Stanford scientists see the logic in the illogical behavior of neurons

Nov 08, 2010 By Louis Bergeron
Electrical engineering postdoctoral student Mark Churchland, left, and associate professor of electrical engineering Krishna Shenoy.

(PhysOrg.com) -- Neurons in your brain trigger the physical movements of your body, but some of them seem to fire in a crazy quilt pattern just before and during the movement. But Stanford researchers say there is method in the apparent madness.

You've decided to kick a soccer ball. But before your muscles even twitch, your brain has to kick into gear to direct the action.

Just what happens inside your brain during that kick-in process, though, has puzzled researchers because some of your neurons – which generate the electrical signals to trigger your muscles – sometimes work to create the opposite motion.

Now Stanford researchers have found out why, and it turns out the contrarian neurons aren't contrary at all – they just have a different way of getting to their goal.

"A classic idea is that the neurons are coded according to a sort of blueprint, in which each neuron has a movement that it 'prefers,'" said Mark Churchland, a postdoctoral researcher in electrical engineering. That means that a given neuron would be most active before and during its preferred movement. For example, if you wanted your leg to make a rightward movement, all your neurons would be active, but the rightward-preferring would be the most active.

"But what we found is that a neuron could be very active before, say, a rightward movement, but then actually shut down just before the rightward movement," Churchland said. Or it could be completely inactive before a leftward movement, but then become strongly active during the leftward movement.

"If you're trying to relate the activity of a single neuron to the action that takes place, it looks crazy," he said.

Method to the 'madness'

"If you said that the neuron was effectively voting for its preferred movement, you'd say it is voting for moving left at this time and a tenth of a second later it is voting for moving right and a tenth of a second after that it is voting for something else," Churchland said. "It would not make any sense at all."

But if you compare the neuron's behavior to a pendulum in a clock, things begin to make sense, he said. In order to get a pendulum to swing to the right, you first have to pull it to the left. And as a pendulum swings back and forth, it will be moving in different directions at different times, even as all its activity is directed at keeping the proper time.

"Whereas a vote is something that should stay nice and consistent across time, a pendulum may swing different directions at different times. But a pendulum has dynamics that are consistent across time even though the position of the pendulum is not," Churchland said.

"It basically comes down to don't think of planning a movement as something that involves creating an explicit blueprint," Churchland said. "Think of it as getting your motor system wound up in just the right way so that when you release it, it does just the right thing."

Churchland is the lead author on a paper explaining the research in the Nov. 4 issue of Neuron. Krishna Shenoy, an associate professor of electrical engineering, is the senior author.

Shenoy and Churchland conducted their research with rhesus macaque monkeys. The monkeys were connected to a computer that monitored their neural activity.

Video game test

While the monkeys were monitored, they played a video game that required them to move their hand from an initial point on the screen to a small square that appeared elsewhere on the screen. "The square initially jiggles onscreen, sort of like a buzzing fly. Then it stops jiggling, which we call landing, and then the monkey's job is to swat the fly," Churchland said.

"The important thing is that the tasks allowed us to record neural activity not just during the movement, but also during the period when the monkeys are getting ready to make the movement," Churchland said.

So if there is no blueprint for your brain to follow each time you decide to move a muscle, does the brain have to figure everything out from scratch, starting at the beginning each time?

"That is an excellent question," Churchland said. "We do know that the has difficulty winding up its 'neural pendulum' exactly the same way every time. That is probably part of what makes golf so challenging. However, we really don't have a full answer to that question yet. It is very much an avenue for future research."

Explore further: Abuse in childhood tied to migraines in adulthood

Related Stories

Neurons cast votes to guide decision-making

Oct 08, 2010

(PhysOrg.com) -- We know that casting a ballot in the voting booth involves politics, values and personalities. But before you ever push the button for your candidate, your brain has already carried out an ...

Monkey's Thoughts Make Robot Walk from Across the Globe

Jan 17, 2008

In a first-of-its-kind experiment, the brain activity of a monkey has been used to control the real-time walking patterns of a robot halfway around the world, according to researchers at Duke University Medical Center.

Researchers identify new neurological deficit behind lazy eye

Sep 10, 2010

Researchers at New York University's Center for Neural Science have identified a new neurological deficit behind amblyopia, or "lazy eye." Their findings, which appear in the most recent issue of the Journal of Neuroscience, shed a ...

Recommended for you

Smartphone thumb skills alter our brains

Dec 23, 2014

When people spend time interacting with their smartphones via touchscreen, it actually changes the way their thumbs and brains work together, according to a report in the Cell Press journal Current Biology on Dec ...

User comments : 11

Adjust slider to filter visible comments by rank

Display comments: newest first

javier_marasco
not rated yet Nov 08, 2010
Nice article, comes to my mind a 'crazy' idea. Isn't possible that some neurons are voting for the right movement and others are voting for the opposite movement just as a way to equilibrate the resulting movement? thinking about AI you need to pass a sigmoid function to change a neuron, maybe a way to get a good tuning of that sigmoid function is this way in which the neurons "equilibrate" the resulting "vote". It's just a guessing but would like to know what other people thinks on this idea.
DamienS
not rated yet Nov 08, 2010
Not sure if this is related, but often adding noise to a system can improve its acuity/performance, which is similarly counterintuitive.
kevinrtrs
1 / 5 (2) Nov 09, 2010
So if there is no blueprint for your brain to follow each time you decide to move a muscle, does the brain have to figure everything out from scratch, starting at the beginning each time?

