Month: April 2010

New Scientist has an interesting article on whether single cells can be considered intelligent. The piece is by biologist Brian Ford who implicitly raises the question of how we define intelligence and whether it is just the ability to autonomously solve problems. If so, then individual cells such as neurons might be considered ‘intelligent’ even when viewed in isolation.

However, he finishes on a bit of an odd flourish:

For me, the brain is not a supercomputer in which the neurons are transistors; rather it is as if each individual neuron is itself a computer, and the brain a vast community of microscopic computers. But even this model is probably too simplistic since the neuron processes data flexibly and on disparate levels, and is therefore far superior to any digital system. If I am right, the human brain may be a trillion times more capable than we imagine, and “artificial intelligence” a grandiose misnomer.

It’s odd because it reads like blue-sky speculation when, in fact, the idea that neurons could work like “a vast community of microscopic computers” is an accepted and developed concept in the field supposedly doomed by this idea – namely, artificial intelligence.

Traditionally, AI had two main approaches both of which emerged from the legendary 1956 Dartmouth Conference.

One was the symbol manipulation approach, championed by Marvin Minsky, and the other was the artificial neural network approach, championed by Frank Rosenblatt.

Symbol manipulation AI builds software around problems where data structures are used to explicitly represent aspects of the world. For example, a chess playing computer would have a representation of the board and each of the pieces and in its memory and it works by running the simulation to test out and solve problems.

In contrast, artificial neural networks are ideal for pattern recognition and often need training. For example, to get one to recognise faces you put a picture into the network and it ‘guesses’ whether it is a face or not. You tell it whether it is right, and if it isn’t, it adjusts the connections to try and be more accurate next time. After being trained enough the network learns to make similar distinctions on pictures it has never seen before.

As is common in science, these started out as tools but became ideologies and a fierce battle broke out over which could or couldn’t ever form the basis of an artificial mind.

At the time of the Dartmouth Conference, the neural network approach existed largely as a simple set-up called the perceptron which was good at recognising patterns.

Perceptrons were hugely influential until Minksy and Seymour Papert published a book showing that they couldn’t learn certain responses (most notable a logical operation called a XOR function).

This killed the artificial neural network approach dead – for almost three decades – and contributed to what is ominously known as the AI winter.

It wasn’t until 1986 when two young researchers, David Rumelhart and James McClelland, solved the XOR problem and revived neural networks. Their approach was called ‘parallel distributed processing‘ and, essentially, it treats simulated neurons as if they are a ‘a vast community of microscopic computers’ just as Brian Ford proposes in his New Scientist article.

Artificial neural networks has evolved a great deal and the symbol manipulation approach, although still useful, is now ironically called GOFAI or ‘Good old fashioned artificial intelligence’ as it seems, well, a bit old fashioned.

How we define intelligence is another matter and saying that individual cells have it is actually quite hard to dismiss when they seem to be solving a whole range of problems they might never have encountered before.

Artificial intelligence seems cursed though, as true intelligence is usually defined as being just beyond whatever AI can currently do.

Link to NewSci on intelligence and the single cell (thanks Mauricio!)

The Economist has an interesting article on the ‘CSI effect’ which suggests that television crime dramas are altering jurors’ expectations of the relevance and power of scientific evidence and hence affecting how court judgements are made.

The article is largely based on a forthcoming paper to be published in Forensic Science International that argues the ‘CSI effect’ is influencing how forensic evidence is interpreted and understood by professionals and the public alike.

Nevertheless, both The Economist piece and the academic article in Forensic Science International are notable for the fact they are largely based on anecdotes.

Actually, empirical (shall we say, forensic?) evidence for the effect is harder to come by. One of the few people who have systematically investigated the effect is trial judge and law professor Donald Shelton who came to significantly less alarming conclusions.

