What if we could reliably measure children’s brain circuits to predict reading ability just as we measure their height and weight to predict physical development? That is a question Brian Wandell has been exploring for the past 30 years – how to use neuroimaging techniques like MRI to quantify the properties and activity of living brain tissue, in particular in the visual cortex.
This year’s recipient of the CNS George A. Miller Prize, Wandell of Stanford University has recently turned his interests to reading and the developing brain. His latest work includes a longitudinal study of children aged 8 to 12 years old, in which he and colleagues have been measuring the development of brain structures and responses in the visual cortex. The goal is to create brain circuitry profiles that can predict reading success and to ultimately inform training interventions for those who have difficulty reading.
In a preview of the award lecture he’ll deliver next month at the CNS conference in New York City, Wandell spoke with CNS about his work on reading development, the growing body of computational neuroimaging work in the visual cortex, and how human neuroimaging studies could help inform animal research (yes, that’s right – human studies → animal ones!).
CNS: How do you define computational neuroimaging to a non-neuroscientist?
Wandell: Computational neuroimaging mean that we intend to use our work to build computable models (e.g., software programs) that predict the neuroimaging measurements from knowledge of the stimulus and general experimental conditions (e.g., the task). Perhaps contrasting this approach with a different approach will be helpful.
The media frequently reports on clinical evaluations that compare two treatments. Is it better to take this drug or that? Do people lose more weight with this diet or that? Which reading therapy helps poor readers more? These comparisons benefit shoppers, and they offer practical advice given the current options.
Basic scientists ask a different question. How does the system work? When we take this drug, which cells are influenced? And then, why does changing the activity in these cells influence our ability to think, feel, and act? The difference is between receiving a review of cars (clinical evaluations) and detailed drawings that show how the system works (basic science).
Basic scientists add the word computational when we aim to understand the system with enough precision that we can implement a computer program to predict the system’s performance. The enormous growth in computational power has motivated many different scientific fields to aim for this objective. In the context of neuroimaging, we call the approach computational neuroimaging.
CNS: Your CNS 2016 award lecture is titled “Computational neuroimaging: Quantifying brain tissue and modeling activity in the living human brain.” What does it mean to “quantify brain tissue”?
Wandell: Many neuroimaging studies rely on comparisons between two signals. For example, the neuronal cell bodies in cortex look a little darker than the myelinated axons that emerge from these cells. Hence, we call the cell bodies gray matter and the axons white matter.
Such qualitative comparisons are helpful but limited. For example, a neurologist or radiologist reads an MRI scan of a multiple sclerosis patient and makes clinical recommendations using the number of UBOs. It comes as a disappointment to learn that UBO means “unidentified bright object.” It would be like hearing that you bought an item on sale, but not knowing how much you paid.
Around the world, scientists and engineers are developing neuroimaging methods that provide absolute amounts, not just relative levels. There has been real progress in measuring certain quantities – such as the density of cell membranes, the orientation of these membranes, and the chemical composition of these membranes. Improving these methods and simplifying them for practical use is what I mean by quantifying brain tissue.
CNS: Why is the visual cortex the focus of your work?
Wandell: Vision science has a long and beautiful history of computational modeling. It is often the first part of neuroscience to develop advanced methods. As the great scientist William Rushton wrote “You will need no words of mine to convince you how precious are your eyes.” Vision science is an important field that offers an opportunity to develop quantitative methodologies on a firm foundation and in a context that matters deeply.
CNS: What from your most recent findings will you be sharing in your award lecture that you can share with us now?
Wandell: My talk will provide an encouraging review of the state of neuroimaging in visual cortex, documenting the successes over the last 25 years, and what I think is a fruitful next step. I will then turn to the specific issue of understanding the neural circuitry that is essential for rapidly recognizing word forms. There are new findings and modeling on the reading circuits that build upon the methods and findings of the last 25 years.
CNS: Can you give an example that illustrates the unique contributions of the computational approach to understanding some aspect of brain function/health that was previously elusive with noncomputational approaches?
Wandell: The visual cortex comprises about 20 percent of the human brain. Visual cortex is organized into a collection of about 25 small regions, some about the size of a matchbook, some larger. Each responds in its own characteristic way to signals from the retina, either the whole retina or just the central, high-acuity region. Computational methods have been used to identify the position of these regions in the living human brain, and we are at a level of precision that we could not have imagined when the research program began in the early 1990s. The field is now actively engaged in understanding the processes in these regions and how the regions communicate with one another.
CNS: Which near-term and long-term applications are you most excited to see come to fruition with computational neuroimaging?
Wandell: Our goal is to develop methods to measure the living human brain structures and activity that match the level of quality and capabilities of our much older partner disciplines, such as chemistry and physics. A specific application that guides my evaluation of the progress is to use these findings to better diagnose reading disabilities so that the appropriate training can be found for each child.
CNS: What do you most want the public to understand about your work?
Wandell: The visual regions in cortex and the connections between these regions develop enormously as children grow. Right now I am passionate about using computational neuroimaging to understand what is different about children who have difficulty learning to read. Each child is unique, and computational neuroimaging methods permit us to measure each child’s circuitry and compare it with the circuitry in other children. Just as we can measure a child’s temperature or height with respect to other healthy children, we should be able to measure the status of the brain circuits for reading. Computational neuroimaging can provide a biological basis for understanding each individual, and then finding the most effective training method for that child.
CNS: What do you most want fellow neuroscientists to understand about this work?
Wandell: Rigorous computational neuroimaging measurements in the human brain can motivate neuroscience experiments in animal systems. Neuroscience frequently begins with studies in rodent systems and then hopes that these findings apply to humans. Starting from the human findings, we are more likely to close the loop between electrophysiology, cells, and molecular measurements in animal systems and human measurements.
-Lisa M.P. Munoz
Brian Wandell will give his George A. Miller Prize lecture on Saturday, April 2, 2016, 4:00-5:00 pm, in the Grand Ballroom East at the New York Hilton Midtown Hotel, as part of the CNS 23rd annual meeting.