9 posts / 0 new
Last post
Gerodea
Gerodea's picture
Stimulus-Response Compatibility

Stimulus-Response (S-R) Compatibility is usually defined as the ‘naturalness of the connection between the stimulus and the associated response'.   Reaction Time for skill performance is faster the more compatible the S-R pairs.

For example, the S-R Compatibility for turning a car steering wheel to the left in response to an upcoming left turn stimulus is very close.    This is not the case for a person placed for the first time in a sailing boat who is asked to turn the boat to the left - they generally will have a short “incompatibility delay” while they work out that the tiller needs to be moved to the right in order to turn the boat to the left.

Practice and High-Level Performers

 A highly practiced performer can overcome many things, including the disadvantage of many S-R choices and low S-R Compatibility.   For example the skilled sailboat racer almost instantly moves the tiller to the left as soon as it is obvious he needs to turn the boat to the right.

Research has shown that both the amount of practice and the nature of practice can effect Choice Reaction Time.   With extreme amounts of practice, high-level performers can produce reactions that approach automatic processing; not only are these reactions very fast, but they are slowed down little, if at all, as the number of S-R choices increases.

That’s the good news:  S-R Non-Compatibility can be overtrained and therefore compensated for.  The caveat, of course, is that long-term training is required for this overtraining to occur.  Short-term practice, such as is needed in occupational training  contexts rather than lifestyle training contexts, may not provide the over-training needed to compensate for S-R compatibility bias.

Natural Defensive Gestures and S-R Compatibility

Consider how a person naturally responds to a sudden stimulus, such as a splash of water aimed at their face.....    Now, consider (if you have experience of it) how some Self-Defence training attempts to teach how a practitioner “should” block a punch.   Often the “natural” response to the stimulus is seen as somehow less-optimal than a neater, more technically alluring or complex movement, however what the sport-science is telling us is that if the “blocking move” does not closely match the naturally-connected response, then the Reaction Time will be longer.

Needless to say, in the case of a punch to the face, a longer Reaction Time carries serious risk.   What then is the most naturally-associated response to a punch in the face?

Cell Press, in an article in 2004 noted new research which was being published on the machinery of that automatic, natural response called the startle reflex; “...a critically important protective mechanism by which animals and humans instantly protect themselves against threats ranging from an attacking predator to an incoming golf ball.”

For some years, a scientist called Graziano has been studying in a number of experiments how certain movements, consistent with defending the head or body from impending threats are automatically evoked.  These movements include:  “a squint and facial grimace that was more pronounced on the side of the sensory receptive field, a turning of the head away from the side of the sensory receptive field, a rapid movement of the hand to an upper lateral location as if blocking an object in the sensory receptive field, and a turning outward of the palm”.

He went on to note that “These movements had a machine-like repeatability over hundreds of trials” suggesting that they are highly reliable and do not atrophy over time.   He also asked whether there might be a stored set of postures in the brain which controls limb-movements to evoke these inherited defensive gestures.

Pahlavian, researching similar fear effects and publishing his findings in 2000, found similarly:  “...stimuli such as pain or fear automatically elicit patterns of terminal motor states corresponding to fight or flight, initiating processes of preparation of spatially oriented movements which are automatic”

Critically, these movements are extremely fast: "For a protective movement in which the hand moved to an upper lateral position and turned outward as if to block an impending threat to the head, the velocity was remarkably fast, appropriate for a defensive gesture (230 cm/s at peak speed...)" [from Graziano]

So, it would seem, that incorporating natural defensive gestures into our Breakaway and Self-Defence training programmes may indeed be a way to overcome some of the issues with improving Technique Recall, Performance Effectiveness and reducing Choice Reaction Time.

miket
miket's picture

Good post Gerodea, this is a subject that I have been thinking a lot about recently...

I think the key to your observations is centered on the length of training time. I agree that S/R (which I personally prefer to think of as S/R/R) is possible to develop to the levels that you identify.  I once had a JKD instructor illustrate this to me by tossing his keys to me as part of a discussion we were having on the subject-- I caught them 'almost without thinking'.  His reply was to say that this is the objective in our martial arts training-- to attempt to make reaction reach near-autonomous levels.

