Neuroscience
Month
The best way to learn is to teach. Now a classroom robot that helps Japanese children learn English has put that old maxim to the test.
(Image: Sinopix/Rex Features)
Shizuko Matsuzoe and Fumihide Tanaka at the University of Tsukuba, Japan, set up an experiment to find out how different levels of competence in a robot teacher affected children’s success in learning English words for shapes.
They observed how 19 children aged between 4 and 8 interacted with a humanoid Nao robot in a learning game in which each child had to draw the shape that corresponded to an English word such as ‘circle’, ‘square’, ‘crescent’, or ‘heart’.
The researchers operated the robot from a room next to the classroom so that it appeared weak and feeble, and the children were encouraged to take on the role of carers. The robot could then either act as an instructor, drawing the correct shape for the child, or make mistakes and act as if it didn’t know the answer.
When the robot got a shape wrong, the child could teach the robot how to draw it correctly by guiding its hand. The robot then either “learned” the English word for that shape or continued to make mistakes.
Robot learner aids learningMatsuzoe and Tanaka found that the children did best when the robot appeared to learn from them. This also made the children more likely to want to continue learning with the robot. The researchers will present their results at Ro-Man - an international symposium on robot and human interactive communication - in September.
"Anything that gets a person more actively engaged and motivated is going to be beneficial to the learning process," says Andrea Thomaz , director of the Socially Intelligent Machines lab at the Georgia Institute of Technology in Atlanta. "So needing to teach the robot is a great way of doing that."
The idea of students learning by teaching also agrees with a lot of research in human social learning, she says. The process of teaching a robot is akin to what happens in peer-to-peer learning, where students teach each other or work in groups to learn concepts – common activities in most classrooms.
Source: NewScientist
Ignacio Arganda, a young researcher from San Sebastián de los Reyes (Madrid) working for the Massachusetts Institute of Technology (MIT) is one of the driving forces behind Fiji, an open source platform that allows for application sharing as a way of improving biomedical-image processing. Arganda explains to SINC that Fiji, which has enjoyed the voluntary collaboration of some 20 developers from all over the world, has become a de facto standard that assists laboratories and microscope companies in their development of more precise products.
Ignacio Arganda is a postdoctoral researcher at the Laboratory of Computational Neuroscience of the Massachusetts Institute of Technology (MIT). Along with a group of researchers he implemented Fiji, a platform that allows for applications to be shared in order to improve and advance in the processing and analysis of biomedical imaging. “All of this in open source,” outlines Arganda.
The platform was built from a previous one, ImageJ, which was well known in the industry at the time. ImageJ was not an open source platform but it was publicly accessible. According to Arganda, it had the advantage that any person working in medical imaging could easily create small software applications to resolve their particular problems and then incorporate it into the platform by means of a plug-in (an application which is linked to another providing a new or specific function).
Nonetheless, the researcher adds that this platform became too chaotic with applications of all kinds, some of which were not related to biomedical-imaging. It also began being used to handle astronomical images, in video tracking, etc. “There was a significant lack of control and structure,” he says.
Therefore, “in a spontaneous manner and without any help” this group of researchers decided to create the new open source platform that could put order to that already in place, reusing what was of interest and useful in their work.
"We created a webpage organised like Wikipedia where people could contribute and use their knowledge to help others. To our surprise, it became very popular," he ensures. According to Ignacio Aranda, Fiji currently has 127,000 unique visits (20,000 each month).
Mathematical models developed by scientists at the University of Bristol are providing new insights into why the placebo effect exists and when it should occur. Their research is published in the journal of Evolution and Human Behaviour.
A placebo – such as a sugar pill – is a treatment which is not effective through its direct action on the body but works because of its effect on the patient’s beliefs. But if individuals are capable of recovering without external aid, why do they rely on an external cue? In other words, why have individuals not evolved the ability to get better immediately on their own?
Members of the Modelling Animal Decisions group in the University of Bristol’s School of Biological Sciences built mathematical models of the placebo effect which examine the trade-off between the costs and benefits of an immune response when faced with a health problem.
The work is based on an idea proposed by the theoretical psychologist Professor Nicholas Humphrey. He proposed that, as it can be beneficial to hold the immune system back from full operation due to uncertainties about the state of the world (such as the possibility of starvation), cues which indicate a change can therefore lead to an altered level of immune response.
The models take this argument even further and demonstrate that the placebo effect is modulated by the patient’s expectations. Previous studies measuring brain activity using functional magnetic resonance imaging (fMRI) provide experimental evidence which support the models, by showing correlations between the placebo effect and regions of the brain associated with expectation.
The models show why changes to the perceived cost of getting well, the value of being well or external environmental factors can induce the placebo effect.
