Closing In On Why Bogie Quit Talking


[After 25 years or so of much talking and companionship, Bogie, the African Grey in our house, has suddenly quit talking – boo-hoo — a sad state of affairs.]


It’s time for another paradigm change. “These findings suggest that a fundamental assumption of current theories of motor coding requires revision,” as the Abstract of a new paper in the Proceedings of the National Academy of Sciences concludes. Neuroscientists from Emory University have uncovered another coded signaling system, this time in nerves and muscles. The paper’s categories include “Computational Neuroscience” and “Information Theory.”

Neurons and muscles have a strong relationship. To get a bicep to flex, or a diaphragm to bend for breathing, the muscles involved need to be triggered. The triggers come from nerves connected to the muscle fibers. Until this paper came along, most neuroscientists figured that the brain just sped up the “spike rate” of pulses to the muscle to get them to respond. The emerging view is much richer.  It’s not just the rate; it’s the timing.

A crucial problem in neuroscience is understanding how neural activity (sequences of action potentials or “spikes”) controls muscles, and hence motor behaviors. Traditional theories of brain function assume that information from the nervous system to the muscles is conveyed by the total number of spikes fired within a particular time interval. Here, we combine physiological, behavioral, and computational techniques to show that, at least in one relatively simple behavior–respiration in songbirds–the precise timing of spikes, rather than just their number, plays a crucial role in predicting and causally controlling behavior. These findings suggest that basic assumptions about neural motor control require revision and may have significant implications for designing neural prosthetics and brain-machine interfaces. [Emphasis added.]

Working with six male Bengalese finches that were anesthetized, the researchers monitored their breathing while recording neural spikes to the lungs. They were able to stimulate the motor neurons arbitrarily in vivo and watch what happens. This is delicate work; they had to work at 250 micro-amp levels. To locally block certain nerve-muscle junctions, they applied curare — the compound Brazilian hunters use on poison darts — but not enough to paralyze the poor birds! (How do you say that in scientese? “Applying too much curare and fully paralyzing EXP [expiratory muscle group] would endanger the wellbeing of the animal.”)

Next, they analyzed triplets of spikes where the middle spike was variable. They wanted to test whether a “neural code” exists in the train of spikes. To do this, they had to measure interspike intervals (ISIs) at millisecond resolution. If the brain controls these intervals, and the muscles respond accordingly (for instance, with changes in air pressure), it would signify the presence of a neural code.

With these techniques they were able to isolate properties of the neuromuscular response for a variety of experimental tests. In particular, they were looking for the effects of different signal patterns. “Therefore, we believe that our muscle stimulation experiments were only activating the axons of motor neurons and were not activating muscle fibers directly,” they say. “This finding allowed us to make insightful comparisons between the results of our spike pattern and stimulation analyses.” After gathering large data sets and crunching them with software, they came to the conclusion they had found a code — not just in songbirds, but all animals:

Overall, we have shown that respiratory motor unit activity is controlled on millisecond timescales, that precise timing of spikes in multispike patterns is correlated with behavior (air sac pressure), and that muscle force output and the behavior itself are causally affected by spike timing (all on similar temporal scales) (Figs. 2D, 3C, and 4C). These findings provide crucial evidence that precise spike timing codes casually [sic, causally] modulate vertebrate behavior. Additionally, they shift the focus from coding by individual spikes (1, 14, 19) to coding by multispike patterns and from using spike timing to represent time during a behavioral sequence (20, 21) to coding its structural features. Put another way, although it is clear that earlier activation of neurons would lead to earlier activation of muscles, this relationship only accounts for encoding when a behavior happens (10, 22). Here, we show that changing the timing of a single spike within a burst by ∼1 ms can also affect what the animal will do, not just when it will do it. Furthermore, we showed that the effect of moving a single spike is stable across animals (Fig. 2). We believe that this precise spike timing code reflects and exploits muscle nonlinearities: spikes less than ∼20 ms apart generate force supralinearly (SI Appendix, Fig. S12), with stronger nonlinearities for shorter ISIs [interspike intervals]. Thus, changing the first ISI from 12 to 10 ms significantly alters the effect of the spike pattern on air pressure (Fig. 2B). Such nonlinearities in force production as a function of spike timing have been observed in a number of species (23⇓-25), highlighting the necessity of examining the role of spike timing codes in the motor systems of other animals. Importantly, our findings show that the nervous system uses millisecond-timescale changes in spike timing to control behavior by exploiting these muscle nonlinearities, even though the muscles develop force on a significantly longer timescale (tens of milliseconds as shown in Fig. 3B).

They speak of the “surprising power of spike timing to predict behavior,” indicating that patterns of spikes coming down the nerves are the determining factor in behavior, not just how fast they come.

Is this really a code? Well, count the number of times they refer to coding directly, beside the suggestion in the title, “Motor control by precisely timed spike patterns.” Result: 29 times. “Information,” a related concept in coding, gets 51 mentions. “Precision” and related terms, important for conveying information, gets 14 mentions.

Take a moment to watch this video of a nightingale singing on YouTube and prepare to say Wow!

How much information does the forebrain have to send to the vocal muscles to achieve that kind of performance? The authors note in their concluding discussion, “Because respiration is critical to vocalization in songbirds, it will be of special interest to record respiratory timing patterns during singing….” Indeed!

Think of the possibilities this discovery opens for further research. A multitude of questions come to mind: how does the brain know what pattern to send to a distant muscle to get it to act in a certain way? Are the codes inherited or learned? How reproducible are the patterns from one animal to another? Can a spike code from one bird sent to the nerves of another bird make it sing the same song? How does a human mind interact with the brain to turn a choice into an action? What translates the thought “I must run” into a spike timing pattern that makes you run? How rich, do you think, is the spike timing code in a performance of Chopin’s Fantaisie-Impromptu? (See the video at the top.)

Being a new discovery, this “spike timing code” will undoubtedly prompt much more research on more animals in more settings.


Article credit: Discovery Institute


Don Johnson – January 2017


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 )

Google+ photo

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

Connecting to %s