Neuroscientists Identify Physiological Link Between Trial And Error And Learning

ScienceDaily (Mar. 26, 2009) Learning through trial and error often requires subjects to establish new physiological links by using information about trial outcome to strengthen correct responses or modify incorrect responses. New findings, which appear in the latest issue of the journal Neuron, establish a physiological measure linking trial outcome and learning.

"Our results open a new door to understanding the important role of trial outcome in the learning process," said Wendy Suzuki, a professor at New York University's Center for Neural Science and a co-author of the study.

The study's other co-authors included researchers from France's National Center for Scientific Research, the Harvard Medical School, and the University of California, Davis. The study was supported by a grant from the National Institutes of Health.

For the past half century, scientists have examined the role of the brain's medial temporal lobe in learning. Previous scholarship has determined that a critical function of the medial temporal lobe is to successfully acquire new information about facts and events ("declarative learning") by making new associations between initially unrelated items ("associative learning").

The researchers on the study published in Neuron sought to understand if there is a link between how the brain functions in associative learning and in processing information about trial outcome. Specifically, they were interested in cell activity in a portion of medial temporal lobe called the hippocampus. Earlier research had found that hippocampal neurons are involved in associative learning, such as matching names with faces.

To investigate this process, the researchers had primates play a computer memory game in which the subjects matched particular object-place combination with motor responses. When they associated the correct object-place association with the correct response, the primates were rewarded with their favorite fruit juice. During these sessions, the researchers recorded the activity of the primates' hippocampal neurons.

The results showed that a surprisingly large proportion of recorded hippocampal cells50 percentdifferentiated between correct and error responses. This finding was striking since previous learning or memory studies in the hippocampus showed lower proportions of active cells in task-related activities. Moreover, their findings showed many of these cells also came to respond more strongly to particular object-place combination as learning improved. This suggests that the cells' ability to make distinctions between correct and incorrect trial outcomes may influence new learning by changing a cell's sensitivity to the stimuli being learned.

Journal reference:

1. Sylvia Wirth, Emin Avsar, Cindy C. Chiu, Varun Sharma, Anne C. Smith, Emery Brown, Wendy A. Suzuki. Trial Outcome and Associative Learning Signals in the Monkey Hippocampus. Neuron, 2009; 61 (6): 930-940 DOI: 10.1016/j.neuron.2009.01.012

http://www.sciencedaily.com/releases/2009/03/090325132155.htm

Sammy, Just out of curiosity how is your putting? I know you know your stuff when it comes to what Geoff teaches and have read his book and you seem to easily comprehend his writing, so how has it all transfered to your actual putting stroke and putts per round?

I would say that on a level of 1-10, 10 being able to incporporate 95-100% of what Geoff teaches into my putting, I would say I am at maybe 4-5 ( I still have lots of room for improvement). So far what has helped me the most is the idea of instincts. Not trying to outthink a putt. Getting my line and letting my brain and insticts pull my putter back as far as it needs to and let it go. When you can trust it, it really works. I have yet to really delve into his line reading and as I have told Geoff, I don't know if I will ever go to a 100% gravity stroke. I 100% accept the ideas behind it, I have seen Geoff putt with it and it works, it just takes some commitment and I am building my putting house one brick at a time.

Learning From Mistakes Only Works After Age 12, Study Suggests

ScienceDaily (Sep. 27, 2008) Eight-year-old children have a radically different learning strategy from twelve-year-olds and adults. Eight-year-olds learn primarily from positive feedback ('Well done!'), whereas negative feedback ('Got it wrong this time') scarcely causes any alarm bells to ring. Twelve-year-olds are better able to process negative feedback, and use it to learn from their mistakes. Adults do the same, but more efficiently.

Brain areas for cognitive control

The switch in learning strategy has been demonstrated in behavioural research, which shows that eight-year-olds respond disproportionately inaccurately to negative feedback. But the switch can also be seen in the brain, as developmental psychologist Dr Eveline Crone and her colleagues from the Leiden Brain and Cognition Lab discovered using fMRI research. The difference can be observed particularly in the areas of the brain responsible for cognitive control. These areas are located in the cerebral cortex.

