Source : McGill University
When you are about to collide into something and manage to
swerve away just in the nick of time, what exactly is happening in
your brain? A new study from the Montreal Neurological Institute
and Hospital – The Neuro, McGill University shows how the brain
processes visual information to figure out when something is moving
towards you or when you are about to head into a collision. The
study, published in the Proceedings of the National Academy of
Sciences of the USA (PNAS), provides vital insight into our sense
of vision and a greater understanding of the brain.
Researchers at The Neuro and the University of Maryland have
figured out the mathematical calculations that specific neurons
employ in order to inform us of our distance from an object and the
3D velocities of moving objects and surfaces relative to
ourselves. Highly specialized neurons located in the brain’s
visual cortex, in an area known as MST, respond selectively to
motion patterns such as expansion, rotation, and deformation.
However, the computations underlying such selectivity were unknown
until now.
Using mathematical models and sophisticated recording
techniques, researchers have discovered how individual MST neurons
function. “Area MST is typical of high-level visual cortex,
in that information about important aspects of vision can be seen
in the firing patterns of single neurons. A classic example is a
neuron that only fires when the subject is looking at the image of
a particular face. This type of neuron has to gather information
from other neurons that are selective to simpler features, like
lines, colors, and textures, and combine these pieces of
information in a fairly sophisticated way,” says Dr. Christopher
Pack, neuroscientist at The Neuro and senior author.
“Similarly, for motion detection, neurons have to combine
input from many other neurons earlier in the visual pathway, in
order to determine whether something is moving toward you or just
drifting past.” The brain’s visual pathway is made up of
building blocks. For example, neurons in the retina respond to very
simple stimuli, such as small spots of light. Further along the
visual pathway, neurons respond to more complex stimulus such as
straight lines, by combining inputs from neurons earlier on.
Neurons further along respond to even more complex stimulus such as
combinations of lines (angles), ultimately leading to neurons that
can respond to, or recognize, faces and objects for example.
The research team found that a remarkably simple computation
lies at the heart of this sophisticated neural selectivity: MST
neurons appear to be capable of performing a multiplicative
operation on their inputs. These inputs come from neurons one step
earlier in the visual pathway, in a well-studied area known as MT.
In other words, the inputs of MT neurons are multiplied in order to
get the output of MST neurons. This turns out to be
remarkably similar to what has been observed in other brain areas
and in other species, suggesting it may reflect a general strategy
by which brains process sensory information. “One interesting
aspect of the computation is that it appears to be about the same
as what other people have found in flies and beetles, suggesting
that evolution solved this problem once, at least a few hundred
million years ago.”
“We developed a new motion stimulus with a morphing pattern flow
(e.g. dots on a screen that are expansive, swirl around, circle to
the right, contract etc) and recorded MST neurons responding to
these stimuli,” says Patrick Mineault, Ph.D. candidate at The
Neuro and primary author on the study. “We circumvented the issue
of increasing complexities of calculations along the various steps
of the visual pathway by incorporating known data from neurons just
one step earlier in the pathway - area MT, which precedes MST. As
we now had measurements of the output of the MST neurons from the
study’s recordings, and already knew the input of MT neurons, we
could calculate the math linking these two functions – and it turns
out to be a multiplicative function.” The mathematical models
successfully account for the stimulus selectivity of some of the
brain’s complex motion neurons - which are vitally important in
helping navigate us through the world.
This work was supported by the Canadian Institutes of Health
Research, Le Ministre du Développement économique, de l'Innovation
et de l'Exportation du Québec , the National Science Foundation,
the Fonds de recherche du Québec – Santé, and the Fonds de
recherche du Québec – Nature et technologies.
The Montreal Neurological Institute and
Hospital :
The Montreal Neurological Institute and Hospital — The Neuro, is
a unique academic medical centre dedicated to neuroscience. Founded
in 1934 by the renowned Dr. Wilder Penfield, The Neuro is
recognized internationally for integrating research, compassionate
patient care and advanced training, all key to advances in science
and medicine. The Neuro is a research and teaching institute of
McGill University and forms the basis for the Neuroscience Mission
of the McGill University Health Centre. Neuro researchers are
world leaders in cellular and molecular neuroscience, brain
imaging, cognitive neuroscience and the study and treatment of
epilepsy, multiple sclerosis and neuromuscular disorders. The
Montreal Neurological Institute was named as one of the Seven
Centres of Excellence in Budget 2007, which provided the MNI with
$15 million in funding to support its research and
commercialization activities related to neurological disease and
neuroscience.