Writ Large – and Small


You are here

Writ Large – and Small

When you jot notes on a pad, you move different sets of muscles than when you write in large letters on a blackboard. Yet your handwritten a on the pad is formed with the same unique curves as the a on the blackboard. How does our brain encode the complex motions of writing? Does it employ a single set of instructions for both notepads and boards – or multiple ones? This question has been debated in neurobiology circles for a while; but a team of Weizmann Institute scientists that included both mathematicians and neurobiologists has now provided a surprising answer. Their results appeared recently in Neuron.

One of those scientists is research student Naama Kadmon Harpaz, whose work spans both mathematics and neurobiology. She began her graduate studies in the Neurobiology Department and is now in the Computer Science and Applied Mathematics Department  working in the group of Prof. Tamar Flash. Also participating in the study was neurobiologist Dr. Ilan Dinstein, a former postdoctoral fellow in Prof. Rafi Malach’s lab who is now on the faculty of Ben-Gurion University of the Negev.
To conduct their study, Kadmon Harpaz, Dinstein and Flash asked volunteers to trace out three letters, in both large and small versions, on a concealed touch screen. While the volunteers were writing, their brains were being scanned in the Institute’s functional magnetic resonance (fMRI) apparatus to capture the brain activation underlying the physical act. In addition, the team analyzed the movements themselves: the geometrical and temporal features of the writing and its kinematics (paths and speeds).

When the results were added together and analyzed, the findings were conclusive: Certain areas in the brain encode the act of writing both small and large letters in a highly similar way. In mathematical terms, the code is “scale invariant” – that is, the pattern of activation looks very similar for large and for small, fast and slow. A one-size-fits-all code meshes with earlier research which had suggested that, even though the muscles used are different, the basic kinematics of the movement when writing large or small are the same.
Top: Movement traces from single trials of three representative subjects. Orange and blue are small and large letters, respectively. Bottom: Mean movement traces after performing the Procrustes transformation (small mean is scaled to the size of the large mean) revealing a similarity in the path shape across scales


Simplified control

(l-r) Prof. Tamar Flash and Naama Kadmon


The fMRI results highlighted two areas of the brain that exhibited this “scale-invariant” encoding: One, called the anterior intraparietal sulcus (aIPS), is known to be important for such functions as hand-eye coordination and the planning of movement; the other, the primary motor cortex (M1), is thought to be the executor of the hand’s action. In the hierarchy of the brain, aIPS is said to be “upstream” – that is, it deals with more abstract information. Thus one might expect this brain area to encode symbols in a scale-invariant manner. But the scientists were surprised to find that the downstream area, M1 – thought to be a source of purely instantaneous neural commands for the more mechanical aspects of movement that are sent to the spinal cord and from there to the muscles – also encodes the ensuing action with scale invariance.
The researchers think that the encompassing scale invariance they identified acts to simplify the “control problem” and increases the efficiency of neural computation.  

Prof. Flash: “Concerning the generation of movement, the brain is thought to function from the top down, from abstract representation to physical output; yet we found this abstract coding in a supposedly downstream area. We think, on the one hand, that the brain is more of a network and less of a strictly top-down hierarchy than previously thought and, on the other hand, that our brain extends its abstract coding patterns to our seemingly concrete actions in the world around us.”
This experiment, say the researchers, is among the first to use fMRI to investigate motor control in humans. Most fMRI research looks at the brain’s reaction to such input as photos or video clips, while research into output – e.g., motor control – most often involves electrodes recording the activity of single or multiple neurons in non-human primates. By using fMRI to examine motor control in human subjects, says Dinstein, “we were able to see many areas in the brain at once. We could observe what each was doing in relation to the other.” 
According to Flash, these findings may have relevance for a number of areas of scientific research. For example, the insight they provide into the workings of the brain could be applied to robotics and bio-robotics to improve efficiency and enable a wide range of complex movements. In addition, they may aid in understanding those movement disorders that originate in the brain, among them Parkinson’s disease. They may also lead to a better understanding of the “motor memory” we use daily without a conscious thought. Further questions to be answered include: At what stage does learning to write enable a scale-invariant activity? How does such motor learning take place in the brain?    

Prof.Tamar Flash is the incumbent of the Dr. Hymie Moross Professorial Chair.