Michael Quoc Founder & CEO · Axioms.ai
Vector 1 of 7: The threat response under strikes

Axiomatic Intelligence · An investigation

The Physics of Striking

When a strike comes at your head the body fires a brainstem defence before any fear exists and sets its strength by how close the strike is predicted to land and what covers the face; over the next seconds adrenaline and heart rate climb far more than the energy being spent, because the heart is tuned for anticipated work and not the work being done; the tenfold figure's only located provenance is an activity code for the work of fighting against seated rest, with fear's share of that excess a small fraction and no panic-specific source found anywhere; thinking is not destroyed but made dearer and pushed toward speed, action and the threat in front of you, with the hands largely intact and the judgment of what an approach is the part fear damages most; training does not switch the alarm off but keeps the hands, the read and the exits working while it runs, a protection that held for months with anxiety still high, and what it builds is a stimulus-bound safety memory that time, a new room or one bad exchange can undo; a defense is a margin on a clock set by pre-strike body cues, a near-fixed response floor, decision time and range, so no one defends the punch they see and the skilled defender's edge is reading the trunk before the fist moves, not faster reflexes; a strike's force is a property of what it hit, only contact speed and impulse travel between targets, a head on a neck can receive only a few kilograms' worth of push from any punch, and a kick delivers more impulse than a punch on one pad with no art's kick measured harder than another's; forward pressure and precise placement are two operators on the defender's exits rather than opposite styles, output is what a fighter spends while accuracy, defensive success and position are what he buys, errors under pressure arrive by the clock not by panic, the ruleset ranks aggression last, and Rodtang and Tawanchai have never fought each other; head impacts in sparring are the high-count low-magnitude arm of a right-skewed dose whose counts depend on the sensor, one ordinary session changes the motor cortex and memory within the hour and one bout raises injury markers past a two-week rest with no symptoms, no single g-figure separates a concussing blow from a harmless one, over years the thalamus and corpus callosum shrink and processing slows in the population record, stopping is followed by partial recovery, and headgear lowers a peak in the laboratory while establishing no safe dose; and a skill practiced less decays per component, the movement pattern persisting for years while repeat-effort capacity drains in weeks, accuracy and reading over months, tissue tolerance returns after force production, relearning is savings not restored precision, and reading an opponent is kept only by a live opponent.

The question this page asks: what happens in a body under incoming strikes, and how is composure inside that chaos actually trained?

“When someone is throwing strikes at your head, your body wants to panic. Panic is expensive - it burns energy at ten times the normal rate and destroys your ability to think. The entire discipline is learning to override that response.”

“Rodtang - relentless forward pressure. Tawanchai - geometric precision. The lesson is finding composure inside chaos.”

This page rests on seven research vectors and 56 axioms, forged on October 4, 2026 under the current version of the method. Every axiom lists the mechanism, the underlying physics, and the primary sources it rests on. Nothing on this page is taken on faith; it can be read one piece at a time.

What happens in a body and a mind in the seconds strikes come at the head, what does that response cost in energy and thinking, and against what baseline could a tenfold figure hold?

The threat response under strikes · eight axioms

The mechanism under test is the acute threat response under incoming strikes: startle, arousal, perception and thinking, and its energy cost against a stated baseline. The easy answer this vector was told not to settle for: “Fight or flight floods you with adrenaline and you cannot think”
1.1 Defensive reflex gain is a controller set by predicted proximity, threat probability and cover, not an alarm that fires at a thresholdVERIFIEDA cortical estimate of predicted proximity, threat probability and cover sets brainstem defensive reflex gain, so a screen or a learnt guard turns the reflex down.Open this axiomClose this axiom
The axiom in full

In healthy adults the brainstem hand-blink reflex is enhanced when the hand enters the face's defensive peripersonal space (S2); the enhancement follows a step function of proximity over a half-ellipsoid on the face (S3, model); it appears only when a stimulus is expected on the near hand and is suppressed when a thin wooden screen stands between hand and face, a screen that changes no afferent input (S1); in 13 boxers against 13 naive adults the near-face enhancement is suppressed in a static guard-like position, more with more years of boxing, and returns when the hand moves (S4). Population: healthy adults and boxers; outcome: reflex magnitude by EMG; class: intervention (within-person manipulation) with one cross-sectional expert contrast. Boundary: a proxy reflex to electrical stimulation, not a measured response to a punch; the boxer effect is a dose correlation, not a training assignment.

The physics underneath

A subcortical defensive circuit is gated by a cortical estimate of where harm can arrive and what covers it; the estimate is written by expected stimulus probability, hand-face geometry and the presence of a barrier, and a learnt protective posture becomes one of its inputs. Because this layer fires before any appraisal exists, subjective panic cannot cause the first defensive action; it can only shape the ones that follow.

The sources
This is the first axiom on the page.Next axiom: 1.2, Threat raises cardiac drive far more than oxygen uptake →Link to this axiom: #ax-1-1
1.2 Threat raises cardiac drive far more than oxygen uptakeVERIFIEDMental stress at rest adds about 20 to 30 percent to energy expenditure while cardiac output exceeds what that oxygen uptake predicts.Open this axiomClose this axiom
The axiom in full

Mental stress raised VO2 by 20 percent of the resting rate in 11 lean women under a clamp, and propranolol abolished the rise in 6 restudied women (S5, intervention, 2000); energy expenditure was 1.3 times rest during mental arithmetic in 44 male students (S6, 2007); cardiac output during stress substantially exceeded the value predicted from each person's own exercise calibration at the same oxygen uptake, in young men measured by Doppler and mass spectrometry (S7, 2009; S8, 22 men, 2010); at a matched running speed of 7.3 km/h a 5 m height threat added 65 mL/min of VO2 (95 percent CI 7.1 to 123.5), about 3 percent of the running uptake, and 4.1 bpm in 19 adults (S9, 2014). Populations: resting women, students, young men, recreational adults; outcomes: VO2, cardiac output, heart rate; classes: intervention and within-person manipulation. Boundary: no study measured oxygen uptake while strikes arrived (S12 states the mask cannot be worn in a contest); the stressors were arithmetic, speech and height, not punches; additivity of the fear cost onto a fighting workload is assumed.

The physics underneath

The sympathetic response is beta-adrenergic and its cardiac arm is sized for anticipated work rather than the work being done, so cardiac output and heart rate rise out of proportion to oxygen flux; by the Fick relation heart rate alone cannot fix VO2 because stroke volume and extraction also move, and under stress the calibration between heart rate and energy that holds during exercise breaks.

