‘s group at Stanford University, California, studies girls aged 10 to 14 years whose mothers suffer from depression. Such girls are thought to be at , in part because they may inherit their mothers’ tendency to “amplify” unpleasant information. Although none of the girls has yet experienced a depressive episode, Gotlib has found that their brains already overreact to negative emotional stimuli – a pattern they share with their mothers and other depressed people.

Gotlib is studying whether these young subjects can use interactive software and brain-imaging hardware to “rewire” their brains by unlearning this negative bias. In a pilot experiment, eight girls used a neural feedback display to learn how to control activity in a network of interrelated brain regions that have been linked to depression – these include the dorsal anterior cingulate cortex, anterior insula and dorsolateral prefrontal cortex.

The level of activity in this network was measured using an functional MRI scan and displayed to the girls in the form of a thermometer on a computer screen. The girls were shown sad or negative pictures that might ordinarily raise their “temperature”, and tried to lower that “temperature” by adopting more sanguine mental states. They were then advised to try to recreate that mindset in their daily lives.

A control group unknowingly watched someone else’s scan output instead of their own, so they didn’t actually learn how to control their brain activity.

Accentuate the positive

Another set of girls in the pilot experiment received their training through a simple computer game instead. In this game, a pair of faces appeared on a screen every few seconds: they would be either neutral and sad, or neutral and happy. Then a dot replaced one of the faces, and the “game” was to click on the dot. For the eight girls in the control group, the face replaced by the dot was selected at random, but for eight girls in the experimental group, the dot always replaced the more positive face in the pair. Over a week of playing this game daily, these girls were in effect being trained to avoid looking at the sad faces.

Gotlib himself originally found this concept, called attentional-bias training, so simplistic that he bet , a psychologist at the University of Western Australia in Perth who pioneered the technique, that it would not alter psychological symptoms. Gotlib lost his bet.

In his pilot study, both kinds of training significantly reduced stress-related responses – for example, increases in heart rate, blood pressure and cortisol levels – to negative stimuli. These stress responses are a key marker of depression, and they diminished one week after training. The girls in the experimental groups also developed fewer defensive responses to negative faces, such as startled blinking. Control groups showed no such improvement.

, head of Harvard University’s clinical psychology programme, was impressed by the findings despite the small sample size: “This is highly innovative work,” she said. “Ian is breaking new ground here.”

Gotlib is adding more subjects to the training programme and plans to compare their long-term mental health with a parallel cohort of 200 girls, half of whom have depressed mothers, who aren’t participating in the study.

He presented his results at the annual meeting of the in Boston in September.

Unlike electronic debit and credit card transactions, which merely report the amount spent to the appropriate bank or credit agency, the digital cash system actually assigns a value to strings of data that can be exchanged like virtual coins. “Your computer will maintain a portfolio of these, like the balance you have in your wallet,” says David Chaum, the president of Digicash. “When you want to make a payment, it will take out the right set of coins and send them off.”

Mark Twain Bank’s computer stamps each digital coin with a unique, encrypted serial number. The customer’s computer also tags a personal serial number to each digital coin, which other computers, including the bank’s, cannot trace. So users can either keep their transactions anonymous, or they can disclose their personal serial number to the recipients of their digital payments.

Digicash’s electronic cash has been tested in a fun-money experiment for about a year (“Banking on electronic money”, 8 April 1995), and in some closed environments. But Mark Twain will be the first bank to guarantee a value for electronic cash on the open market. The Swedish bank Sweden Post has also licensed the technology, although it has not yet announced a launch date.

The designers of the house at the Colorado-based firm Winston International believe that it will make life safer for millions of people who live on flood plains, both in the US and in developing countries. But a row with the US government’s Federal Emergency Management Agency (FEMA) – which provides flood insurance for people living in affected areas – has so far stymied plans to build prototype homes.

Houses built on flood plains in the US are repeatedly knocked down by rising waters. Some authorities, including FEMA, contend that the best solution is to move everyone away from the water’s edge. But many people’s livelihoods depend on cultivating the rich soil on the flood plains, so they refuse to leave.