For each technique.smiley

It's a great illustration but a daunting objective.  Personally, I have come to refer to this as the 'at will' deployment of mechanics--  As soon as your mind demands it, your body can do it.

If you read Miller's unpacking of Boyd's OODA concept (See Facing Violence), he articulates, as have others, that basically the 'middle' 'OD' of the OODA process can be shortened with a process operant conditioning.  So, a 'response' is paired to a 'stimulus' and repeatedly drilled, then rewarded.  That's pretty typical in most martial arts.

The 'problem' with many organized martial arts, however, is that they attampt to pair multiple stimuli with multiple responses, such that the requisite training time you mention becomes virtually impossible to 'set' the skill at that requisite (near-autonomous) level, let alone maintain that skill.  If I know fifty ways to block a straight punch (or even 10!), I have significantly increased the amount of stimulus-response pairings I am required to 'know' and maintain at a near-autonomoous level.  That's quite a challnege.

This very quickly brings us to a situation where Hick's law starts to bear on perfornance, although personally I believe that Hick's is somewhat of a simplification due to the human ability to 'chunk' like-kind things together, such that the elemental builidng blocks of 'a', 'b' and 'c' become simply 'one' unit we might describe with a capital 'A'.  But essentially the science confirms what intuition tells us:  the 'more' stuff we 'know' how to do in response to attack X, at a minimum, the 'more' material we have to sift through in application; the 'more' we have to train to that point of lifestyle ownership you describe; the 'more' material we have to maintain from degrading; and the 'more' training time we must necessarily devote to all of the above.

Further, to the degree that we have **NOT** adequately eliminated the middle steps of the OODA process with veritable **tons** of OC training, at least theoretically, the 'more' reaction time we will loose 'orienting' to a perceived stimulus and 'deciding' what to do about it.

Chunking, I believe, only becomes truly possible when the skill set is a) kept small or b) closely related, or c (preferably) both.  So for instance, I teach eight 'structured parries' for students in defending straight punches-- literally 'eight methods' of punch deflection.  BUT-- I also very much use drills that FORCE students outside of 'structured' responses and into a zone where the eight methods are focused on 'fusing' to become ONE method:  to dynamically parry.

Effectively, I could skip the eight BEGINNING steps, and just throw them to the wolves and throw straight punches at them while telling them to "block".  The purpose of the structured (organized) methods, however, is simply to IMPROVE the underlying performance of the 'one' UNSTRCTURED method by lending it something we might call PHYSICAL structure and efficiency.  Where I believe many martial arts fall apart  is with the assumption that the 'eight methods' will manifest purely in 'real' combat-- or worse, with the assumption that it will be possible to OPERATE at a level where the ability to discriminate at the level of eight POSSIBILITIES for dealing with the same thing even exists.  But chunking IS possible, I believe, and to prove it one only needs to look at boxing, where even a mediocre boxer can punch six to eight times per second.  At the reaction times dictated by Hicks, this would be a physical impossibility as Hock Hochheim has noted.  So, 'chunking' in skill acquisition is both possible and important.  But, it is equally important to note that even chunking is  UN-likely to solve the self-created problem of most martial arts which I would here assert is:  attempting to match **too many** responses with too many stimuli.

The very idea of S/R/**R** is typically placed on the resonse.   Peronsally, I tell students that this is by definition a "third choice" (of four possibilities) for encounter management.  i.e. The idea that we a) recognize an attack, and b) CHOOSE a response-- what we might call 'riposting' (blocking and countering)--  becomes desirable as a choice only AFTER the INITIATVE control options of  1) preemption of the threat and 2) simultaneous attack have ALREADY been eliminated.  Such a 'response' (riposting) further precedes only one option, what I call 'recovery initiaitve' which is those situations concerned primarily with the ambush where we are effecively 'in' a fight before we are aware of the threat and need to 'recover' the initiative by 'making' action, not reacting to the threat.  i.e., it becomes imperative in recovery situations that we a) stop or minimize damage and 2) give HIM somthing to 'react' to by going on the attack.