The sources
← Previous axiom: 1.1, Defensive reflex gain is a controller set by predicted proximity, threat probability and cover, not an alarm that fires at a thresholdNext axiom: 1.3, The 'ten times' figure is a workload ratio, the coded cost of fighting activity against seated rest, and the fear's share of that excess is about one to two percent →Link to this axiom: #ax-1-2
1.3 The 'ten times' figure is a workload ratio, the coded cost of fighting activity against seated rest, and the fear's share of that excess is about one to two percentVERIFIEDFighting work drives oxygen use to ten to fifteen times seated rest over a round while fear adds a few tenths of a MET; the tenfold figure measures work.Open this axiomClose this axiom
The axiom in full

The Compendium of Physical Activities codes moderate-pace martial arts (judo, karate, kick boxing, tae kwon do, Muay Thai boxing) at 10.3 METs, boxing sparring 7.8 and boxing in a ring 12.8, where 1 MET is the seated 3.5 mL O2/kg/min convention (S10, 2011); 11 elite boxers ran above 90 percent of VO2max in every round of a simulated 3 x 3 min contest, about 15 METs for a VO2max near 55 mL/kg/min (S12, 2018); a 3 x 2 min semicontact bout cost about 680 kJ over its 8 minutes in 10 boxers, 77 percent aerobic, which is about 85 kJ/min against a conventional resting 5.2 kJ/min for 70.6 kg, about 16 times rest (S13, 2014; ratio derived from the paper's totals); fear alone adds 0.2 to 0.3 METs (S5, S6) and threat at matched work about 3 percent (S9). Excess of the fight over rest about 14 METs, excess of the fear about 0.2 to 0.3 METs, a share of about 1.4 to 2 percent. Populations: amateur and elite boxers, resting adults; outcomes: VO2 and total metabolic energy; classes: within-person measurement, compendium codes, intervention for the fear term. No leg located a source for a panic-specific tenfold figure.

The physics underneath

Energy conservation puts the fight's bill in the legs, trunk and fists: external work, braking, posture and co-contraction consume oxygen at ten to fifteen times rest over a round, while the affective response, acting through beta-adrenergic thermogenesis and cardiac work, adds a few tenths of a MET; attributing the fight's bill to the fear is a category error between a workload ratio and a threat increment.

The sources
← Previous axiom: 1.2, Threat raises cardiac drive far more than oxygen uptakeNext axiom: 1.4, A metabolic multiplier has no meaning until its denominator and time window are fixed: the measured resting denominator is 35 percent below convention and the muscle-level ratio over seconds exceeds 100 →Link to this axiom: #ax-1-3
1.4 A metabolic multiplier has no meaning until its denominator and time window are fixed: the measured resting denominator is 35 percent below convention and the muscle-level ratio over seconds exceeds 100VERIFIEDA metabolic ratio changes from 1.3 to over 100 as the reference state and window change, so a multiplier without both named measures nothing.Open this axiomClose this axiom
The axiom in full

Measured resting VO2 averaged 2.6 +/- 0.4 mL/kg/min in 642 women and 127 men by ventilated-hood calorimetry, so the 3.5 convention overstates resting oxygen uptake by 35 percent and resting energy by 20 percent on average (S11, 2005, correlational measurement); muscle ATP turnover can rise more than 100-fold from rest to maximal exercise, phosphocreatine is mostly spent in 10 to 15 s and glycolysis is acid-limited within 30 to 90 s (S14, 2020 review); over one 3-minute round the whole-body ratio is about 15 (S12) and over an 8-minute bout with breaks about 16 (S13). Population: heterogeneous adults, elite and amateur boxers; outcomes: resting VO2, ATP turnover, bout energy. Boundary: muscle figures are per kilogram of muscle, not whole body; the Compendium and the bouts use the 3.5 convention.

The physics underneath

A ratio is a property of an integration window and a reference state, not of a condition: the same body yields 1.3 (fear at rest), 15 (a round) or over 100 (one exchange at the muscle) depending on what is divided by what over how long, and pulmonary gas exchange cannot see the stored phosphocreatine term over seconds, so a seconds-scale expenditure is not even measurable by the instrument the convention assumes.

The sources
← Previous axiom: 1.3, The 'ten times' figure is a workload ratio, the coded cost of fighting activity against seated rest, and the fear's share of that excess is about one to two percentNext axiom: 1.5, Catecholamine concentration and heart rate are signals, not fuel →Link to this axiom: #ax-1-4
1.5 Catecholamine concentration and heart rate are signals, not fuelVERIFIEDAn adrenaline level several times baseline and a heart at 90 percent of maximum carry no energy of their own and predict neither expenditure nor collapse.Open this axiomClose this axiom
The axiom in full

Plasma epinephrine rose from 702 at baseline to 1,871 after kata and 3,269 pmol/L after 3 min of kumite in 10 competitive karate athletes, a 4.7-fold concentration ratio sampled within 10 minutes after, with no oxygen measurement (S15, 2012, randomised crossover); under mental stress VO2 rose 20 percent while cardiac output rose 25 percent and exceeded its oxygen-predicted value (S5, S7); the mean task heart rate across a shoot-or-withhold task under shock threat was 68.5 bpm in 76 participants, with threat producing preparatory slowing then action acceleration rather than sustained tachycardia (S16, 2019); the On Combat claim that fighters are exhausted because they burned all their adrenaline (S26, argument) is physically impossible. Populations: karate athletes, resting adults, civilians and police recruits; outcomes: plasma catecholamines, VO2, cardiac output, heart rate; classes: intervention and within-person measurement, one argument source audited.

The physics underneath

A plasma concentration is secretion minus clearance over a distribution volume, not a flux of fuel; catecholamines circulate at nanomolar levels and their oxidation yields no usable energy, so a doubled or quintupled signal implies neither a doubled energy flux nor an emptied tank; heart rate is a control signal moved by sympathetic drive and cardiac loading, and a given rate means different things depending on why it arrived.

The sources
← Previous axiom: 1.4, A metabolic multiplier has no meaning until its denominator and time window are fixed: the measured resting denominator is 35 percent below convention and the muscle-level ratio over seconds exceeds 100Next axiom: 1.6, Under threat a trained person's performance changes as a policy shift (faster, a looser criterion, a contracted range) rather than a capacity loss, and accuracy can fall, hold or rise with the task →Link to this axiom: #ax-1-5
1.6 Under threat a trained person's performance changes as a policy shift (faster, a looser criterion, a contracted range) rather than a capacity loss, and accuracy can fall, hold or rise with the taskVERIFIEDThreat raises the price of a decision and biases the criterion toward fast action, so accuracy shifts by task while discrimination, gaze and prepared skill largely survive.Open this axiomClose this axiom
The axiom in full

Against an opponent firing soap cartridges, 7 police officers lost shooting accuracy while acting faster, making themselves smaller and blinking more (S17, 2010); with a shoot-back cannon on, 36 officers shot more often at surrendering suspects, lost shot accuracy and responded faster to armed suspects while gaze behaviour was unaffected (S18, 2012); in 22 civilians and 54 recruits threat produced opposing accuracy effects, better when shooting was required and more withholding errors on phone trials (S16, 2019); in 22 soldiers anxiety reduced shooting accuracy, decision making and arithmetic, exercise-raised arousal prevented the shooting decrement, and soldiers more often failed to shoot when fired at (S22, 2014); in 9 soccer players exercise at adrenaline threshold and at maximal power left decision accuracy unchanged and made decisions faster than at rest (S20, 1999); attentional control theory predicts efficiency fails before effectiveness through inhibition and shifting (S19, 2007). Populations: police officers and recruits, civilians, infantry soldiers, college athletes; outcomes: accuracy, speed, response bias, gaze; classes: within-person manipulations and one theory. Boundary: firearms use-of-force tasks, not strikers; samples of 7 and 36 in the two officer studies; effects are task-specific and cannot be averaged into one 'thinking' score.

