5. Cerebellum
Part 3 of 4
James Knierim, Ph.D.
.
Cerebellum and Control Systems
What do the various symptoms of cerebellar damage have in common that reveal the function of the cerebellum? A number of different theories have been proposed. Recall the discussion in Chapter 1 of the ubiquitous use of sensory information in motor control. The cerebellum receives extensive sensory input, and it appears to use this input to guide movements in both a feedback and feedforward control manner.
In a feedback controller, a desired output is compared continuously with the actual output, and adjustments are made during the execution of the movement until the actual movement matches the desired movement. A common example of a feedback control system is the thermostat in your home (Figure 5.9).
Figure 5.9 A feedback control system, such as the thermostat in your home, is sufficientfor slow movements, such as posture. The myotatic reflex is an example of a feedback control system in the spinal cord. |
The thermostat is set to a desired temperature (e.g., 72°), and a thermometer measures the current temperature in the room. If the thermostat (the comparator) detects that the room is cooler than the desired temperature, it sends an error signal that turns on the furnace. If the comparator detects that the room is warmer than the desired setting, its sends an error signal that turns on the air conditioner.
Feedback control systems can produce very accurate outputs; however, in general they are slow. In order to change the output, the effector must wait until information is transmitted from the sensor to the comparator and then to the effector. At this point, another comparison is made, and the process continues. Consider further the thermostat example. If the temperature reads 5° cooler than desired, the thermostat can instruct the furnace to turn on at a moderate heat. It reads the new room temperature, and, if it is still too cool, it instructs the furnace to deliver more heat, and so on. Although this will eventually produce an accurate room temperature at the desired point, it takes a number of cycles to reach that point. One possible solution for quicker results would be to turn an enormous furnace on full-blast, such that is heats the room very quickly. This solution, however, can generate another problem. It will tend to cause the system to oscillate if the feedback pathways are slow. For example, assume that the furnace can heat the room at the rate of 5° per second, but that it takes 2 seconds for the thermometer to adjust to the new temperature, and for the new error signal to turn the furnace off. In those 2 seconds, the furnace has heated the room up 10°, and now it is too warm. So the error signal turns on the air conditioner, and it cools the room at 5°/sec. Of course, it also takes 2 sec to receive the feedback, and by the time it is told to shut off, it has cooled the room by 10°. You can see what happens: the system will be sent into an endless oscillation of being 5° too hot and 5° too cold. In order for a feedback system to work well, the transmission time of sensory information through the comparator to the effector must be rapid compared to the time of the action.
Feedback control systems work well only when the sensory feedback about the actual output is fast relative to the actual output. If the actual output is faster than the sensor’s ability to provide feedback, then the system will tend to oscillate between overshooting and undershooting the desired output. Thus, a feedback controller is useful for slow movements, like postural adjustments. The role of the myotatic reflex in posture maintenance is an example of a feedback controller in the spinal cord, and the cerebellum plays a role in coordinating these postural adjustments. Feedback control is not effective for most of the fast movements we make routinely (such as an eye movement or reaching out for a cup). For these movements, a feedforward controller is needed.
Feedforward control systems
In a feedforward control system, when a desired output is sent to the controller, the controller evaluates sensory information about the environment and about the system itself before the output commands are generated. It uses the sensory information to program the best set of instructions to accomplish the desired output. However, in a pure feedforward system, once the commands are sent, there is no way to alter them (i.e., there is no feedback loop). The advantage of a feedforward system is that it can produce the precise set of commands for the effector without needing to constantly check the output and make corrections during the movement itself. The main disadvantage, however, is that the feedforward controller requires a period of trial-and-error learning before it can function properly. In most biological systems, it is hard (perhaps impossible) to pre-program all of the possible sensory conditions that the controller may encounter during the life of the organism. Furthermore, the environment and conditions under which actions are made are constantly changing, and the feedforward controller must be able to adapt its output commands to account for these changes.
A feedforward control system is needed for fast movements, because a feedback system is too slow. |
Figure 5.10 |
Let us extend the thermostat example to see how a temperature controller operating as a feedforward system would work to raise the temperature of a room from 70° to 75°. The controller would use diverse sensory information about the environment before sending its command to the furnace (Figure 5.10). For example, it would read the current temperature, the current humidity level, the size of the room, the number of people in the room, and so forth. Based on this information, it would direct the furnace to turn on for a pre-set period of time, and that’s it. There would be no need to continually compare the current temperature with the desired setting, as the system has predetermined how long the furnace needs to be working in order to achieve the desired temperature. How did the controller obtain this information? A feedforward controller requires a large amount of experience in order to learn the appropriate actions needed for each set of environmental conditions. If on one trial it turns the furnace off too soon and the room does not reach the desired temperature, it adjusts its programming such that the next time it encounters the same environmental conditions, it turns the furnace on for a longer period of time. Through many such instances of trial and error learning, the feedforward system creates a “look-up table” that tells it how long the furnace needs to be active under the current conditions. The key distinction between a feedback and feedforward system is that the feedback system uses sensory information to generate an error signal during the control of a movement, whereas a feedforward system uses sensory information in advance of a movement. Any error signal about the final output is used by the feedforward system only to change its programming of future movements.
Contact the author(s) at nba_course@uth.tmc.edu |
