May 13, 2007

Cerebellum: Electronically Speaking

an EPROM chipSometimes journeying into physiology textbooks can be a deja vu in electronics. The other day I was studying the way cerebellum learns from its past 'karma' and memorizes the best setting for an action. I was surprised to find that its learning resembled a lot with the op-amp (operational amplifier) ic's (LM741, for example) and its memorizing resembled to that of the EPROMs (erasable programmable read only memory, a picture of which is shown here on the left).

The cerebellum has one of the largest nerve cells of the body, the Purkinje cells, in addition to other types of cells. Purkinje cells have numerous tree-like branching dendritic processes. They receive two types of electrical connections; from the mossy fibers (about 250000 to 1 million fibers for each Purkinje cell) and climbing fibers (ONLY one for each cell).

The cerebellum fine tunes the movement that accompanies a certain task in the following way. The brain (lateral portions of the cerebellum and the basal ganglia) 'plans the action' even before we start an action. After the task has been executed, the cerebellum calculates the 'error', the difference between the planned trajectory and the achieved output in much the same way a does.
In the figure on the left, the op-amp is configured as a negative feedback, through resistance R2 which feeds a portion of output voltage back into its inverting input (since input is fed into its - terminal) amplifier. This way, the op-amp can have both a fraction of the output and input at the same pin, thereby the gain and other parameter remain 'stable'.

A still better analogy phase locked loop operation is that of the phase locked loop, the output phase of which 'locks' to the input frequency (diagram on the left). Likewise, the cerebellum too matches its expected action with that of the resulting action. The cerebellum is 'happy' with the performance when the error is minimum. At this moment of bliss, the climbing fibers fire for a long duration and with a characteristic waveform: a spike followed by a long trail. Now the information is written permanently.

Doesn't our experience of programming a 2716 (EPROM) tell a similar story; a peak programming voltage and a normal working voltage? When we wanted to program an EPROM, we needed to give it a high programming voltage in the beginning, followed by a steady 5 voltage thereafter till the EPROM 'learned'.

It doesn't end here. In the cerebral cortex, memory is consolidated in sleep, when it receives a spike voltage: the k-complex in the NREM stage of sleep. It seems that we have a micro, or may be a pico-controller in our brain.

To be contd.
Post a Comment