Intuition says NO, since one develops habits and in the habit one has stored a blueprint of how to perform a certain action. The same kind of learning behaviour can be experienced by using the normally non-dominant hand/side to perform the action. Thru repetition a neural pathway is established that guides the firing of the neurons.
kevinrtrs
1.2 / 5 (6) Nov 09, 2010
Mmmhhhh... this article raises the question of just HOW does the brain "know" that there's an arm or leg or other limb to be moved in the first place? does it all come hardwired during the pregnancy. If so then that means that there's information stored in the DNA that tells the brain where and how to access the muscles. Which then begs the question, where does that information come from? Random physical processes does NOT produce such structured information. Of course I'm referring to the random mutations required by evolution to bring forth a human being from that single ancestor.
WHERE does the information come from?
DamienS
5 / 5 (4) Nov 09, 2010
HOW does the brain "know" that there's an arm or leg or other limb to be moved in the first place? ... WHERE does the information come from?

Sensory feedback.
Skeptic_Heretic
5 / 5 (3) Nov 09, 2010
If so then that means that there's information stored in the DNA that tells the brain where and how to access the muscles.
If that were the case, you'd be born able to run.
Random physical processes does NOT produce such structured information.
Good thing physical processes aren't random. They follow rules, or laws, some of which we know, others we're attempting to discover.
Of course I'm referring to the random mutations required by evolution to bring forth a human being from that single ancestor.
WHERE does the information come from?
Non-random survival of randomly generated organisms.

I think the most notable pieces of your posting is that you showed that you're not that stupid in your first post, then had to revisit and post your tripe. Says only one thing to the rest of us Kevin, you're trolling.
javier_marasco
not rated yet Nov 09, 2010
Mmmhhhh... this article raises the question of just HOW does the brain "know" that there's an arm or leg or other limb to be moved in the first place? does it all come hardwired...


Actually your brain "learns" about your entire body while you are a baby. That's how your brain know about your limbs, simply trying to use them and failing. Failing is what makes you progress and learn (in this topic we are discussing).

Nice article, comes to my mind a 'crazy' idea. Isn't possible that some neurons are voting for the right movement and others are voting for the opposite movement just as a way to equilibrate the resulting movement? thinking about AI you need to pass a sigmoid function to change a neuron, maybe a way to get a good tuning of that sigmoid function is this way in which the neurons "equilibrate" the resulting "vote". It's just a guessing but would like to know what other people thinks on this idea.


Does my idea makes any sense? anyone?
CHAN
not rated yet Nov 10, 2010
May be the brain behaves like sectors of memory stacks connected to a central processor with simple reflective ends. Signals from conceptions passes to the memory stacks and interferes with the existing data in the stacks and seeks for equilibrium. If the feedback is nutral, the signal will be kept for a while and disappeared with the strength of signal - short-term memory; if the feedback is positive, it becomes permenant memory. if it is negative and strong enough, it will eliminate the old data and replaced with updated data, if the negative signal is not strong enough to eliminate the old data, it will only partial reduce the strength of old data, i.e. you are in doubt. If the stacks are empty, the process is absolute learning. As the stacks grow, they begin to reject/reimpress/doubt.
No prove for the above said.
CHAN
not rated yet Nov 10, 2010
the other parts of brain linking to motors of our body recognize the combinations of reflections from the stacks and transform the signals into motoring signals for minimum energy consuming actions in order to nutralize the remaining signals.

That is conscious route for signal travelling when we are awake - close circuit.
There is an unconscious route of signal travelling when we are asleep - open circuit. There are two modes:
1- Energy Saving Mode (Exhausted)
2- Signal Free Mode (Have nothing to worry about)
perhaps there is 3rd mode - the hybrid of the two.
Mode 1 can not be reactivated by stimulation of perceptions.
Mode 2 can be reactivated by certain kinds of perceptions.
Mode 3 depends.
Natural Die-ing of Brain:-
Signals from perceptions are too weak to travel through the stacks. If the stacks are first in last reached, the fresh memories will be the easiest to be forgot.
so are the basic behaviors.
Aging and the selfrepair capabilities of body are not related.
PJK
not rated yet Nov 14, 2010
If so then that means that there's information stored in the DNA that tells the brain where and how to access the muscles. Which then begs the question, where does that information come from? Random physical processes does NOT produce such structured information. Of course I'm referring to the random mutations required by evolution to bring forth a human being from that single ancestor.
WHERE does the information come from?


If it were "hardwired", the information would come from the same place as information about eye colour, no. of fingers and so forth. Your parents DNA.

You seem to be under the impression that evolution by natural selection requires that everyone's DNA is randomly generated. If that were the case, then your objection concerning the origin of information would of course be valid.

But that is definitely not the case.
marjon
not rated yet Nov 14, 2010
More headline hubris from physorg.com.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.