The physics underneath

Threat first raises the cost of a decision and shifts what the system optimises: speed over accuracy, a criterion biased toward acting because a missed attack is costly, and a range policy that treats the opponent as closer; inhibition and shifting pay first while perception and prepared action largely survive, so a missed defense cannot identify which stage failed and overall accuracy cannot separate a criterion shift from lost discrimination.

The sources
← Previous axiom: 1.5, Catecholamine concentration and heart rate are signals, not fuelNext axiom: 1.7, Perceptual narrowing under threat is a reallocation toward the threat axis, documented by retrospective self-report whose prevalence swings with the instrument, co-occurs with heightened detail, and establishes the experience, not a performance loss →Link to this axiom: #ax-1-6
1.7 Perceptual narrowing under threat is a reallocation toward the threat axis, documented by retrospective self-report whose prevalence swings with the instrument, co-occurs with heightened detail, and establishes the experience, not a performance lossVERIFIEDThreat re-weights visual bandwidth toward the threat axis, so recalled tunnel vision and heightened detail co-occur, a reallocation rather than a loss of perception.Open this axiomClose this axiom
The axiom in full

Across 113 shootings recalled by 80 officers, tunnel vision was reported in 51 percent of incidents, heightened visual detail in 56 percent, both in 15 percent, auditory blunting in 82 percent, heightened hearing in 20 percent and slow motion in 56 percent (S23, 2006, retrospective interviews, denominators are incidents); under return fire 7 officers blinked more and spent more time with eyes closed (S17, 2010); under anxiety 36 officers' gaze behaviour was unaffected while decisions shifted toward threat-based inference (S18, 2012); a fearful-face cue improved contrast sensitivity at the cued location in laboratory adults (S21, 2006, intervention, non-combat). Populations: police officers, laboratory adults; outcomes: recalled distortions, blink and gaze measures, contrast sensitivity. Boundary: the field figures are memories of a shooting, not measurements during it, and prevalence of tunnel vision moves between about 51 and 79 percent across surveys; none of these studies measured what the officer could see.

The physics underneath

When heightened detail and tunnel vision are reported at near-equal rates in the same incidents, the visual system is re-weighting bandwidth toward the threat rather than failing; the sampling policy changes (shorter dwell, more blinks) and the cost falls only on information off the threat axis, which is the perceptual face of the same gain controller that sets reflex strength.

The sources
← Previous axiom: 1.6, Under threat a trained person's performance changes as a policy shift (faster, a looser criterion, a contracted range) rather than a capacity loss, and accuracy can fall, hold or rise with the taskNext axiom: 1.8, Heart-rate bands (115, 145, 175 bpm) as thresholds of cognitive collapse are refuted: skilled perceptual-motor performance proceeds at 90 percent or more of maximal heart rate, and threat effects on decisions appear at a mean of 68.5 bpm →Link to this axiom: #ax-1-7
1.8 Heart-rate bands (115, 145, 175 bpm) as thresholds of cognitive collapse are refuted: skilled perceptual-motor performance proceeds at 90 percent or more of maximal heart rate, and threat effects on decisions appear at a mean of 68.5 bpmVERIFIEDHeart rate is a correlate re-labelled by its cause, so bpm bands cannot set cognitive thresholds; boxers perform at 90 percent of maximum and threat shifts decisions at 68 bpm.Open this axiomClose this axiom
The axiom in full

The 1997 Siddle-Grossman chart places complex-skill breakdown above 145 bpm and cognitive collapse above 175 bpm, names no experiment, and exempts exercise-induced heart rate (S24, argument; S26); a trainer's published critique quotes Lewinski that the idea a high heart rate alone causes loss of fine motor skills is a myth (S25, argument); 11 elite boxers held peak heart rate at or above 90 percent of maximum in every round while executing the full striking and defending repertoire of a simulated contest (S12, 2018); the shoot-or-withhold task ran at a mean heart rate of 68.5 bpm and threat still shifted accuracy and speed in 76 participants (S16, 2019); exercise-raised arousal prevented rather than caused the anxiety-driven loss of shooting accuracy in 22 soldiers (S22, 2014). Populations: elite boxers, civilians and recruits, infantry soldiers; outcomes: heart rate beside skill, accuracy and decisions. Boundary: no study has run strikers at 190 bpm through a decision battery mid-round; the refutation is of the chart's causal reading, not of a claim that extreme arousal never impairs.

The physics underneath

The same heart rate means different things depending on why it arrived, which the chart concedes by exempting exercise; a variable that must be re-labelled by its cause before it predicts anything is a correlate, not a cause, and the causal variables are the threat appraisal and the attentional policy, not beats per minute.

The sources
← Previous axiom: 1.7, Perceptual narrowing under threat is a reallocation toward the threat axis, documented by retrospective self-report whose prevalence swings with the instrument, co-occurs with heightened detail, and establishes the experience, not a performance lossNext axiom: 2.1, Training preserves threatened execution while the threat response stays up (modulation, never elimination) →Link to this axiom: #ax-1-8

What governs whether sparring, stress inoculation and breathing training preserve decisions under attack, how far gym composure transfers to an unfamiliar threat, and whether composure can be the whole discipline?

Training composure · eight axioms

The mechanism under test is composure training: stress inoculation, sparring dose, breathing and the transfer from gym to real threat. The easy answer this vector was told not to settle for: “You fight how you train; enough sparring and nothing rattles you”
2.1 Training preserves threatened execution while the threat response stays up (modulation, never elimination)VERIFIEDRepresentative practice under threat redirects attention and effort so motor output holds while heart rate, anxiety and catecholamines stay elevated; composure is preserved allocation, not a quieter alarm.Open this axiomClose this axiom
The axiom in full

In police officers trained under a shoot-back opponent (27 officers, four weekly one-hour sessions, 48 rounds; S1), shooting accuracy that fell under threat at pretest no longer fell at posttest and still held at a four-month retention test, while anxiety, heart rate and rated effort stayed elevated under high threat at every time point. In 66 officers (S4), both imagery groups kept accuracy under threat while mean anxiety, heart rate and mental effort were uniformly higher in all groups and the control's accuracy fell. In military survival training (S16, S17), the more resilient Special Forces trainees showed higher, not lower, neuropeptide-Y and norepinephrine under interrogation stress and returned to baseline by 24 hours while non-SF trainees stayed below it. No opened source at any dose shows the threat response eliminated. Boundary: police and military use-of-force populations; outcomes are motor accuracy and biomarkers; INTERVENTION (S1, S4) and CORRELATIONAL (S16, S17); no Thai fighter in a ring was measured.