In a study on the effect published in the National Institute of Journal, Shelton reported that although to effect did appear in places, it mainly effected expectations and the effect on actual decisions was inconsistent and largely insubstantial:

There was scant evidence in our survey results that CSI viewers were either more or less likely to acquit defendants without scientific evidence. Only 4 of 13 scenarios showed somewhat significant differences between viewers and non-viewers on this issue, and they were inconsistent. Here are some of our findings:

* In the “every crime” scenario, CSI viewers were more likely to convict without scientific evidence if eyewitness testimony was available.

* In rape cases, CSI viewers were less likely to convict if DNA evidence was not presented.

* In both the breaking-and-entering and theft scenarios, CSI viewers were more likely to convict if there was victim or other testimony, but no fingerprint evidence.

Law professor Kimberlianne Podlas was even more damning in a paper [pdf] published in the Loyola of Los Angeles Entertainment Law Review, writing:

Notwithstanding the popularity of such claims, they are not grounded in case-studies or statistical data of increases in acquittals. Rather, they are based on anecdotes about cases wherein law enforcement lost their case while believing it should have won. However, anecdotes are not an adequate substitute for empirical evidence or a logical theory of media influence.

The ‘CSI effect’, it seems, probably wouldn’t stand up in court.

UPDATE: Many thank to Mind Hacks reader Brett for emailing to say that the Stanford Law Review published an article on the supposed ‘CSI effect’ and why it lacks evidence last April, which is also notable for tackling the reasons for why it has gained a cultural foothold despite such flimsy support.

Link to The Economist on the ‘CSI effect’ (via @crime_economist)
Link to Forensic Science International paper.
Link to study on ‘CSI effect’
pdf of Podlas’ article on CSI effect ‘fiction’.

The Guardian has an excellent article questioning the widely cited statistic that ‘1 in 4’ people will have a mental illness at some point in their lives. The issue of how many people have or will have a mental illness raises two complex issues: how we define an illness and how we count them.

Defining an illness is a particularly tricky conceptual point and this is usually discussed as if it is an issue particular to psychiatry and psychology that doesn’t effect ‘physical medicine’ but it is actually a concern that is equally pressing in all types of poor health.

The most clear-cut definition of an illness is usually given as an infectious disease that can be diagnosed with a laboratory test. For example, you either have the bacteria or you don’t.

However, you will acquire lots of new bacteria that will continue to live in your body, some of which ’cause problems’ and others that don’t. So the decision rests not on the presence or absence of new bacteria, but on how we define what it means for one type to be ‘causing a problem’. This is the central point of all definitions of illness.

For example, when are changes in heart function enough for them to be considered ‘heart disease’? Perhaps we judge them on the basis of their knock-on effects, but this raises the issue of what consequences we think are serious, and when we should consider them serious enough to count. Death within weeks, clearly, death within two years, maybe, but is still this the case if it occurs in a 90 year-old?

The idea of a personal change ‘causing a problem’ is also influenced by culture as it relies on what we value as part of a fulfilling life.

In times gone past, physical differences that caused sexual problems might only have been considered an illness if they prevented someone from having children. A man who had children, wanted no more, but was unable to have recreational sex with his wife due to physical changes might be considered unlucky but not ill.

The idea of normal sexual function was different, and so the concept of abnormality and illness were also different.

The same applies to mental illness. What we consider an illness depends on what we take for being normal and what someone has the ‘right’ to expect from life.

The fact that the concept of depression as an illness has changed from only something that caused extreme disability (‘melancholy madness’) to something that prevents you from being content is likely due to the fact that, as a society, we have agreed that we have a right to expect that we enjoy our lives. There was no such expectation in the past.

The problem of correctly diagnosing an illness is a related problem. After we have decided on the definition of an illness, there is the issue of how reliably we can detect it – how we fit observations of the patient to the definition.

This is a significant issue for psychiatry, which largely relies on changes in behaviour and subjective mental states, but it also affects other medical specialities.