Another problem (largely training induced, I would argue), is the assumption that combat precedes in a sort of "you go, I go" fashion of alternating turns.  When in fact, the threat in an ambush may land his first THREE (or more) strikes as a unit of 'one'-- literally bam bam bam of a sudden-- which we are then left simply trying to 'orient' to, let alone 'decide' what to do about it (which, as Miller notes, is one aspect behind the tendencey to freeze in such situations).  Meaning, we have not just ONE stimulus to 'orient' to and 'decide' what to do about, we have instead recieved a 'three for the price of one' deal.  Clearly, in such a situation, the opportunity to 'react' is gone before it even really manifests.  He is on to his second (third, fourth, fifth) action while we are still playing mental catch up.

And, in a 'class four' situation, clearly, with the sheer amount of variables and options avaialable to the threat in an ambush (Or when we have allowed ourselves to be overcome by a barrage or blitz attack), it becomes obvious that the idea of 'reactive' pairing of stimulus with response goes right out the window-- at least until a point that we have oriented ourselves to the fact that we are BEING attacked and identified the relative position of the threat, assuming we get that  far.  One of the quotes Rory put in his book that I liked was (paraphrasing from memory), "to the degree that you can turn an 'ambush' into a 'fight', you are doing well."

Meanwhile, as he notes, we are "taking damage" the entire time, which introduces  the concept of pain, thereby further disturbing our abilty to orient.

So, as you begin to intimate above, this is the point where the genius of guys like Tony Blauer advocating "behaviorally based" self-protection concepts enter the discussion.  If, instread of chunking not only **MY** moves, I instead focus in training on categorically chunking HIS attack options to certain primary angles and gross catagories (i.e. "aggressive contact from the front high line")  **AND** on delimiting my so-called 'response' options to those primary attacks as much as possible, I am theoretically, money ahead when faced with the spontaneity and chaos of a 'real' attack. (Incidentally, this is also the approach Rory advocates in his book).  This MAY give me the mental 'space' I need to 'find' the 'recognition point' of something RESEMBLING one of my structured response movements.  Personally, its my belief that structured movements don't manifest in real fighting, not really.  Rather, they both  'underlie" and 'inform' the effectiveness of whatever we attempt to do.

So, if structured training responses are so non-exisitent in real fighting, then why do we do it?  Again, my personal belief is that it is well-formed structured training which is precisely what allows us **TO** identify that split second opportunity RESEMBLING something we already 'know how to do'.  In other words, your 'parry counter punch' combo doesn't manifest, but because you have trained it ad nauseum, assuming you have turned the assault into a fight, you MAY be able to recognize that his attempted headlock (which you have half-counterd in a sloppy bastardized more-panicky way NOT focused on a PARTICULAR structured response, but merely on the idea of 'stopping him from choking me', now!) leaves your ams in a positon RESEMBLING something you have trained to do, and it clicks.  So, you can (at that point) 'connect to' your training and 'bring it out' in application.  

When I was first learning to grapple, I once choked a guy out using a triangle-choke-that-I-did-not-know-was-a-triangle choke. smiley   I was in some top position and the guy swept me such that my shin ended across his shoulders.  No one had ever SHOWN me a triangle choke at that point.  But I had studied the bejeebers out of 'figure-four leverage' using arms, sticks, and a combination of arms and sticks from standing  ** A LOT**.  So, presented with the brief connection, I recognized the 'opportunity' for something that I didn't even know at that point 'existed' which gave me what I would call 'something [spontaneous] to try'.  (emphasis that this was a simple free-roll, not a fight).