The physics underneath

Anxiety loads the goal-directed attentional system and lets the stimulus-driven system pull attention toward threat (S30). Training changes where attention is spent and how much effort is paid, so output holds while the alarm runs; observed calm is an effectiveness reading blind to the efficiency cost paying for it (S30, S1). The elite phenotype is a larger, better-shaped, faster-recovering alarm, not a smaller one (S16).

The sources
← Previous axiom: 1.8, Heart-rate bands (115, 145, 175 bpm) as thresholds of cognitive collapse are refuted: skilled perceptual-motor performance proceeds at 90 percent or more of maximal heart rate, and threat effects on decisions appear at a mean of 68.5 bpmNext axiom: 2.2, Threat moves the decision criterion before it moves the hands, and the practice that rescues execution does not rescue the criterion →Link to this axiom: #ax-2-1
2.2 Threat moves the decision criterion before it moves the hands, and the practice that rescues execution does not rescue the criterionVERIFIEDWhen the judged person is the threat, anxiety lowers the evidence needed to treat approach as attack; practice protects the trigger finger's accuracy, not that threshold.Open this axiomClose this axiom
The axiom in full

Under a shoot-back cannon (S3), officers shot more often at suspects who were surrendering, shot faster when a gun was present and shot less accurately, with gaze behaviour unchanged by anxiety: they acted on threat-related inference rather than on what they were looking at. After three consecutive weeks of shooting-decision practice in four practice groups (S2: 57 officers), anxiety and false-positive shots under high threat were as large at posttest as at pretest in every group. In a realistic lethal-force scenario (S18: 122 active-duty officers, CORRELATIONAL), mean performance was 59 percent of the scored items, at least one lethal-force error occurred in 27 percent of officers, greater stress reactivity predicted poorer performance and years of service did not remove the effect. Boundary: police use-of-force decision tasks in which the judged person is the threat source; outcome is decision accuracy and false positives. Opposing row: under threat of electric shock in a military friend/foe task (S19), arousal rose and decision time and confidence changed but response sensitivity and bias did not, and stimulus uncertainty mattered more than the stressor; the criterion shift is documented when the stimulus to be judged is itself the source of the threat, not when the threat is an unrelated shock.

The physics underneath

A criterion shift is a change in how much evidence is needed to act, not a loss of perception: gaze and distance judgments stay intact while the threshold at which an approach counts as an attack drops (S3, S30). Practice that rehearses the motor act under threat protects execution (Axiom 1) and leaves the prior on danger untouched, because the prior is set by the threat itself.

The sources
← Previous axiom: 2.1, Training preserves threatened execution while the threat response stays up (modulation, never elimination)Next axiom: 2.3, Transfer follows shared elements and representative structure, not stressor intensity; added stressors and long hard doses are the weakest levers →Link to this axiom: #ax-2-2
2.3 Transfer follows shared elements and representative structure, not stressor intensity; added stressors and long hard doses are the weakest leversVERIFIEDUnder load a learner retrieves the solution rehearsed with the same information and affordances; stressor intensity adds arousal without adding structure, so representative variety transfers and hard volume does not.Open this axiomClose this axiom
The axiom in full

In 84 police officers trained in knife defence (S7), the arm trained with added stressors (n=51) reported more stress and effort than the control arm (n=33), yet both arms improved on every rated performance variable in reality-based scenarios (all p<.05) and the stressors added no measurable gain. Across pressure-training trials pooled in S9, long doses had the smallest effect on performance under pressure. Across 37 studies and 1,837 participants (S8), stress inoculation reduced state and performance anxiety and improved performance under stress, with anxiety effects larger than performance effects. Stress training generalised across a change of stressor and of task only when the new task had itself been instructed (S10); consistent training environments raised efficiency and an inconsistent test environment disrupted it (S37); variable practice served transfer to a novel condition and specific practice served retention (S33); affordance similarity, not intensity, predicted transfer between sports (S34). Boundary: police (S7), laboratory learners (S10, S33, S37), pooled sport and tactical trials (S8, S9); outcomes are rated scenario performance, task accuracy and anxiety; INTERVENTION throughout. No trial randomised sparring dose and tested a first unfamiliar assault; that estimand is empty.

The physics underneath

What a learner retrieves under load is the solution rehearsed under the same information and affordances; a stressor that shares nothing with the test adds arousal without adding retrievable structure (S10, S34, S30). Repeated sparring with one partner maximises retention of a partner-specific solution and minimises the variability that buys transfer (S33); uncalibrated exposure also inflates expectations of future stressors (S35).

The sources
← Previous axiom: 2.2, Threat moves the decision criterion before it moves the hands, and the practice that rescues execution does not rescue the criterionNext axiom: 2.4, Habituation is bound to the stimulus that produced it, partial across systems and time-limited; the felt sense of composure habituates first and carries least →Link to this axiom: #ax-2-3
2.4 Habituation is bound to the stimulus that produced it, partial across systems and time-limited; the felt sense of composure habituates first and carries leastVERIFIEDRepeated exposure to one stressor quiets cortisol and self-report but not heart-rate reactivity, keyed to that stimulus and lost within weeks; a novel threat meets an unhabituated or facilitated response.Open this axiomClose this axiom
The axiom in full

When healthy adults face the same psychosocial stressor on consecutive days, about two-thirds habituate their cortisol response (S20, review of repeated-measures designs) while cardiovascular reactivity shows little or no habituation; in one repeat design (S21) cortisol habituated between sessions 24 hours apart and a renewed challenge ten weeks later activated the axis as if new, and across a four-month interval endocrine habituation did not occur while subjective responses habituated faster. In rodents, an axis habituated to a chronic homotypic stressor shows a facilitated, larger response to a novel heterotypic one (S22, S23). In 1,837 pooled participants (S8), anxiety outcomes moved more than performance outcomes. Expertise did not abolish threat cost: 66 officers with more martial-arts experience performed better under threat yet still deteriorated relative to their own low-threat baseline (S31). In 103 elite kickboxers (S36, self-report, the one combat-sport row), arousal control and confidence went with challenge appraisal and worry with threat appraisal and poorer self-rated performance: appraisal, not suppression, is what the athletes report regulating. Boundary: human data are psychosocial laboratory stressors, not sparring; the heterotypic facilitation is rodent and is INFERENCE when applied to an unfamiliar opponent; S31 and S36 are correlational. By the same law, The Physics of Flow States (February 2026) reports arousal regulation in non-combat performers: heterotypic evidence relative to attack, with no presumptive transfer to it.

