Funded Grants

Computing time in the cerebellum

When Bill Mazerowski hit his famous home run to win the 1960 World Series, he did something extraordinary. To help the Pittsburg Pirates defeat the New York Yankees, he swung his bat at exactly the right time to hit the ball a very long way. Had he swung only a few milliseconds earlier, he would have hit the ball foul. Had he swung only tens of milliseconds later, he would have missed the ball entirely. Unfortunately for the Yankees, Mazerowski timed his swing perfectly.

The importance of precision timing is obvious in professional baseball, but it is also important in many everyday motor, perceptual and cognitive acts. When you reach for your coffee, your fingers close just at the time your hand arrives at the cup. If someone pauses too long before answering your question, you wonder about the truthfulness of their response. In speech perception, the only difference between the sound “pa” and the sound “ba” is the voice onset time, the time the vocal cords start vibrating relative to the release of air from the lips. If that time is longer than about 20 ms, we perceive the sound as “pa”; if it is shorter, we perceive “ba.”

Since the accurate perception and execution of timing is of critical importance for so many motor and cognitive acts, the brain must have a way of measuring time. This is not a simple task—until the British clockmaker John Harrison produced his “chronometers” in the mid-18th century, the difficulty of building accurate mechanical clocks was the main limitation to navigation. Our experiments ask how the challenge of accurately measuring time is addressed by the biological hardware of the brain.

Research on this question suggests that there is no central brain clock; rather, time is measured for different purposes at distributed sites in the brain, including the cerebral cortex (in particular parietal cortex), the basal ganglia, the hypothalamus, the brainstem, and the cerebellum. The neural circuits in these brain regions are able to measure time over a range of different time scales, from microseconds to days.

Our research project focuses on the measurement of time in the cerebellum. Previous empirical and theoretical studies suggest that one important function of the cerebellar circuit is the computation of time on a scale of tens of milliseconds to hundreds of milliseconds, in the support of both perception and action. Functional imaging studies of brain activity in humans have reported activation of the cerebellum during tasks that require subjects to judge the duration of events (time discrimination) and during tasks that require subjects to produce movements with particular inter-response intervals (time production). Additional evidence for a role of the cerebellum in timing has come from studies of human patients with cerebellar lesions. These patients are impaired on time production tasks such as rhythmic tapping and the production of discrete movements with a specified duration, and they are also impaired on time discrimination (for a review, see Ivry et al., 2002; Llinas and Welsh, 1993). Thus, a large body of empirical evidence suggests that the cerebellum computes time in the service of perception and movement. However, the mechanism by which the intricate circuit architecture of the cerebellum performs this computation is still a matter of theory.

A compelling model (Medina et al., 2000) proposes that a specific population of interneurons in the cerebellar cortex, called Golgi cells, plays a key role in creating a representation of time in the cerebellum. This model suggests that the pattern of interconnectivity between the Golgi cells and the input neurons of the cerebellar cortex, the granule cells, causes different subsets of granule cells and Golgi cells to be sequentially activated in response to a constant input to the cerebellum. In this way, the specific pattern of active granule cells and Golgi cells could indicate, not just whether a particular cerebellar input is present (reflecting the presence of a particular sensory stimulus), but also how long that stimulus has been present. Downstream neurons could use this neural representation of the time since stimulus onset to appropriately time the motor response to the sensory input. One can imagine that the sight of the Yankee’s pitcher releasing the ball triggered the sequential activation of different subsets of granule cells and Golgi cells in Bill Mazerowski’s cerebellum. After years of practice, the downstream motor circuits had learned to recognize exactly which subsets of neurons signaled the appropriate time to initiate his swing for success in hitting the ball, and the connections from that subset of neurons to the motor circuitry had been strengthened. The same mechanism also may be used for some aspects of perceptual timing, a possibility supported by the observation that the cerebellar cortex is activated during language tasks.

The above model is consistent with lesion studies, which implicate the cerebellar cortex in timing. However, the cerebellar cortex contains seven different classes of neurons, and therefore a more detailed analysis is required to determine the contribution of each of class of neuron in this network to the computation of time. Our project undertakes this detailed analysis, using an approach that combines the power of molecular-genetics with more traditional behavioral and electrophysiological approaches of systems neuroscience. One straightforward approach we will use is simply to record from individual Golgi cells during a simple motor timing task to determine whether the pattern of Golgi cell activation does indeed encode temporal information about the stimulus that triggers the movement. In addition, we will capitalize on the recent development of molecular biological tools for precisely perturbing the function of specific classes of neurons within a circuit. We will use these tools to generate a line of transgenic mice in which the Golgi cells can be functionally removed from the cerebellar circuit, and we will examine the consequences of this manipulation for the coding of time in cerebellar neurons and for the ability to accurately time movements. We will compare the effects of functionally removing the Golgi cells with the effects of functionally removing other classes of neurons from the cerebellar circuit. This systematic dissection of the cerebellar circuit will significantly advance our understanding of the role of each class of neuron in cerebellar computation. In particular, we will gain insight about how the circuit architecture of the cerebellum supports the measurement of time with an accuracy that enables some of us to hit baseballs out of the park.