Opposite case

In children of eight and nine, these areas of the brain react strongly to positive feedback and scarcely respond at all to negative feedback. But in children of 12 and 13, and also in adults, the opposite is the case. Their 'control centres' in the brain are more strongly activated by negative feedback and much less by positive feedback.

Three-way division

Crone and her colleagues used fMRI research to compare the brains of three different age groups: children of eight to nine years, children of eleven to twelve years, and adults aged between 18 and 25 years. This three-way division had never been made before; the comparison is generally made between children and adults.

Unexpected

Crone herself was surprised at the outcome: 'We had expected that the brains of eight-year-olds would function in exactly the same way as the brains of twelve-year-olds, but maybe not quite so well. Children learn the whole time, so this new knowledge can have major consequences for people wanting to teach children: how can you best relay instructions to eight- and twelve-year-olds?'

Ticks and crosses

The researchers gave children of both age groups and adults aged 18 to 25 a computer task while they lay in the MRI scanner. The task required them to discover rules. If they did this correctly, a tick appeared on the screen, otherwise a cross appeared. MRI scans showed which parts of the brain were activated.

Learning in a different way

These surprising results set Crone thinking. 'You start to think less in terms of 'good' and 'not so good'. Children of eight may well be able to learn extremely efficiently, only they do it in a different way.'

Learning from mistakes is complicated

She is able to place her fMRI results within the existing knowledge about child development. 'From the literature, it appears that young children respond better to reward than to punishment.' She can also imagine how this comes about: 'The information that you have not done something well is more complicated than the information that you have done something well. Learning from mistakes is more complex than carrying on in the same way as before. You have to ask yourself what precisely went wrong and how it was possible.'

Is it experience?

Is that difference between eight- and twelve-year-olds the result of experience, or does it have to do with the way the brain develops? As yet, nobody has the answer. 'This kind of brain research has only been possible for the last ten years or so,' says Crone, 'and there are a lot more questions which have to be answered. But it is probably a combination of the brain maturing and experience.'

Brain area for positive feedback

There is also an area of the brain that responds strongly to positive feedback: the basal ganglia, just outside the cerebral cortex. The activity of this area of the brain does not change. It remains active in all age groups: in adults, but also in children, both eight-year-olds and twelve-year-olds.

Journal reference:

1. Anna C. K. van Duijvenvoorde, Kiki Zanolie, Serge A. R. B. Rombouts, Maartje E. J. Raijmakers, and Eveline A. Crone. Evaluating the Negative or Valuing the Positive? Neural Mechanisms Supporting Feedback-Based Learning across Development. The Journal of Neuroscience, 17 September 2008

http://www.sciencedaily.com/releases/2008/09/080925104309.htm

I must say the researchers about children learning dont seem to know much about basic brain development. The process by which the nerves get an insulating coating of myelin, a white matter fatty coating, occurs last in brain development and is the main reason young people before about puberty lack good judgment and reasoning of cause and effect and consequences. This developmental knowledge is in ALL the neuroscience books. There is an obvious difference due to the developmental role played in the brain maturation by the growth hormones that appear during puberty as these hormones complete the myelination. A moral and reasonable human does not really emerge until the developmental stage between onset of puberty and full adulthood, between about 12 or 13 and between 16 and 20. It is called the "juvenile" stage and is no big mystery. Juveniles are NOT treated as adults for this reason. Before that, human children learn more like nice pets, and respond to praise and affection from the parents. After the reasoning comes more online, the learning is BOTH response to praise and figuring out cause and effect independently of the parents. Even so, there is a distinction between withholding priase or reward and administering punishment. Even pet dogs and horses train according to the power of withholding affection or praise upon error.

