Retrieving knowledge from the archives of the mind

21 May 2012

At the time of testing, none could recall any aspect of either language, in which they had once been fluent as children.   They were played recordings of particular sounds and later replayed those recordings and given feedback as to whether they were right or wrong.   After between 15 and 20 sessions, volunteers were able to recall particular phonemes, suggesting that within a relatively short period of time, it is possible to relearn a language. 

“The thing I found intrinsically surprising is the idea that children could entirely lose a language,” says Professor Bowers.  “We were able to show that with relatively little training, people could relearn a specific aspect of the language; so they hadn’t really lost that skill in the most profound sense, even if in the functional sense, they had.  If they wanted to learn that language again, they could, because if they could learn the most difficult aspects, the rest should follow.” 

Professor Bowers’ is especially interested in how we perceive, comprehend and consolidate language in the written and spoken form.  He believes it is down to perceptual learning and the way information is coded in the brain by individual neurons and neural connections: “At a young age, people become experts at the sounds in their language but the process of getting good at one language often leads to a distortion of the perceptual space, meaning that you reorganise the sounds to maximise your skills in one language, which has a corresponding effect that leads you to be bad at another language. 

“It’s a perceptual skill; just as people are good at recognising people who are of the same race, or gender or age, you get good at perceptual domains that are important to you but sometimes that has a cost on other skills.”

In related work, Professor Bowers reviewed a series of neurobiological studies that are typically taken to discredit the idea of grandmother cell theory, which proposes that a single neuron can represent one complex perceptual category and is responsible for our ability to recognise a single object or face, like the face of our grandmother.  The theory was originally dismissed as an oversimplification of human experience.  However recent studies have demonstrated that it is biologically plausible and computer simulated studies carried out by Professor Bowers further support the idea: 

“There is neurobiological research being carried out showing these cells that in the brain respond to highly selective information.  There are lots of animal studies showing the same thing, that you can predict what the animal is looking at based on which neuron is fired.  Using computer simulations we’re trying to show that if you want a system that can code multiple things at the same time and support short term memory, the model has to learn grandmother-like cells, otherwise it is computationally difficult for the model to succeed.”

When it comes to brain training and applying theory to education, Professor Bowers believes that assessing behaviour changes rather than changes in the brain is the most functionally relevant approach, arguing that pure psychology rather than neuroscience should be the focus: “People point to the fact that the brain has changed as evidence that a training procedure worked.  But the only relevant question is whether the child’s behaviour improved – do they read better or not?   Sometimes it’s relevant to do modelling and sometimes it’s relevant to do neuroscience, but often, beautiful psychology experiments by themselves are the most informative.”

Please contact Aliya Mughal for further information.