Rather than permanently teetering on stilts, Winston’s Landlocked Floating House is designed to rest on the ground when the river level is normal, and float like a raft when the floods come. Foam-filled tubs cover the underside of the house to provide the necessary buoyancy, while lubricated steel piers project downward from the house’s corners and rise out of their concrete sockets to serve as anchors. As the house lifts, the plumbing and electrical systems should disconnect automatically and switch over to a stand-by generator and built-in water storage tanks.

When designer Paul Winston first introduced his idea to federal officials, he expected it to be greeted with enthusiasm. “The public needs to have something like this because the government can’t afford to move everybody away from the rivers,” he says. The technology will add around $12 000 to the cost of building a house.

But officials at FEMA are not convinced. “These structures may be sitting around for fifty years and never be called on to float, and then they may not float when there’s a flood,” says Cliff Oliver, a senior engineer at FEMA.

Oliver counters that FEMA’s main concerns are that it will be liable to pay for cleaning out silty muck under the homes when the floods recede, and for damage from any debris that pummels the houses during violent storms.

Moreover, Winston’s house does not meet FEMA’s current insurance requirements, which state that new houses on flood plains must be permanently fixed at 30 centimetres above the 100 year flood line. Although the Landlocked Floating House should rise to over 1.5 metres above the 500-year flood line, its ground-level position technically violates the FEMA policy.

There is also the risk that a floating house will give residents a false sense of security and encourage them to stay put at home when they would be safer evacuating the area.

FEMA will not permit even a prototype house to be built on land it insures; the agency wants to see the technology demonstrated with a weighted barge instead. But Winston’s group wants to launch a full-blown demonstration in order to convince potential investors that the concept is practical.

Negotiations with FEMA are continuing. Winston has won the backing of several members of Congress, including Republican leader Robert Dole.

Oliver says the encouraging results of the latest computer models will be considered, but it will take long-term physical studies to assuage the agency’s concerns over damage from silt and debris. For now, Winston is proceeding with plans to demonstrate the house in a warehouse near Denver, although its results will not be officially recognised by FEMA (see Diagram).

In the current issue of the Journal of the Experimental Analysis of Behavior, psychologist Shigeru Watanabe and his colleagues at Keio University in Tokyo, describe how they trained pigeons to distinguish a Picasso from a Monet, and more generally, impressionism from cubism, by pecking the correct picture.

They trained the pigeons to distinguish between Monet and Picasso with 90 per cent accuracy. Once trained, the pigeons maintained their ability even for works they had never seen before. And when presented with paintings of other impressionists, such as Cézanne and Renoir, the birds lumped these in with Monet’s portfolio but distinguished them from works by cubists such as Georges Braque.

Art critics might protest that these pigeons have not developed any aesthetic sensibility but have merely learnt to respond to simple cues, such as the sharp angles and bold colours of cubism compared with the fuzzy contours and pastel shades of the impressionists.

But Watanabe and his colleagues showed the pigeons remained accurate judges of style even when the images were blurred, or shown only in black and white. “We integrate several cues to recognise which is impressionist and which is cubist,” he says. “The pigeons may do the same thing.”

Indeed, the art critic’s only remaining advantage is that they do not peck the paintings.

The genetic ink is the brainchild of Charles Butland, founder and president of the Los Angeles company DNA Art Guard International. Butland’s team isolated a DNA fragment from a few of Barbera’s hairs, sandwiched it between chemical primers, and amplified it by a standard technique known as the polymerase chain reaction. They then mixed the amplified sequence into an ordinary ink and used it to fill a normal pen, which Barbera presumably keeps in a very safe place. For extra security, each cel’s numbered seal and accompanying certificate are also treated with his genetic material.

Whenever cartoon collectors doubt the authenticity of their signed frames of Fred and Wilma, they need only subject the cels to a quick sweep under one of Art Guard’s handheld scanners. Using a wavelength similar to that of an X-ray, the scanners detect the unique emission spectra reflected back by the artist’s biochemical signature. Since a specific DNA fragment was chosen for the ink, Butland says that the technique “cannot be replicated unless the copyist is aware of the primer and the sequence, even if he is aware of who the donor is.”

