The Roaches Spike Back

This is a post I started about a year ago, but never got around to finishing it, so here it is in due time.

Every so often I teach an undergraduate course which deals with some of the seminal experiments done in the field of neuroscience, and how they shape the way we understand contemporary neurobiology. I usually contrast these to modern versions of the experiment which then either support or overturn the established dogma. Sometimes we try and replicate some of these experiments in class. I’ve written about my experiences in having my students replicate some of Cajal’s original findings about neural structure by doing some of their own neural drawings under a microscope. Today I’m going to tell you about our experiences in replicating some seminal findings by neuroscientist (and nobleman) Lord Edgar Adrian. Adrian did some experiments in the 1920’s that described how the nervous system encodes information. What he did is he basically hooked an isolated frog leg muscle with an attached sensory nerve to a device that amounts to a series of vacuum tube amplifiers. These amplifiers are able to record electrical impulses from the frog nerve and then cause a little bit of mercury inside a capillary tube to bob up and down in proportion to the amount of voltage recorded. This was then captured on film, which allowed him to have a continuous record of electrical activity in the nerve as it changes over time. Here’s a schematic of the device from one of Adrian’s papers (click to enlarge):


Adrian 1926, 1929 – Valve amplifier and capillary electrometer


Adrian then hung little weights from one end of the muscle —thus stretching it— and recorded the output of the sensory nerve. What he then found is one of the central principles in nervous system function. He found that when the muscle was stretched the nerve would fire discrete electrical impulses, and the heavier the weight (meaning the muscle is stretched more) the faster the rate of these impulses was. These impulses are known as action potentials and the phenomenon is known as rate coding. Rate coding basically means that the rate of action potential generation is proportional to the strength of a sensory stimulus. Larger stimuli result in a faster rate of action potentials. This finding is shown below:


Rate Coding, as described by Adrian (click to enlarge). Top graph shows the responses resulting from different weights hung from the frog muscle. The squiggles show changes in voltage over time. Those bumps are the action potentials. Notice that it isn’t that clear cut that these are discrete units, but whatever. The bottom graph shows the relationship between response strength (x-axis) and stimulus strength (y-axis) for several experiments.


The other important observation was that action potentials were discrete units with similar size and shape. So stimulus strength was not represented as larger acton potentials, but rather as more action potentials, as shown here:


These are voltage recordings showing spikes as discrete units. In this example both rate coding and unit consistency is much more evident.


Finally, Adrian also observed that the rate of action potentials tended to slow down if the duration of the stimulus was sufficiently long. Thus, if he pulled the muscle for several seconds, the nerve would fire rapidly, but then would eventually slow down. Adaptation is the reason why you don’t continuously feel your clothes, or why you can still see changes in brightness in bright daylight or have a conversation in a noisy street. The nervous system tends to adapt to continuous stimuli and is better at detecting change (like in Jurassic Park when the dude gets eaten by a T-rex when he makes a run for it, while the guys that stay put don’t get eaten). Adrian’s demonstration is shown below:


Adaptation of spike rate. Here spike rate is in the Y-axis, while the X-axis is time. The bar on the top shows the duration of the stimulus. Notice that the rate of spiking decreases after several seconds.


So now the question was, could we replicate these seminal findings in our classroom? Enter Backyard Brains. Backyard Brains is a nifty little company that makes these somewhat inexpensive amplifiers called spiker boxes, a modern and el-cheapo version of Adrian’s valve amplifier. Rather than vacuum tubes a spiker box uses what most modern amplifiers use, which is a series of differential amplifiers, in order to amplify electrical signals in the nerve of a cockroach leg. A schematic is shown here. You can then poke the little hairs on the leg, which the roach uses for touch, to generate a sensory stimulus and record the resulting activity in the nerve in the leg. We had some trouble at first with the spiker box. The first one didn’t really work, we could hardly get a signal, but they promptly sent us a replacement and we were on our way. Using free audio-processing software we were able to record the waveforms resulting from the neural activity.

Adrian’s main findings were: action potentials as discrete units, rate coding, and adaptation. How did the spiker box hold up? As soon as we stuck the metal pins (known as electrodes in the local parlance) in the leg we observed little spontaneous action potentials, also called spikes, and indeed these seemed to be of roughly uniform shape and size. Here are some examples:



Notice how some of the spikes are big and some are small. Could Adrian be wrong? Probably not, more likely we are picking up activity from a couple of different sensory nerves and the size of the spikes depends on how close we are to the nerve itself. How about rate coding? Below is an example of a response resulting from gently blowing air onto the leg hairs compared with when you blow air vigorously. A weak and strong stimulus:


Rate coding and adaptation in cockroach leg nerve. The horizontal bar shows the duration of the sensory stimulus, a puff of air. The top response is to a gentle puff, while the bottom one is to a vigorous puff.


Notice how the weak puff of air causes the nerve to spike at a slower rate. Also, look at how the stimulus adapts after the puff of air has been going on for a while. This is more noticeable with the strong stimulus. Rate coding and adaptation!

So there we have it, the nervous system still works as we thought it did. This type of experiment is part of a field which neuroscientists call electrophysiology, and it is what I spend a lot of my time doing, but with much more expensive machines. I think it was a nice way to show the students how one can design an experiment to find specific principles and also to give them some first hand experience into how noisy data can look. Although we found the observations we had predicted, it took quite a bit of finesse to get the responses to look just right and to find good examples, and to sort these out from all the noise that goes on in the nervous system. And this is the kind of thing you get only from experience and from fiddling around with your equipment and preparation. Definitely not from a textbook. So go, get yourself a spiker box and see what you can learn.


Further Reading

Adrian, ED. The impulses produced by sensory nerve endings: Part I. J Physiol. 1926 Mar 18;61(1):49-72.

This entry was posted in Uncategorized. Bookmark the permalink.

5 Responses to The Roaches Spike Back

  1. Tim says:

    Thanks for the nice write-up; I was familiar with Adrian’s experiment but not in the detail you described. I am actually working on conduction velocity experiments with earthworms right now…it’s a bit of a trickier preparation than the cockroach as the giant earthworm axons fire very sparsely and are highly sensitive to depth of anesthesia. Plus, for reasons I don’t understand yet, the earthworm recordings are much more sensitive to electromagnetic noise (is it a better antenna?). But, since the earthworm is one long “tube within a tube” it’s easy to measure conduction velocity once the set-up is working. BTW, facts! I never new this, but earthworm axons are actually myelinated.

    • namnezia says:

      Probably all the wet squishiness makes for a good antenna. Are you dissecting the nerve cord and resting it on the electrodes?

      • Tim says:

        At first I was doing the dissection under a scope, as a reality check to make sure I had spikes, but once I knew what to listen for (that sparse coding will really throw you off…one-two spikes every 3-4th time you touch the end of the worm??!!), blindly inserting the needles works fine. I’m writing it up and will post on our experiment section of our website in a month or so. It’s a bit involved as you have to make a two-channel SpikerBox, use a DIY faraday cage, etc..

  2. DrugMonkey says:

    DUDE! You got one???

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s