As to the hippocampus, the relationship between socalled declarative knowledge and associative knowledge, terms made uo by people who are not brain scientists, is really "figuring out what works and why" to the point that the body accepts the correct way as the habit. Neuroscience has long ago tracked the declarative knowledge that dominates at first while the brain sorts out the pattern and then dissipates as the figured out pattern is accepted as a habit of solving the problem. No news here.

Geoff Mangum
PuttingZone.com

Researchers Find New Learning Strategy

ScienceDaily (Aug. 7, 2006) Central to being human is the ability to adapt: We learn from our mistakes. Previous theories of learning have assumed that the size of learning naturally scales with the size of the mistake. But now biomedical engineers at Washington University in St. Louis have shown that people can use alternative strategies: Learning does not necessarily scale proportionally with error.

In so doing, Kurt Thoroughman, Ph.D., assistant professor of biomedical engineering at Washington University, and his graduate student, Michael Fine, have discovered a new learning strategy they call categorical adaptation in which steps of learning are sensitive to the direction of error, but do not scale proportionally with the size of the error. Eventually, their findings could have an impact in the rehabilitation of people with neurological ailments such as strokes by making use of different learning environments.

If you make a movement error in one direction, in makes sense that your next movement would correct toward the opposite direction, in exact proportion to the error. An example would be a pitcher correcting to the right, after missing home plate to the left with a pitch.

"We show that learning does not necessarily scale with error," said Thoroughman. "I think we have uncovered a part of human adaptation that certainly doesn't do that. We are not claiming that all previous theories are false in the behaviors that were captured. It's just that we have for the first time found a part of human adaptation that clearly does not scale with the size of the error."

Thoroughman is interested in how humans learn motor skills incrementally, how information from a single movement can inform the generation of the next movement. He and Fine asked volunteers to make reaching movements while holding the end of a robotic arm. Volunteers were trained for about 40 minutes a day for two days. On each day, subjects were asked to make half-second, 10-centimeter reaching movements, directed away from their bodies.

Subjects learned the baseline task on the first day. On the second day Thoroughman and Fine tricked volunteers by having the robotic arm push the human hand with a perturbing pulse of force in 20 percent of movements. The pulse pushed subjects from their normal trajectory, either to the right or the left, with three different pulse strengths. Thoroughman and Fine observed how the pulse altered that trajectory and how subjects corrected, or adapted, in the very next movement.

"The pulse should induce an error in that movement that scales with the size of the pulse," said Thoroughman, who also has appointments in Neurobiology and Physical Therapy at the Washington University School of Medicine. "And we did see that - big pulse, big error; small pulse, small error. But then we expected, just as previous theories would predict, that the adaptation in the next movement would also scale with the size of the force pulse. But it didn't - the adaptation countered the direction of the pulse but was flat with respect to the size of the pulse."

The results were published in the August 2006 issue of the Journal of Neurophysiology.

Thoroughman said the discovery raises interesting new questions in motor learning and neurophysiology and eventually could have an impact on physical therapy protocols.

"By changing environments in a specific way and by not providing the same environment all of the time, we can change the way that people learn," he said. "We're hopeful that this kind of technology can help in neurological rehabilitation so that stroke patients, for instance, could better relearn movements and reduce recovery time."

Dear sammy,

Thanks for that VERY interesting article on motor learning. However, I must comment that the "discovery" made much ado about by the researchers is not really a discovery in the usual sense of finding a previously unknown continent or learning that the mold on the rye bread cures infection. Strictly speaking, they "found out" for the first time that the brain corrects for error without scaling the size of the correction to the size of the error. What they really discovered is that researchers who thought the corrective scaled to the error were mistaken, plus the empirical fact that the corrective is flat regardless of the size of the error.

Okey dokey. So researchers make bad predictions and have erroneous beliefs about what is true. That's normal, or else science is not needed.