Dr Pete Trimmer, lead author of the work, said: “The placebo effect comes down to expectations about when to take action. Waiting for a useless pill before taking action is not optimal. But the general responsiveness to cues is adaptive, so it is logical for evolved organisms to display the placebo effect.”
The models indicate that under stress it can be better for the immune system to work less effectively. However, the most important finding of the research is that the particular type of belief in the treatment can lead to positive or negative effects. The belief that a treatment will cure, without any need for the immune system to do anything, could have deleterious effects on the patient’s health.
Now that a theoretical approach has laid the foundations of understanding the placebo effect, future empirical work may provide insights as to how the placebo effect can be invoked and controlled in a clinical environment. The Bristol study clearly shows that the focus of future placebo studies should be shifted to the type of belief patients have about their treatment rather than just whether a treatment is helpful or harmful. A better understanding of the placebo effect may change the code of practice for health practitioners and save human lives.
Source: University of Bristol
Researchers at the University of Southern California have devised a method for detecting certain neurological disorders through the study of eye movements.
In a study published today in the Journal of Neurology, researchers claim that because Attention Deficit Hyperactivity Disorder (ADHD), Fetal Alcohol Spectrum Disorder (FASD) and Parkinson’s Disease (PD) each involve ocular control and attention dysfunctions, they can be easily identified through an evaluation of how patients move their eyes while they watch television.
“Natural attention and eye movement behavior – like a drop of saliva – contains a biometric signature of an individual and her/his state of brain function or dysfunction,” the article states. “Such individual signatures, and especially potential biomarkers of particular neurological disorders which they may contain, however, have not yet been successfully decoded.”
Typical methods of detection—clinical evaluation, structured behavioral tasks and neuroimaging—are costly, labor-intensive and limited by a patient’s ability to understand and comply with instructions. To solve this problem, doctoral student Po-He Tseng and Professor Laurent Itti of the Department of Computer Science at the USC Viterbi School of Engineering, along with collaborators at Queen’s University in Canada, have devised a new screening method.
Participants in the study were simply instructed to “watch and enjoy” television clips for 20 minutes while their eye movements were recorded. Eye-tracking data was then combined with normative eye-tracking data and a computational model of visual attention to extract 224 quantitative features, allowing the team to use new machine learning techniques to identify critical features that differentiated patients from control subjects.
With eye movement data from 108 subjects, the team was able to identify older adults with Parkinson’s Disease with 89.6% accuracy, and children with either ADHD or FASD with 77.3% accuracy.
Providing new insights into which aspects of attention and gaze control are affected by specific disorders, the team’s method provides considerable promise as an easily-deployed, low-cost, high-throughput screening tool, especially for young children and elderly populations who may be less compliant to traditional tests.
“For the first time, we can actually decode a person’s neurological state from their everyday behavior, without having to subject them to difficult or time-consuming tests,” Itti said.
Nanoparticles often meet a sticky end in the brain. In theory, the tiny structures could deliver therapeutic drugs to a brain tumour, but navigating the narrow, syrupy spaces between brain cells is difficult. A spot of lubrication could help.
Nanoparticles (green) coated with poly(ethylene-glycol) (PEG) (Image: Elizabeth Nance, Graeme Woodworth, Kurt Sailor)
Justin Hanes at Johns Hopkins University in Baltimore, Maryland, was surprised to discover just how impermeable brain tissue is to nanoparticles. “It’s very sticky stuff,” he says, similar in adhesiveness to mucus, which protects parts of the body – such as the respiratory system – by trapping foreign particles.
It was thought that the adhesiveness of brain tissue limited the size of particles that can smoothly spread through the brain. Signalling molecules, nutrients and waste products below 64 nanometres in diameter can pass through the tissue with relative ease, but larger nanoparticles – suitable for delivering a payload of drugs to a specific location in the brain – quickly get stuck.
Now Hanes and his colleagues have doubled that size limit. They coated their nanoparticles with a densely-packed polymer shield, which lubricates their surface by preventing electrostatic and hydrophobic interactions with the surrounding tissue. “A nice hydrated shell around the particle prevents it from adhering to cells,” says Hanes.
Tracking the particles
Using this approach, they were able to observe the diffusion of nanoparticles 114 nanometres in diameter through live mouse brains and dissected human and rat brain tissue. Hanes believes the true upper size limit now lies somewhere between 114 nm and 200 nm. “Things were starting to slow down at 114,” he says.
But further research is needed before the team can progress to clinical trials in humans. “At this scale, it is very important to understand where our nanoparticles go once injected into the body,” says team member Elizabeth Nance, also of Johns Hopkins University. “We will need to show that, when combined with a therapeutic agent, these particles are getting to our site of interest, are having the intended effect and are not causing any side effects or toxicity to healthy normal tissue.”
