Botond Roska and James Poulet

The neurobiology of gazing out the window of a train

You might feel like you're daydreaming as you watch the landscape flash by through the window of a train, but your eyes and brain are engaged in an active process that is yielding sharp insights into neurobiological organization. In November's edition of the MDC Lectures, Neurobiologist Botond Roska described how neuronal structures form the basis of the brain's ability to "compute" sensory stimuli.

Botond Roska and James Poulet. Image: Dario J. Laganà.

The metaphor of the machine has guided scientific thinking about the brain ever since Descartes divided human nature into the mind and body and pondered how one might control the other. In today's age of sophisticated computers and "artificial intelligence," terms like networks, modules, and computing have very powerfully shaped the way scientists think of mental behavior. If these are more than just metaphors, their underlying processes must be rooted in the physical organization of the brain. This idea has spurred researchers such as Botond Roska, of the Friedrich Mieschner Institute for Biomedical Research, to search for structures within networks of neurons and tissues that serve as the "computational" foundation of the brain's management of sensory input.

Roska visited the MDC on Nov. 25 in the framework of the "MDC Lectures" series, in which each of the institute's research areas hosts one prominent speaker per year. This time the guest was welcomed by James Poulet and Carmen Birchmeier from the program "Diseases of the Nervous System." In a talk entitled "The first steps of vision: cell types, circuits and repair," Roska dissected some of the early steps in visual processing that configure the way mammals perceive and respond to particular stimuli. The work draws on a range of methods including genetics, computational approaches, and the use of laboratory-grown "retinoids" and brain tissue.

Starting at a very basic level within the retina, Roska described arrangements of single rod and cone cells that are initially stimulated by visual signals. The structure of their connections, he said, amount to a first level of neural computation, which is partly handled by the distribution of tasks among over 70 types of cells in the retina. Roska traced the route that nervous impulses take on their way to specific regions of the visual cortex and some of the types of filtering they experience along the way.

Visual stimuli are not just passively received; they also trigger behavioral responses. To study the genetic basis of this connection, Roska's lab needed to find a simple, measurable type of response. They found it in a behavior called the horizontal optokinetic reflex. You can experience this while looking out the window of a train and watching the landscape passing quickly by. The reflex causes your eyes to dart rapidly back and forth, and in the process something interesting happens: in one direction you take notice of objects and gather information, but not in the other.


This reflex is disrupted in a hereditary condition called human congenital nystagmus. People who affected are unable to control the horizontal movement of their eyes, which don't collect information in either direction. The problem has been linked to defects in a gene called FRMD7. The scientists discovered that FRMD7 protein is produced specifically in neurons called starburst amacrine cells, named for their sprawling shapes, which connect the retina to neurons called direction-sensitive ganglion cells. Precise genetic and functional studies have allowed the lab to show how mutations in FRMD7 in mice can disrupt the structure of the connections between these two types of cells. This interrupts the reflex that guides the eye along the horizontal axis, but it leaves vertical activity and information processing unaffected. One thing this reveals, Roska emphasized, is that specialized cell types are assigned precise tasks in the processing that underlies sensory perception and the management of behavioral responses. Further work on the functions of the types and their connectivity should reveal additional steps and layers of the "computational system" that makes this possible.

The talk provoked a stimulating discussion, just one sign of appreciation for the way Roska and his colleagues have approached one of the most challenging themes in modern research. Connecting the brain's structure to behavior will require work on highly structured systems that can be precisely mapped, measured and controlled. The type of elegant, interdisciplinary work Roska is pursuing on the retina and its links to the visual cortex seems like a very good place to start.