The ADDED knowledge, however, is a bit of a puzzle because for the brain NOT to scale the corrective to the size of the error sounds bad to me. So, applying critical intelligence (i.e., "suspicious doubting and further thinking what MIGHT be the real explanation") to this research, I focus on the "task" the researchers used. In particular, the researchers MADE the error, not the test subjects and their brains. This sort of difference in the "task" being tested can "make ALL the difference" in neuroscience research (just ask fMRI researchers about the care required to define the task that gets tested), but apparently "motor science researchers" haven't quite taken this fact about research design totally to heart yet.

The brain probably does not react equally to an error imposed by outside forces and to an error that arises from some organic internal process or sequence of processes. (Let me erase the word "probably" in this sentence.) There is a learned right way to perform and there is an internal "environment" of wrong ways, some of which might have the potentiality seeming like an inevitability of a habit. Apart from just normal human variability (physiologically and psychologically) and medical dysfunctions of one sort or another, the internal errors arise from the "action" getting off the good track onto another erroneous track.

The "track" of the "task" tested in this particular research was the direction of the good arm movement compared to the direction offline of the push, and the response was for the corrective to head back to the good track, just not with a scaled size but with a flat size. This means the brain knows its way back to good AND knows the relative direction between bad and good for any given bad. How would the brain in this research "task" know the way back to good? Because the robot shoved the subject's arm in a specific direction and the subject's brain "felt" which way it got shoved. The corrective for that is quite easy: move back along the same way you felt the shove move you away from what you were doing.

The brain's internal corrective process is not really tested and revealed by this robot-made-the-error-happen-to-you "task". This leaves the issue "undiscovered": maybe the brain's internal processes to correct in light of error DO scale the corrective for the size of the error after all. We'll see.

The "task" or one like it that needs testing to find out is something like a gymnast walking a balance beam or a kid walking along the rail of a railroad track. The imbalances that make the person wobble and regain balance are internally generated by the form of the steps along the beam. The corrective is not really directed in response to the direction and size of the error, but to the brain processes that CAUSED the bad form in the step that resulted in wobbling -- "don't do that again" is the brain's first reaction, and "don't do WHAT again?" is the second reaction and "let's review the last misstep to make sure we understand what made that step bad" is the third reaction in the learning process. So the brain learns that sending the leg too far off to the side, or that allowing the head to lean, or that too much twist in the hips or shoulders, CAUSES the wobbling when the step is made (ultimately teaching about center-of-mass management). The corrective generated internally is analytical diagnosis of what went wrong and why, combined with a comparison of the misstep as diagnosed and understood as a cause of imbalance with the correct in-balance step to learn why the good step does not cause a wobble, and THEREBY the brain learns the "task" of walking the beam in balance with good steps that don't cause wobbles.

The main point is that if I am correct in characterizing the brain's learning process when encountering internally-generated errors, then this process is not a simple response to the direction and size of an externally-imposed error. Instead, as I suppose, the internal corrective process is predominately "get back to the good step" if you know how, and leave the erroneous step behind as it is not likely to recur. Although OTHER errors will undoubtedly occur, exactly that last misstep may or may not be likely. It takes a lot of errors to "learn" what SIZE or DIRECTION of error to expect as the most likely. The brain probably waits to see what error is happening before it scales anything, and the scaled response is first-in-line going to be the usual scaled response to the usual error, and only in rare cases will the brain bet all its success-or-failure money in getting the specific error that is happening scaled just so when it makes a correction. Instead, the brain will likely fire the usual corrective and hope that gets the specific error covered enough. if not, a second corrective is not far behind. This is the basic way the eyes dart from one location to an intended location -- a little "slop" factor in the mismatch between what is usual and what is specifically required, so at first there might be a little miss of the intended target by the initial main eye movement followed by a quick follow-on eye movement to get the miss taken care of.

Neuroscience (and usually motor science) is well familiar with the difference in the feedback corrective processes versus the feedforward corrective processes. The robot-did-it "task" investigated in this study is likely prompting a "feedback" corrective process. This is a response dependent upon the perturbing signal, not the internal goal. A feedforward corrective process is faster as depends upon the internal signal / representation in the brain of what should be happening, not what is or has been messing it up.