Butland hopes to adapt the system to stop the counterfeiting of driving licences, immigration cards, passports, and even currency. In these situations, he admits that labelling each person’s property with their own DNA would be prohibitively expensive and time-consuming. But he says that DNA sequences are so complex that a strand of DNA from a single source could be used repeatedly on thousands of documents – it need not be unique to one person. For example, a single sequence could be used on all driver’s licences to prove that they are genuine.

While this would make it extremely difficult to create counterfeit identification documents, it could not prevent a thief from stealing the originals and using them for themselves.

The local union, however, has responded angrily to the move, claiming that sergeants, who have traditionally been responsible for keeping officers on the straight and narrow, can do the job better.

A $21 000 neural network system installed by the department’s internal affairs division has already identified 91 officers whose behaviour suggests they may become a liability to the police department. But while the internal affairs department insists that the system is benign or even helpful to officers, representatives of the local police union are furious. “What they are saying is that this department cannot supervise correctly,” says Bill Nolan, president of Chicago’s Fraternal Order of Police.

Nolan does not believe that the computer can predict serious misconduct more accurately than supervisors. “Who knows the officer better than the guy working with him out on the street day in and day out?” But he admits that the opinions of supervisory sergeants can be inconsistent.

Because neural networks are modelled loosely on the human brain, they can “learn” in much the same way as humans do, working round missing information (Technology, 3 December) and becoming more reliable as the number of variables increases. The department chose a neural network system called BrainMaker, developed by California Scientific Software.

Raymond Risley, assistant deputy superintendent of internal affairs, says that the size of the Chicago force – 12 500 employees – combined with the frequent transfer of supervisors, has led to “an awful lot of slippage” in the department’s ability to police itself. He says that the goal of the neural network is not to seek out officers already involved in blatant corruption, but to identify officers whose minor misbehaviour could signal major trouble ahead. By offering them counselling, the department will try to steer them back on course before anything serious happens.

His team originally considered using straightforward computer programs to track down “at risk” officers. But they found that such programs could not handle the number of variables needed to describe an officer’s behaviour and could not cope with the amount of information often missing from the records.

To train the neural network to spot problem police, Risley’s team fed it the records of 191 former officers who had been sacked, along with nearly 2000 records of well-behaved officers. The program learnt to spot the warning signs – such as insubordination or unexplained absences – that characterised the behaviour of officers who ultimately lost their badges.

Now the department intends to run 1.5 gigabytes of information on current officers through the neural network four times a year. The names of officers who could be heading for trouble will be sent to the personnel department, which will review their records and possibly refer them for counselling for anything from alcohol abuse to financial problems.

Negotiations with the police union continue, but in the face of such opposition, the project has lost its earlier momentum. Thomas Sadler, the director of personnel, has yet to see the first list of 91 names, which was generated by the computer system in Octobr.

Jerome Klein of Boston University Medical Center and Karen Farbman,
now at the Rhode Island Hospital, tested the plant against a variety of
infectious microorganisms. They found that freshly pressed garlic extract,
even diluted to one part in 250, proved effective against all the organisms
in the laboratory study, including drug-resistant strains of bacteria.

‘The garlic extract not only inhibited, but also killed the bacteria,’
says Klein. The team’s findings were published in the Pediatric Infectious
Disease Journal (vol 12, p 613).

Klein says that the plant’s effectiveness is probably due to allicin,
a simple organic disulphide that is known to be an antibacterial agent.
Allicin is also responsible for garlic’s pungent odour. Klein cautions
that while the laboratory trials give reason for optimism, it is not yet
clear how much allicin reaches sites of infection when we eat garlic, or
even whether it acts as an antibiotic if it does get there. But with resistance
to commercial antibiotics becoming an increasing problem, he says it is
time for this ancient remedy to step out of the annals of folk medicine.