The physics underneath

Habituation is non-associative learning keyed to the known parameters of a stimulus (partner, room, rules, the guarantee it stops); the endocrine layer quiets to the familiar, the cardiovascular layer barely does, and the subjective layer quiets fastest, so a person can feel composed in a familiar room on a measurement that does not travel (S20, S21, S22).

The sources
← Previous axiom: 2.3, Transfer follows shared elements and representative structure, not stressor intensity; added stressors and long hard doses are the weakest leversNext axiom: 2.5, Breathing and HRV regulation are resource-allocation levers bounded by stressor type and task headroom; an HRV change is not a certificate that panic was disabled →Link to this axiom: #ax-2-4
2.5 Breathing and HRV regulation are resource-allocation levers bounded by stressor type and task headroom; an HRV change is not a certificate that panic was disabledVERIFIEDSlow breathing frees attentional capacity under psychological threat without lowering felt stress or exertion heart rate, so it helps decisions when the task has headroom and fails under metabolic load.Open this axiomClose this axiom
The axiom in full

In a police critical-incident simulator (S12, randomized pairs), the tactical-breathing group scored higher on rules-of-engagement and communication and the gain was not associated with lower self-reported stress. In 12 special-response officers randomized to a three-day package of breathing with biofeedback, education, imagery and scenarios (S5), warehouse-scenario correct decisions averaged 3.83 of 4 against 2.33 of 4 for controls, while heart-rate advantages were not significant in both scenarios. In 57 officers given HRV biofeedback during lethal-force scenarios (S6), errors fell and stayed down at 12 months while maximum heart rate relative to rest rose from pre to post. In 100 student officers (S13), tactical breathing raised the first shot by 1.9 points of a 10-point shot score at 10 m with no difference on the easier later shots. In 38 officers after anaerobic fatigue (S14), about 45 seconds of breathing changed nothing: not accuracy, decisions, heart rate, RMSSD or HF-HRV. In 351 Marines (S11, cluster-randomized), breathing with biofeedback raised respiratory sinus arrhythmia and lowered arousal during multimedia exposure, with PTSD screens as endpoints. In 80 cadets (S15, four arms, four weeks), breathing and mindfulness raised shooting scores and the paper cautions that LF/HF is not a sympathetic index. Boundary: police and military trainees; outcomes are scenario decisions, shot scores and HRV; INTERVENTION throughout; gains appear under psychological threat with task headroom and vanish under metabolic load (S14) and on easy tasks (S13's later shots); breathing was bundled in S5 and S11, so its standalone share is unidentified there.

The physics underneath

Slow respiration raises vagal drive on the sinoatrial node and frees processing capacity under anxiety (S30 on efficiency; S12 on resources without lower felt stress); it does not reduce the exertion-driven heart rate of a fight round, and a favourable HRV reading can be produced by the breathing itself, so HRV is not independent evidence that threat regulation improved (S15).

The sources
← Previous axiom: 2.4, Habituation is bound to the stimulus that produced it, partial across systems and time-limited; the felt sense of composure habituates first and carries leastNext axiom: 2.6, Trained composure has a retention horizon →Link to this axiom: #ax-2-5
2.6 Trained composure has a retention horizonVERIFIEDHeld at four and twelve months, returning toward baseline by eighteen without re-dosing.Open this axiomClose this axiom
The axiom in full

In 57 front-line officers (S6, longitudinal within-subjects), reductions in lethal-force errors and in the extent and duration of autonomic arousal after HRV-biofeedback scenario training were maintained at 12 months and began returning to pre-intervention levels at 18 months; maximum heart rate relative to rest was lowest at 12 months and plateaued at 18. In 27 officers (S1), the protection of shooting accuracy under threat held at a four-month retention test. In healthy adults, cortisol habituation to a repeated stressor was gone ten weeks later (S21). Within the same programme's five-day block (S5 lineage), average heart rate fell and controlled respiration improved across scenarios as they grew more graphic, and lower perceived stress arrived before performance improved. Boundary: police use-of-force trainees; outcomes are lethal-force errors, shot accuracy under threat and arousal; INTERVENTION; the 18-month curve is one cohort of 57 with no untrained comparison across the interval; no combat-sport cohort measured retention of threat-regulated decisions.

The physics underneath

Composure is a maintained regulation skill: the error reduction arrives first and the arousal profile reorganises over a year, then both drift back when the exposure that maintained them stops; it behaves like a conditioned capacity with a half-life, not a one-time acquisition (S6, S21).

The sources
← Previous axiom: 2.5, Breathing and HRV regulation are resource-allocation levers bounded by stressor type and task headroom; an HRV change is not a certificate that panic was disabledNext axiom: 2.7, Perceptual anticipation is a separate acquired axis →Link to this axiom: #ax-2-6
2.7 Perceptual anticipation is a separate acquired axisVERIFIEDExperts read pre-attack kinematics, which calm does not supply.Open this axiomClose this axiom
The axiom in full

On the mat, karate experts blocked and counterattacked faster and more accurately than novices and detected a recurring attack pattern after its fifth repetition, which novices never did (S24). Elite officers shot more accurately than rookies (74 percent against 53.8 percent of trials in the opened record, S25) and made very few decision errors in the cell-phone condition, with longer quiet-eye durations on the weapon or where it would appear and more fixations on pre-attack actions of the limbs and hands. Across expert-novice visual search studies in individual and combat sports (S26, meta-analysis), experts used fewer fixations of longer duration on fewer, central body locations. In 35 close-quarters trainees (S32, pre-post, no control), 12 hours of training improved performance and tactical behaviour while gaze behaviour and response time did not significantly change and heart rate and salivary stress fell. In a reality-based training study (S7) both arms improved regardless of added stress, so skill acquisition, not the stress layer, did the measurable work. Boundary: karate experts, police elite-vs-rookie, combat-sport athletes; outcomes are block and counter latency, decision accuracy and fixation metrics; CORRELATIONAL expert-novice contrasts (S24, S25, S26) and INTERVENTION (S32, S7); expert-novice contrasts cannot separate training from selection.

The physics underneath

An expert's apparent calm is partly a consequence of earlier, more usable information: anchoring central vision on the trunk lets peripheral vision pick up distal limb motion, and pattern knowledge turns an ambiguous approach into a readable one (S24, S26). A novice made fearless by any means would still be struck, because the information pickup is not there; composure permits the skill to be deployed and does not contain it.