If a waiter is walking by with a level tray poised on his palm and the tray is weighed down with beer mugs and a person lifts a mug off the tray, what happens to the waiter's arm and tray depends upon whether the waiter expects this to happen or not, one mug at a time as he walks thru the crowd. The "good" movement is to keep the tray steady and level. The simple response without caring about the good is for the arm and tray to buoy upwards when the mug is removed. An unawares corrective to the sudden and unexpected removal of a mug of beer from the tray is probably a non-scaled "hold on, no more of this error stuff until we figure out exactly what is happening". That will look like a non-scaled corrective headed back in the direction of the good like the one seen in this recent research, but it is basically just a stopgap. The corrective is different if the waiter expects the mugs to go missing one at a time in a certain pattern as he moves from person to person across the room.

Applying all this to learning what to do about a poor stroke, the golfer first needs to know the difference between a good stroke and a bad stroke or between a good stroke and OTHER strokes of all sorts that are not as good. Golfers don't start here but must first learn what a good stroke is. That takes years, or a good teacher, because a good putting stroke is not a usual movement that arises frequently in normal daily activities. (Sweeping trash with a broom comes close, but the manner of holding the broom and holding the putter handle are not even close, so sweeping doesn't really serve as a normal movement suited to putting. Sweeping with a bucket of water at the end of a rope is probably closer to a putting stroke, but this is clearly not something humans usually do on a daily basis.)

Even if a golfer knows what a good stroke should be and has made many, many good strokes, he still does not necessarily know what sort of bad stroke is likely to happen. That is another problem entirely, but it is basically revisiting the "bad" strokes the golfer made BEFORE he learned what was a good stroke. Then he "learns" in what respect the usual bad is different from the intended good.

The "usual bad stroke" is a pull. The learned good stroke is OTHER than a pull. The goodness of the good stroke comes from making a non-pull stroke look and feel and happen naturally (i.e., habitually). the masterful golfer on the green has this figured out in terms of what promotes the good and what allows or causes the bad, in terms of setup posture, muscle tone, muscle movement recruitment patterns, and the timing of the forces of the body and putter in motion with balance and comfort for consistent accuracy that generates the good stroke.

In this sense, a "push" stroke that compares poorly to a good stroke is actually a learned stroke, and not at all a natural and habitual movement. A "push" in putting is the result of poor learning trying to get away from the habitual pull.

The inapplicability of the recent motor science research to golf putting is mostly seen upon the realization that in putting there NEVER is an external disturbing force that alters the stroke away from good in the middle of making the stroke. Never. No one walks up to the golfer while putting and shoves the stroke out of pattern. A gust of wind perhaps, but nothing else in the world ever messes with the golfer. The golfer knows the deal and is solely responsible for any error that happens in the stroke, either from setup, or muscle activation, or focus on making the good movement pattern as previously learned. The focus is never on correcting a specific error, but on understanding why the good movement did not happen, so that the NEXT stroke is good. You can either just keep trying to repeat good strokes, or you can learn from the errors MORE ABOUT WHAT IS REQUIRED TO MAKE GOOD STROKES MORE OFTEN. I personally say, "get busy improving not grooving, or get out of the way."

Cheers!

Geoff Mangum
Putting Coach and Theorist

Offering Free Podcast Tips for Putting Every Friday on GolfSmarterTips.com.

The best putting instruction book in golf history is now available for purchase in hardback or as an immediate ebook download: Optimal Putting: Brain Science, Instincts, and the Four Skills of Putting (2008, 282-pages)

Geoff Mangum's
PuttingZone
PuttingZone Clinics
Flatstick Forum
PuttingZone Channel on YouTube
PuttingZone Picasweb Image Gallery


Golf's most advanced and comprehensive putting instruction -- you're either in the PuttingZone, or not.

Over 2.5 million visits -- 200,000 monthly from 50+ countries -- and growing strong.