I have since gone on to learn three or four variations of this 'move', and I'll be honest--  I have a hard time hitting any of them on someone of my own level of ability unless I go looking for one in particular and specifically work to set it up.  I have certainly NEVER hit one with the split second clarity of mental recognition that I had when I 'accidentally' fell into it.  There again, I think this is a function of  training---  i.e. it is a function of NOW having multiple psoition-based pairings of stimulus and response with inadequate mainenance, whereas previously I had simply identified a concept that I had chunked and therefore understood at a more fundamental level.   And, in fairness, I hit triangles on people who DON'T know them, i.e. on lesser-experienced opponents with some regularity.  But against other grapplers who have a similarly developed level of sensitivity?  Its a lot harder.

A lot of guys I know seem to assume that the average thug will not have their level of experience or trained sensitivity.

I'm not saying that I belive your average 'thug'will have the technical repetoire of a good martial artist.  However, I also believe that a 'good' thug may be a LOT more adept at using, and familiar with so-called 'real' violence than the average martial artist.  A 'good' thug may have a much smaller arsenal, and be a LOT more practiced at using it in 'live' spontaneous situations than you are.  They may be a LOT more "innocuated" to stress than you are-- unless you train for those elements.

If you look at those four classes of initiative I mention above,  the interesting thing I point out to students is that the first three assume a THREAT AWARE encounter context.  In other words, you cannot physically preempt, crash, or ripost a threat unless you first know its there and have oriented yourself to it.  And yet, the statistical reality of violent assault is that many are frequently UNDER assault before they recognize it is even happening.

Point:  training frequently assumes initiave classes 1-3 are the fight context.  In reality, class four is (I would assert) the more common scenario, which means-- all of our 'threat aware' S/R pairings may be so much junk, unless we have trained AT LENGTH in how to 'find' and employ them under the less than perfect criteria that exist in class four situations.

Another training maxim I like, which unfortunately I cannot attribute to anyone in particular, being that I just have it scrawled without attribution on a pad, is "When things go right, anything will work". smiley

Training MUST balance out structured responses to controlled sceanrios with un-structured responses to open ended (Iain:  'all-in') objective-oriented scenarios.

Nice article.

JWT
JWT's picture

Ger, Mike - good stuff. :)

miket
miket's picture

Thinking about this more today...

All theory aside, it occurred to me that physical skill 'training' as a concept could be succinctly summarized as an incremental process focused on, and culminating in, what we might call the "actual doing" of a specific target activity.

A second flash of clarity was---  if we think about the 'doing' of something, its kind of hard to 'do' without imagining a place and context of the doing.  You don't "do" kayaking in your tub, you do it in a lake, the river or an ocean.  You also learn it there, mostly BY doing it in gradually increasing--i.e. incrementally more challenging-- tasks.

As such, if our objective is 'incrementally proceeding toward actual doing of something', two final points become possible to articulate:  1) There is no way we can 'know' if we have arrived at the ability TO 'do' unless we ATTEMPT that doing (or the closest facsimile we can safely achive).  We will 'know' if we 'can' (or cannot) 'do' based upon our success or failure in the task environment being tested for.  And

2) its possible to see that training activities that do not DIRECTLY support this incremental process of doing can be, at least at some levels, considered to be impediments that exist between us and that incremental doing of the target activity.  That doesn't mean that they don't have VALUE, just that their value needs to be understood within the  context of supporting whatever ELSE these activities support.  Or, it rests with the acknowledgement that they support the devlopment of useful supporting attributes (i.e. the strength developed by weight lifting SUPPORTS the application of leverage in combat, but has no direct connection to 'technical' combative development).

So, in re-reading this (meaning the above few sentences, and not the entire thread), this all seems maybe rather mundane and obvious...  Except that I think that many martial artists (self-included), tend to get wrapped up in a resource based approach to combatives (working with  what we have been given/ learned), instead of starting with the 'what am I trying to do' question and working backward.

So sorry if this addition is somewhat OT, but it seemed kind of related to me...  Relating it to S/ R, I guess what I mean is:  you'll 'know' if your S/R training is 'working' or not if you find yourself gaining facility with the target activity.    A great part of the 'debate' that exists between martial artists-- this method or that method [of training], this system or that system [i.e. forms of training]-- can be vastly simplified via this simple question: what am I trying to accomplish?