The sources
← Previous axiom: 2.6, Trained composure has a retention horizonNext axiom: 2.8, The sparring dose that habituates a fighter to being hit is a head-impact dose, taxed on the processing substrate composure protects →Link to this axiom: #ax-2-7
2.8 The sparring dose that habituates a fighter to being hit is a head-impact dose, taxed on the processing substrate composure protectsVERIFIEDHead impacts in sparring trigger hours of corticomotor inhibition and, in volume, processing-speed loss, so the dose that habituates to being hit taxes the capacity composure is meant to preserve.Open this axiomClose this axiom
The axiom in full

In 20 amateur boxers and Muay Thai athletes after three three-minute sparring rounds, against 20 controls doing mock sparring (S27), one hour later the sparring group showed greater corticomotor inhibition on transcranial magnetic stimulation, altered motor control and lower cognitive test performance, with values back toward baseline by 24 hours. In 224 professional fighters with 22 controls (S28, cohort), more fights, more years or a higher Fight Exposure Score went with lower thalamic and caudate volume and slower processing speed, processing speed falling 2 to 3 percent per one standard deviation of regional volume lost. In 92 active professional MMA fighters (S29, cross-sectional, conference report), more sparring rounds per week went with greater white-matter hyperintensity volume after adjustment for age, sex, education, fight count and scanner. Pooled pressure-training doses that were long produced the smallest performance effect (S9). Boundary: amateur boxers and Muay Thai athletes (S27, INTERVENTION, acute), professional fighters (S28, S29, CORRELATIONAL); outcomes are corticomotor excitability, cognitive scores, brain volume and white-matter lesions; S29's underlying paper was not opened here. the brain under head impact vector owns the injury physics and thresholds; this axiom carries only the training-variable conflict.

The physics underneath

Threat-learning exposure and head-impact exposure are different variables bundled in one activity; the impacts prompt an inhibitory GABAergic response that slows corticospinal transmission for hours (S27) and, cumulatively, cost the processing speed that anxiety forces a performer to overspend (S30, Axiom 1) and that resolving an approach into a read requires (Axiom 2). Hard sparring volume is the weakest pressure-training dose (S9) and the most expensive one (S28), dominated on both axes.

The sources
← Previous axiom: 2.7, Perceptual anticipation is a separate acquired axisNext axiom: 3.1, Peak force is a stiffness readout of the coupled collision; impulse is the quantity the target keeps →Link to this axiom: #ax-2-8

What governs the force, impulse and effective mass of a punch and a roundhouse kick, and how do target properties determine the deformation and loading the impact delivers?

Strike force and impact · eight axioms

The mechanism under test is strike mechanics and impact: how a punch and a roundhouse kick make force and what the impact does mechanically to the target, ending at tissue deformation and head loading. The easy answer this vector was told not to settle for: “Power comes from the hips; turn your hip over and the kick lands harder”
3.1 Peak force is a stiffness readout of the coupled collision; impulse is the quantity the target keepsVERIFIEDInterface stiffness sets how a fixed impulse is spread in time, so softer contact lowers peak force and loading rate while the target's momentum change stays the same.Open this axiomClose this axiom
The axiom in full

For a strike of closing speed u and reduced mass μ = m_s m_t/(m_s + m_t), the target's momentum change is J = (1 + e) μ u and does not depend on the interface stiffness k, while peak force scales as u√(kμ), contact time as π√(μ/k) and peak loading rate as uk; softening the interface (glove, pad, foam, yielding tissue) lowers peak force by the square root of the stiffness ratio and lengthens contact by its inverse while leaving delivered impulse nearly unchanged. Measured on mechanical drop rigs (INTERVENTION, controlled physical measurement), not on people: a 50 mm EVA covering in place of 25 mm lowered peak force and lengthened contact while force-time area stayed within 2.5 percent in 22 of 24 glove and drop-height combinations (S4, 2018); pneumatic gloves cut peak force 25 to 40 percent, cut peak loading rate 47 to 63 percent and lengthened contact more than 50 percent, with peak-force change tracking contact time at R² above 0.96 (S5, 2018); 26 to 34 percent of a 253-drop upward drift in peak force came from the mat ageing, not the impactor (S6); mat thickness and compaction changed net force by multiples under an 8 kg weight dropped from 15 to 50 cm (S21, 2024). Boundary: the restitution and series compliance of glove, skin and neck for a fist on a head were not separately measured in any opened source; the algebra is assumed to transfer.

The physics underneath

A compliant contact is a mass on a spring for the duration of the collision (S26, S27). Momentum conservation fixes the impulse from μ, u and e alone; the spring only sets how that impulse is spread in time. Putting the measured 25 to 40 percent peak reduction into the half-sine model gives a stiffness ratio r of 0.36 to 0.56, predicting a loading-rate reduction of 44 to 64 percent and contact lengthening of 34 to 67 percent; the rig measured 47 to 63 percent and more than 50 percent (S5), three independently measured ratios with no free parameter.

The sources
← Previous axiom: 2.8, The sparring dose that habituates a fighter to being hit is a head-impact dose, taxed on the processing substrate composure protectsNext axiom: 3.2, Effective mass is a direction-dependent contact-point inertia defined by the collision model, and the three operational definitions disagree by up to an order of magnitude for the same punch →Link to this axiom: #ax-3-1
3.2 Effective mass is a direction-dependent contact-point inertia defined by the collision model, and the three operational definitions disagree by up to an order of magnitude for the same punchVERIFIEDDelivered impulse is (1+e)μu and μ is capped by the lighter body, so a head sees only a few kilograms however much body is coupled behind the fist.Open this axiomClose this axiom
The axiom in full

In striking studies 'effective mass' names three estimands: peak force over fist acceleration on a rigid plate, impulse over incoming hand speed on a padded plate, and the inertial mass back-computed from a surrogate head's momentum change. For straight punches they read 30.39 to 31.17 kg in 30 trained boxers (S10, 2025) or 18.50 to 18.95 kg, about 21 percent of body mass, in elite boxers (S11, 2024); 2.60 kg in 50 untrained adults (S1, 2021); and 2.9 ± 2.0 kg in Olympic boxers striking a Hybrid III head (S7, 2005). The gap is mechanical, not a training difference: the target receives (1 + e) μ u, and against a head of a few kilograms the reduced mass μ cannot exceed the head's own mass however much body is coupled behind the fist, while against a wall-mounted plate μ tends to the striker's coupled inertia (S26, S27). The force-over-acceleration definition is in addition algebraically anti-correlated with acceleration, so the findings that effective mass falls as fist acceleration rises and that hooks read lower effective mass than straights despite higher peak force (S10, S14), and that a side kick reads 73.09 percent of body mass against 20.12 percent for a turning kick on a twofold acceleration denominator (S23, 2024), are partly definitional. Populations: trained, elite and Olympic boxers, untrained adults, Taekwon-Do practitioners; outcome: effective mass by each definition; class: CORRELATIONAL cross-sections and within-person descriptives, the MODEL supplying the reconciliation. No study measured all three definitions on one punch.

