few that don’t have a difficult past or unstable emotions, life quickly become difficult after regular heroin use sets in.

If you can stay high, you’ll be less affected by the consequences of both long-standing problems and your chaotic lifestyle. If you stop, you feel the full massive force of that emotional distress.

It’s vicious circle that is often set in motion by past trauma but requires a meeting with a drug and the right social circumstances. Just taking the drug until you develop tolerance and withdrawal is unlikely to addict most people.

For example, a Vietnam War study found that just under half of soldiers reported trying heroin, 1 in 5 developed full blown dependency while in Vietnam but only about 5-10% of the dependent soldiers continued using when they arrived home. Most said they gave up without any help and only a small minority had ongoing addiction problems.

In fact, some of you reading this may have been addicted to heroin and not known it. Heroin, under its medical name diamorphine, is commonly used as a painkiller after major surgery. It’s not uncommon that patients develop tolerance and go into withdrawal after they leave hospital but just put it down to ‘feeling poorly’ or ‘recovering’.

But for persistent addicts, the ‘short-term solution that maintains the long-term problem’ cycle is not the whole story and it’s important to remember the neurological effects of the drug and how it interacts with, and changes, the brain.

Addiction is associated with difficulties in resisting cravings and making flexible decisions. This is likely to be caused by a combination of genetics, earlier experience and the ongoing impact of the drug and the drug-focused lifestyle – all of which affect brain function.

A recently popular approach is the ‘disease model’ of addiction which says that the brains of those who become addicted are more susceptible to compulsive drug use because of genetic susceptibility and / or brain changes due to early experience that ‘prime’ the brain for addiction.

It’s probably true to say that the extreme version of the ‘disease model’ – which says addiction is entirely explained by these changes and is best characterised as a ‘brain disease’ – is an exaggeration of what we know about the neuroscience of addiction, but this is not to say that neuroscience is not important.

But either way, there is no clear relationship between an aspect of behaviour being best explained in neurobiological terms and not having any control over that behaviour. For example, most genuine addicts usually give up, on their own, without any assistance and don’t relapse. They still have brains, of course.

Unfortunately though, the ‘disease model’ approach is often used precisely because some think it implies addicts have less control, possibly because they feel (probably wrongly) that it is less ‘stigmatising’ to think of heroin users in this way.

Instead, we know that self-efficacy is one of the best predictors of recovery, so denying people’s role in their own decisions just undermines one of their most important tools for recovery – alongside medication, social support and other forms of therapy.

So to say an addict has ‘no choice’ over their actions is just to misunderstand addiction but to pretend these choices are like any others just misses the fact that they can sometimes be impossibly hard decisions.

Unfortunately though, people find it hard to separate any admission of addicts being able to choose their actions from blame and moral accusation.

Blaming someone for their addiction is like shaming someone for being wounded by an abusive partner. Whatever the circumstances that caused the problem, they deserve respect and treatment, and working with them to help them regain control of their circumstances and promote their own autonomy is an important and valuable way forward.

Revenge is not sweet

An interesting paper in the snappily titled International Journal of Offender Therapy and Comparative Criminology examines what we know about the psychology of revenge.

It has a fascinating section where it discusses how often people take vengeful actions and whether they actually bring any relief.

It seems that taking revenge is rare, but when it happens, it is not only remarkably unsatisfying but counter-productive in terms of dispelling the desire for retribution.

Empirical research by Crombag, Rassin, and Horselenberg (2003) showed that most people do not actually take revenge but merely have thoughts, feelings, and fantasies about it (see also Crombag, 2003). Most people become reconciled with the offender and many people decide to let bygones be bygones. Some of the people who did take revenge could not explain their reason for doing so…

It should be noted that, in the study of Crombag et al., the group of people who took revenge even after a period of time still struggled with more vengeful feelings than the people who did not take revenge. Although 58% experienced satisfaction and 16% experienced triumph, only 19% reported their vengeful feelings to be completely gone, compared with 40% of the people who did not take revenge.

A 2008 study found that one reason that people who do take revenge find it hard to move on is that taking action keeps them ruminating about the events.
 

Link to locked paper on the psychology of revenge.

2014-01-31 Spike activity

Quick links from the past week in mind and brain news:

Nautilus discusses how music hijacks our perception of time.

What the Dunning-Kruger effect is and isn’t. Good in-depth discussion of this often misunderstood effect from [citation needed].

The Atlantic has a fascinating piece on mental illness in Ancient Greece and Rome.

Should a robot decide when to kill? asks The Verge. To the bunkers, you say?

Do Deaf People Hear an Inner Voice? Fascinating discussion on The Voices Within.

The New York Times discusses recent research on how we’re genetically part Neanderthal.

Why the social construction of madness is not as simple as it seems. Excellent piece on the Discursive of Tunbrige Wells blog.

Nature releases the latest edition of the excellent NeuroPod podcast.

An article on the history of the ‘Satanic abuse’ panic of the 1980s is mysteriously taken offline by Psychiatric Times. Gary Greenberg takes up the case.

New Scientist has a oddly-titled article (mind-reading?) on genuinely interesting research looking at how the brain makes sense of phonemes.

Fantastic YouTube video of a moving sculpture that gives the illusion of a rotating head.