Martial arts aren't 'wrong' in their pursuit of doing.  It's just, unless they are combat sports, that they generally 'stop short' of 'deploying' the doing in actual contexts, so the 'doing' never gets tested.  So I learn to "knife block" (or whatever) really really well.  But I don't ever learn to block an assault simulation by finding/using/deploying my knife block in less than perfect circumstances-- i.e. by using the acquired tool to some purpose.

This perceived shortcoming is where I think the RBSD and behaviorally focused guys really get it right. 

JWT
JWT's picture

HI Mike

Good points.  This reminds me of a thread I think we had on the old forum where we discussed whether to teach Kata or bunkai first.  I can remember making the analogy to swimming and the need to experience water before breaking down technique and targetting weaknesses with dry training.

As McArdle Katch & Katch said long ago, "practise doesn't make perfect - perfect practice makes perfect".  I am a strong believer in doing more slow training with proper distance and target penetration to create good form, interspersed with high pressure fast training, than the 'norm' of setting classes at mainly medium speed.  I can remember writing a short piece about speed in training for Jissen 7 (republished here: http://www.practicalkarate.co.uk/JissenSpeed.html ) going through the pros and cons of different training paces.  This form of slow training allows safe hitting with good form against people rather than solely against pads, since when you only practise hitting against pads you are not getting the best psychological or physical preparation for making contact in either a combat sport or in self defence.

Over the last 12 months I've been running simulation days in PPE for martial artists outside my club.  As you say, using the tools in a context really helps people put things into perspective.  This little training clip (excuse some of the bad acting) shows a good Karate instructor preempting an attack a little too early, and a BJJ student being given a gentle reminder by a JuJitsu Instructor (thankfully) why going to the ground in a multiple person fight isn't a good idea:

Some lessons are learned best in context.  

miket
miket's picture

Nice, Jon! 

Question-- As a gross percentage of training time, how much do you dedicate in your group between the following three areas:

1.  'Structured formal learning' [of skills TO deploy in context]-- i.e. mechanic formation training 'HOW TO do a ___ ' and its reinforcement.

2.  Semi-choreographed drills (light contact semi-sparring drills, 'attacker option' (i.e. their choice) one steps, open ended play, etc., )  i.e. -- all the stuff you do that is not 'fully arranged' but still 'not-yet-all-in-scenario testing'.

3.  All-in Scenario testing with gear

Personally, while I have found group composition tends to impact this, I would say we are about 20%-60%-20%; maybe 25-50-25 or 30-50-20.  I definitely tend to privelege open ended drills that are not all-in scenarios like your illustration, then use the latter as kind of a periodic exam... kinda like you'd learn math.

miket
miket's picture

PS:  To try and keep this on topic, I added a split for a couple other random ideas I've been having about S/R relationships in training here...

http://iainabernethy.co.uk/content/change-state-and-mixing-martial-arts-0

JWT
JWT's picture

Hi Mike

That's a tricky question.  I would have trouble distinguishing between the 3 categories the way our training is structured.  

We do (1) through structured paired drills and paired padwork that mimics the drills.  In these drills contact is made - either light through slow motion hitting through the training partner, or medium - high contact with both wearing armour, or full contact with the pad in place of a head (held as a guard by the attacker). 

We hardly ever do (2) per se because (1) accounts for (2) in that the drills are devised so that multiple outcomes can flow from positions depending on angles achieved etc and the combative aims.  Alternative endings for basic drills are trained as part of the syllabus.

(3) Is quite rare, mainly because class locations mean that the verbal element is restricted.  We do the physical element regularly as an extension of 1/2.

So in conclusion I would describe our training as 90% (1): 10% (3) or 90%(2):10%(3) depending on perspective.

miket
miket's picture

Got it, thanks.  I like the statement "devised so that multiple outcomes can flow from positions depending..."