The physics underneath

The contact-point inertia in direction n is [nᵀ A M⁻¹ Aᵀ n]⁻¹, a property of configuration, segment inertias, joint coupling over the collision timescale and support constraints (S26); it is neither fist mass nor body mass. Coupled inertia and endpoint speed trade against each other because stiffening a joint to recruit proximal mass also loads the distal segment during acceleration, so the optimised quantity is the product m_e v and the optimal split depends on target mass (S14). Reactive stiffening cannot act inside a 5 to 15 ms contact, so only pre-set activation and skeletal alignment raise coupled inertia, consistent with training tenure correlating with effective mass (S14); this is an inference from timing, not a measured latency.

The sources
← Previous axiom: 3.1, Peak force is a stiffness readout of the coupled collision; impulse is the quantity the target keepsNext axiom: 3.3, The kinematic input to the collision is the contact-point velocity at first contact, which sits well below the limb's peak speed and is the striker variable that tracks delivered force →Link to this axiom: #ax-3-2
3.3 The kinematic input to the collision is the contact-point velocity at first contact, which sits well below the limb's peak speed and is the striker variable that tracks delivered forceVERIFIEDOnly the contact point's velocity at first contact enters the collision, it sits 22 to 59 percent below the limb's peak, and it is the striker lever that never saturates.Open this axiomClose this axiom
The axiom in full

Fist or foot speed peaks before contact and falls by the time the strike lands: in 16 elite and junior boxers mean peak fist speed was 7.16 m/s against a contact speed of 5.56 m/s (elite) and 6.32 against 4.87 m/s (junior) (S2, 2022); in 24 skilled Muay Thai, Karate and Taekwondo practitioners foot speed at contact was 41 to 54 percent of its earlier maximum (S3, 2017). Speed at contact is the striker-side quantity that covaries with delivered load on one rig: foot velocity at impact against relative impact force r = 0.66 in those 24 practitioners (S3); pre-impact hand velocity explained 39 and 38 percent of punch-force variance in 42 amateur boxers against 10 percent for leg forces (S12, second-hand through a 2013 review); elite boxers' higher contact speed came with about 1,508 N against 1,035 N for juniors on one target (S2). Populations: trained boxers and kickers; outcome: contact velocity and impact force within one rig; class: CORRELATIONAL. Bound: speed enters impulse linearly and deformation energy as its square, ½ μ (1 − e²) u², so against a light target that caps μ, speed is the one striker lever that never saturates (S27).

The physics underneath

Contact velocity is the vector sum of shoulder or hip translation and each segment's absolute angular velocity crossed with its lever arm; segments can reinforce or oppose one another, so fast trunk rotation does not guarantee a large component into the target, and the pre-contact deceleration in S2 and S3 is the chain decoupling or the striker braking before the pad (S26). Peak speed measured anywhere on the limb is not the collision input; the contact point's velocity in the contact normal at first contact is.

The sources
← Previous axiom: 3.2, Effective mass is a direction-dependent contact-point inertia defined by the collision model, and the three operational definitions disagree by up to an order of magnitude for the same punchNext axiom: 3.4, The ground supplies the external impulse and the couple but does no work at a planted foot; joint work funds the strike, the chain redistributes it, and the leg contribution is a rig-dependent regression coefficient, not a fixed share →Link to this axiom: #ax-3-3
3.4 The ground supplies the external impulse and the couple but does no work at a planted foot; joint work funds the strike, the chain redistributes it, and the leg contribution is a rig-dependent regression coefficient, not a fixed shareVERIFIEDGround friction and the foot couple supply external impulse and angular impulse, joint work supplies energy, and the chain redistributes both toward the contact point without creating either.Open this axiomClose this axiom
The axiom in full

A striker's whole-body linear momentum changes only through the friction impulse at the feet, capped by μ_s N Δt, and whole-body angular momentum only through the moment of the ground-reaction pair about the centre of mass; joint torques redistribute momentum between segments but cannot create it, and an ideal planted foot does no mechanical work (P = R · v_foot = 0), so muscle work funds the limb's kinetic energy (S26, S27; MODEL). The measured chain shows the redistribution: in 16 boxers trunk angular velocity rose as lead-leg ground force rose and rear-leg ground force fell, peaking at impact, the lead leg acting as the pivot (S2, 2022). How much punch force the legs explain depends on sample and rig: rear- and lead-leg peak-force models accounted for roughly half of cross peak-force variance in 12 elite amateur boxers on a suspended chin-height plate at 12,000 Hz (S13, as read from the abstract), but leg forces explained 10 percent against 39 and 38 percent for hand velocity in 42 amateur boxers (S12, second-hand); the widely quoted 38.46 / 37.42 / 24.12 percent leg, trunk and arm split is a 1985 dynamometer result no leg opened (S12). Rear-leg drive associated with the cross as the largest pre-fatigue output and with its resilience to fatigue (S29, 2025). Populations: elite and amateur boxers; outcome: ground-reaction variables against punch force; class: CORRELATIONAL regressions plus the MODEL.

The physics underneath

The articulated equation M(q) q̈ + h = τ + A_gᵀ R − A_cᵀ F keeps ground reaction, joint torque and contact force as separate terms; nothing travels up the chain unchanged (S26). A large vertical foot-force peak does not by itself set horizontal fist momentum because direction, point of application and duration of the ground reaction all enter the impulse. Proximal-to-distal sequencing maximises endpoint speed by progressively decoupling distal segments from proximal inertia, the same lever that lowers coupled inertia at contact.

The sources
← Previous axiom: 3.3, The kinematic input to the collision is the contact-point velocity at first contact, which sits well below the limb's peak speed and is the striker variable that tracks delivered forceNext axiom: 3.5, A strike's force number is comparable only within one rig with a stable interface; across rigs only impulse, momentum and limb kinematics transfer →Link to this axiom: #ax-3-4
3.5 A strike's force number is comparable only within one rig with a stable interface; across rigs only impulse, momentum and limb kinematics transferVERIFIEDTarget stiffness, mounting and sensor chain set a peak-force reading more than the striker does, so force compares only within one rig while impulse and limb speed cross rigs.Open this axiomClose this axiom
The axiom in full

Reported peak punch force across 23 studies spans 989 ± 117 to 5,008.6 ± 76.3 N, a fivefold range dominated by target stiffness, mounting, padding, sampling rate and event definition rather than by who punched (S17, 2025 review); on one wall-mounted dynamometer the ladder 4,800 ± 227 (elite), 3,722 ± 133 (intermediate) and 2,381 ± 116 N (novice) is a valid rank order on that plate and nothing more in absolute terms (S17, second-hand from Smith et al. 2000); within one rig repeatability can be high (ICC 0.995 to 0.999, typical error at or under 5 percent on an instrumented water bag, S24, 2022) yet for absolute peak impact force the standard error of measurement most often exceeded the smallest worthwhile change (S25, 2023); commercial wrist and glove trackers correlated 0.23 to 0.43 with criterion peak force in 10 trained and 10 untrained punchers (S20, 2023); and the rig itself drifts, 26 to 34 percent of a 253-drop rise in peak force coming from the mat (S6). Populations: boxers across levels, trained and untrained punchers, drop rigs; outcome: peak force and its reliability; class: CORRELATIONAL device studies and reviews, INTERVENTION drop tests.