Contrary to popular belief, most ‘physical’ illness are not diagnosed with lab tests. As in psychiatry, while lab tests can help the process (by excluding other causes or confirming particular symptoms) the majority of diagnoses of all types are made by what is known as a ‘clinical diagnosis’.

This is no more than a subjective judgement by a doctor that the signs and symptoms of a patient amount to a particular illness.

For example, the diagnosis of rheumatoid arthritis depends on the doctor making a judgement that the mixture of subjectively reported symptoms by the patient and objective observations on the body amount to the condition.

The key test of whether an illness can be counted is how reliably this process can be completed – or, in other words, whether doctors consistently agree on whether patients have or don’t have the condition.

This is more of an issue for psychiatry because diagnosis relies more heavily on the patient’s subjective experience, but it is wrong to think that bodily observations are necessarily more reliable.

For example, the Babinski response is where the toes curl upward after the plantar reflex is tested by stroking the bottom of the foot. It is commonly used by neurologists to test for damage to the upper motor neurons but it is remarkably unreliable. In fact, neurologists agree on whether it is present at a far lower rate than would be acceptable for the diagnosis of a mental illness or psychiatric symptom.

The problem of reliably diagnosing a condition is relatively easy to overcome, however, as agreement is easy to test and refine. The problem of what we consider an illness is a deeper conceptual issue and this is the essence of the debates over how many people have a mental illness.

The ‘1 in 4’ figures seems to have been mostly plucked out of the air. If this seems too high an estimate, you may be surprised to learn that studies on how many people qualify for a psychiatry diagnosis suggest it is too low.

There is actually no hard evidence for one in four ‚Äì or any other number ‚Äì because there’s never been any research looking at the overall lifetime rates of mental illness in Britain. The closest thing we’ve had is the Psychiatric Morbidity Survey, run by the Office of National Statistics. The latest survey, done in 2007, found a rate of about one in four, 23%, but this asked people whether they’d suffered symptoms in the past week (for most disorders).

We don’t know what the corresponding rate for lifetime illness is, although it must be higher. Several such studies have been done in other English speaking countries, however. The most recent major survey of the US population found an estimated lifetime rate of no less than 50.8%. Another study in Dunedin, New Zealand, found that more than 50% of the people there had suffered from mental illness at least once by the age of 32.

Psychiatry has a tendency for ‘diagnosis creep’ where unpleasant life problems are increasingly defined as medical disorders, partly due to pressure from drug companies who develop compounds that could genuinely help non-medical problems. The biggest market is the USA where most drugs are dispensed via insurance claims and insurance companies demand an official diagnosis to fund the drugs, hence, pressure to create new diagnoses from companies and distressed people.

Whenever someone criticises a diagnosis as being unhelpful a common response is to suggest the critic has no compassion for the people with the problem or that they are wanting to deny them help.

The most important issue is not whether people are suffering or whether there is help available to them, but whether medicine is the best way of understanding and assisting people.

Medicine has the potential to do great harm as well as great good and it is not an approach which should be used without seriously considering the risks and benefits, both in terms of the individual and in terms of how it shifts our society’s view of ourselves and the share of responsibility for dealing with personal problems.

So when you hear figures that suggest that ‘1 in 4’ or ‘50%’ of people will have a mental illness in their lifetime, question what this means. The figure is often used to try and destigmatise mental illness but the most powerful bit of The Guardian article shows that this is not necessary:

People who experience mental illness often face stigma and discrimination, and it’s right to oppose this. But stigma is wrong whether the rate of mental illness is one in four, or one in 400. We shouldn’t need statistics to remind us that mental illness happens to real people. By saying that mental health problems are nothing to be ashamed of because they’re common, one in four only serves to reinforce the assumption that there’s something basically shameful about being “abnormal”.

If you want more background on the ‘1 in 4’ figure or discussion about how we understand what is mental illness and who has it, an excellent three part series on Neuroskeptic tackled exactly this point.

Link to ‘How true is the one-in-four mental health statistic?’
Parts one two and three of excellent Neuroskeptic series