"The effect of this work should be long-term," says Paul Wilson at the University of Warwick in Coventry, UK. The result represents significant progress in the battle to administer drugs within the brain, he says. "More effective and longer-lasting treatments against brain diseases, such as tumours and strokes, will no doubt soon follow."
Source: NewScientist
An unhealthy diet could lead to Alzheimer’s disease by triggering a form of insulin resistance dubbed “type three diabetes”, scientists claim.
Photo: Getty Images/Peter Macdiarmid
High levels of the hormone insulin, brought on by a bad diet, may harm the brain in the same way that the muscle, liver and fat cells are affected by type two diabetes. Exposing the brain to too much insulin could cause it to stop responding to the hormone, hampering our ability to think and create new memories and ultimately leading to permanent damage, researchers said.
A diet high in fat and sugar has long been linked to a higher risk of Alzheimer’s, while studies of health among large populations have shown that a healthy Mediterranean diet may offer some protection. In type two diabetes, eating too much fatty and sugary food raises our insulin levels to such a consistently high degree that our muscles, fat and liver cells are no longer affected by the hormone.
This means that the amount of glucose and fat in our blood is allowed to increase unchecked, forcing the pancreas to produce even more insulin to try to cope. Ultimately it becomes exhausted and production drops to very low levels.
A small-scale trial on human patients at Washington University found that those who were given a nasal spray containing insulin were better at remembering details of stories, had longer attention spans and were more independent. A further trial on 240 volunteers showing early signs of dementia will provide further clues as to whether the spray can protect memory and learning ability and keep track of brain changes in patients.
A study on rats by experts from Brown University suggest that a similar process could affect the brain, which relies on insulin to regulate nerve signals related to memory and learning and to produce energy from glucose. Researchers found that blocking insulin from rats’ brains made them disorientated and unable to find their way out of a maze because they could not remember where they were.
Examination of their brains showed the same pattern of deterioration seen in Alzheimer’s patients, including increased levels of the amyloid plaque which is a key hallmark of the condition. If the theory is correct, it means eating more healthy foods and exercising more could reduce the risk of Alzheimer’s, and potentially reverse or slow down the memory loss in patients with the condition.
Dr Suzanne de la Monte, who led the study, told New Scientist magazine: “[The rats] were demented. They couldn’t learn or remember. “I believe [Alzheimer’s] starts with insulin resistance. If you can avoid brain diabetes you’ll be fine. But once it gets going you are going to need to attack on multiple fronts.”
ScienceDaily (Aug. 28, 2012) — Deep-brain stimulation (DBS) may stop uncontrollable shaking in patients with Parkinson’s disease and essential tremor by imposing its own rhythm on the brain, according to two studies published recently by University of Alabama at Birmingham researchers in the journal Movement Disorders. An article addressing brain stimulation for essential tremor was published online August 28; a related article on Parkinson’s disease was released May 30.
DBS uses an electrode implanted beneath the skin to deliver electrical pulses into the brain more than 100 times per second. Although this technology was approved by the Food and Drug Administration more than 15 years ago, it remains unclear how it reduces tremor and other symptoms of movement disorders.
With the help of electroencephalography or EEG — electrodes placed on the scalp — study authors used new techniques to suppress the electrical signal associated with the DBS electrode. That enabled the first clear, non-invasive EEG measurements of the underlying brain response during clinically effective, high-frequency brain stimulation in humans.
The results show that nerves in the cerebral cortex, the outer layer of the brain, fire with rapid and precise timing in response to individual stimulus pulses. This suggests that DBS may synchronize the firing of nerve cells and break the abnormal rhythms associated with involuntary movements in Parkinson’s disease and essential tremor.
The newly identified rhythm was captured during effective DBS treatment, so it could represent a new physiological measure of the stimulation dose, say the authors. If validated, such a yardstick could help to guide the fine-tuning of DBS stimulator settings in patients for more lasting relief, fewer side effects and less-frequent battery-replacement surgeries.
"Though it’s clear that more work is needed to better understand these initial observations, we’re very excited by our findings because they may provide a biological marker for improvement in the symptoms of these patients," says Harrison Walker, M.D., assistant professor in the UAB Department of Neurology’s Division of Movement Disorders and lead author of the study.
In current clinical practice, stimulator settings are adjusted by trial and error, requiring careful observation of changes in symptoms over multiple clinic visits. But such immediate, visual feedback may not be available as DBS is applied to neurological or psychiatric conditions such as epilepsy, severe depression or obsessive compulsive disorder. In these diseases, an effective dose measurement could be especially useful in optimizing DBS therapy.
Devices could reveal inner workings of neurons and how they communicate with each other.
Automated assistance may soon be available to neuroscientists tackling the brain’s complex circuitry, according to research presented last week at the Aspen Brain Forum in Colorado. Robots that can find and simultaneously record the activity of dozens of neurons in live animals could help researchers to reveal how connected cells interpret signals from one another and transmit information across brain areas — a task that would be impossible using single-neuron studies.
A robot that can access the internal workings of neurons could be scaled up to allow 100 cells to be studied at a time. MIT McGovern Institute/E. Boyden/Sputnik Animation
The robots are designed to perform whole-cell patch-clamping, a difficult but powerful method that allows neuroscientists to access neurons’ internal electrical workings, says Edward Boyden of the Massachusetts Institute of Technology in Cambridge, who is leading the work.
ROBOTS developed in the safety of a laboratory can be too slow to react to the dangers of the real world. But software inspired by biology promises to give robots the equivalent of the mammalian amygdala, a part of the brain that responds quickly to threats.
(Image: SuperStock)
STARTLE, developed by Mike Hook and colleagues at Roke Manor Research of Romsey in Hampshire, UK, employs an artificial neural network to look out for abnormal or inconsistent data. Once it has been taught what is out of the ordinary, it can recognise dangers in the environment.
For instance, from data fed by a robotic vehicle’s on-board sensors, STARTLE could notice a pothole and pass a warning to the vehicle’s control system to focus more computing resources on that part of the road.
"If it sees something anomalous then investigative processing is cued; this allows us to use computationally expensive algorithms only when needed for assessing possible threats, rather than responding equally to everything," says Hook.
This design mimics the amygdala, which provides a rapid response to threats. The amygdala helps small animals to deal with complex, fast-changing surroundings, allowing them to ignore most sensory stimuli. “The key is that it’s for spotting anomalous conditions,” says Hook, “not routine ones.”
STARTLE has been tested in both vehicle navigation and robot health monitoring. In the latter, it can be trained to respond to danger signs, such as sudden changes in battery power or temperature. It has also been tested in computer networks, as a way to detect security threats, having been trained to identify the pattern of activity associated with an attack.
"A robot amygdala network could be useful," says neuroscientist Keith Kendrick of the University of Electronic Science and Technology of China in Chengdu. "Such a low-resolution analysis will sometimes make mistakes, and you will avoid something needlessly." But a slower, high-resolution analysis is also carried out, he says, which can override the mistakes.
Hooks says that STARTLE could be useful for any robots in complex environments. For example, a robot vehicle would be able to spot other drivers behaving erratically, a major challenge for conventional computing.
Source: NewScientist
Measurements of five protein biomarkers in the cerebrospinal fluid helped to differentiate Alzheimer’s disease from Parkinson’s disease with dementia and from dementia with Lewy bodies in a cross-sectional study of individuals at Swedish neurology and memory disorder clinics.
The diagnostic accuracy of this panel of tests in distinguishing Alzheimer’s disease from dementia with Lewy bodies “is at least in the same order of magnitude as that obtained with dopamine transporter imaging, and with a lower cost,” Dr. Sara Hall of the department of clinical sciences, Lund (Sweden) University, Malmö, and her associates wrote in a study published Aug. 27 in Archives of Neurology.
In addition, one of the five biomarkers in this panel appears to differentiate Parkinson’s disease from atypical parkinsonism such as that seen in progressive supranuclear palsy, multiple system atrophy, or corticobasal degeneration, the researchers noted.
Their results confirmed those of previous studies postulating that CSF total tau (T-tau) and phophorylated tau (P-tau) levels are higher in Alzheimer’s than in the other two dementias, whereas amyloid-beta (Abeta) 1-42 levels are lower in Alzheimer’s than in the other two dementias.
Columbia neurophysiologist David Sulzer took his first piano lessons at the age of 11 and was playing his violin and guitar in bars by age 15. Later he gained a national following as a founder of the Soldier String Quartet and the Thai Elephant Orchestra—an actual orchestra of elephants in northern Thailand—and for playing with the likes of Bo Diddley, the Velvet Underground’s John Cale and the jazz great Tony Williams.
From left, Brad Garton and David Sulzer discuss turning brain waves into music on WHYY/PBS in Philadelphia.
It was only after arriving at Columbia, however, that the musician-turned-research-scientist embarked on perhaps his most exotic musical venture—using a computer to translate the spontaneous patterns of his brain waves into music.
With the help of Brad Garton, director of Columbia’s Computer Music Center, Sulzer has performed his avant-garde brain wave music in solo recitals and with musical ensembles.
Last spring, Sulzer presented a piece entitled Reading Stephen Colbert at a conference in New York City sponsored by Columbia and the Paris-based IRCAM (Institut de Recherche et Coordination Acoustique/Musique), a global center of musical research.
Sulzer, a professor in the departments of Psychiatry, Neurology and Pharmacology, wore electrodes attached to his scalp to measure voltage fluctuations in his brain as he sat in a chair reading a book by the comedian. Those fluctuations were fed into a computer program created by Garton, which transformed them into musical notes.