The physics underneath

A plate behind padding records F_plate = F_contact − dP_pad/dt, so surface and plate peaks differ while their impulses agree if the pad returns to rest; an accelerometer on a rotating segment reads a_G + α × r + ω × (ω × r), not contact force, and recovering force needs a dynamical model, not an assumed mass (S26); a bag's swing is a global momentum response that hides the local pressure field. A defensible impact record needs synchronized contact force and contact-point motion, a known target constraint, characterized compliance, adequate bandwidth and a defined integration window; absent those the number is the rig's.

The sources
← Previous axiom: 3.4, The ground supplies the external impulse and the couple but does no work at a planted foot; joint work funds the strike, the chain redistributes it, and the leg contribution is a rig-dependent regression coefficient, not a fixed shareNext axiom: 3.6, Head loading from a punch is kinematic →Link to this axiom: #ax-3-5
3.6 Head loading from a punch is kinematicVERIFIEDA few kilograms of effective mass at 9 to 11 m/s produce tens of g and thousands of rad/s², the rotational share set by the contact's moment arm about the head's centre and bled by the neck.Open this axiomClose this axiom
The axiom in full

Olympic boxers' straight punches to a Hybrid III frangible face produced punch force 3,427 ± 811 N at hand velocity 9.14 ± 2.06 m/s with effective punch mass 2.9 ± 2.0 kg, jaw load 876 ± 288 N, head translational acceleration 58 ± 13 g, rotational acceleration 6,343 ± 1,789 rad/s² and neck shear 994 ± 318 N (S7, 2005); professional boxers' hooks produced the largest hand-velocity change (11.0 ± 3.4 m/s), punch force 4,405 ± 2,318 N, 71.2 ± 32.2 g and 9,306 ± 4,485 rad/s² (S8, 2005). The nonzero neck shear shows the head is not a free body over the contact, so the free-collision Δv is an upper bound on head velocity change; rotational acceleration is α ≈ F d / I_head, a lever-arm quantity, so the contact's position relative to the head's centre changes angular loading by factors at constant force. The compiled finding that a roundhouse kick loads the head harder than a hook (130.11 ± 51.67 g vs 71.23 ± 32.19 g; HIC15 162.63 clench axe kick vs 24.10 uppercut) is cross-rig, a helmeted Hybrid II against a bare Hybrid III with different necks and mountings, and gives a direction consistent with the kick's higher contact speed and larger segmental inertia, not a ratio (S9, 2013). Populations: Olympic and professional boxers striking surrogate heads, taekwondo athletes on a different surrogate; outcome: head linear and rotational acceleration, jaw load, neck shear; class: INTERVENTION (controlled physical measurement on surrogates), the kick comparison CORRELATIONAL across studies. The axiom ends at loading; the brain's response belongs to the brain under head impact vector.

The physics underneath

Head momentum change is the contact impulse plus the neck impulse; head angular momentum change is the contact impulse's moment about the centre of mass plus the neck moment (S26). A freely moving head's inverse normal effective inertia at an off-centre contact is 1/m_head + (r_c × n)ᵀ I⁻¹ (r_c × n), so rotation gives the contact point extra freedom to move, and the neck removes some of it. Translational acceleration loads the skull in compression; angular acceleration shears the brain within it, and a single accelerometer at the head's centre records only the first (S9). Contact durations in finite-element head models ran 0.17 to 3.29 ms (S22). The headform inertia behind the ~5 cm lever-arm estimate was not opened by any leg; that estimate is order-of-magnitude only.

The sources
← Previous axiom: 3.5, A strike's force number is comparable only within one rig with a stable interface; across rigs only impulse, momentum and limb kinematics transferNext axiom: 3.7, Torso loading is viscous, head loading is kinematic and surface loading is stress: three loading variables, three different optimal strikes →Link to this axiom: #ax-3-6
3.7 Torso loading is viscous, head loading is kinematic and surface loading is stress: three loading variables, three different optimal strikesVERIFIEDSoft organs fail on deformation velocity times compression, the head on impulse and moment arm, surfaces on force per area, so each target rewards a different strike.Open this axiomClose this axiom
The axiom in full

In blunt chest and abdomen impacts on biological surrogates, injury tracked the viscous response VC, the maximum of deformation velocity times compression, with VC = 1.0 m/s the serious-injury level frontally and maximum compression 32 to 40 percent at the same risk (S18, 1988), and VC 1.47 m/s (chest) and 1.98 m/s (abdomen) at 25 percent probability of critical injury laterally with compressions of 38 and 44 percent (S19); a compression survivable at low speed is fatal at high deformation velocity because the viscoelastic body's resisting force rises with loading speed (S18). So the torso is loaded by penetration depth times penetration rate, the head by impulse and moment arm, and a bone or skin surface by stress F/A, where the same total force over a smaller or stiffer contact raises local stress (S1's caution on contact area and impactor rigidity). A fast, low-coupled-mass whip maximises VC; a slower, high-coupled-mass drive maximises head Δv against a braced target; a knuckle or shin edge maximises F/A. Populations: cadaveric and animal surrogates in automotive impact testing, 1988 to 1989; outcome: VC and compression at injury; class: INTERVENTION (controlled impact). Only one leg opened these sources, so SIGNAL is the honest claim even though the viscous criterion is standard injury biomechanics.

The physics underneath

Normal force is pressure integrated over the contact area, so F/A is only the average pressure; local stress needs geometry, tissue thickness, support and a rate-dependent constitutive law (S1, S26). Energy entering the target is the integral of F over indentation, not F_max (S27), and the lost translational energy ½ μ (1 − e²) u² partitions into deformation, damping, heat and vibration; the viscous criterion is the empirical finding that for soft organs the rate of deformation, not its depth alone, fails tissue.

The sources
← Previous axiom: 3.6, Head loading from a punch is kinematicNext axiom: 3.8, The roundhouse kick delivers more impulse than a punch on the same rig through larger segmental inertia and higher contact speed, but the discipline label does not change what the pad receives, and expertise shows as distance-insensitivity, not a bigger force number →Link to this axiom: #ax-3-7
3.8 The roundhouse kick delivers more impulse than a punch on the same rig through larger segmental inertia and higher contact speed, but the discipline label does not change what the pad receives, and expertise shows as distance-insensitivity, not a bigger force numberVERIFIEDLarger leg inertia and faster shin contact raise both factors of impulse against a restrained pad, while expertise shows as force that holds across distance rather than a bigger number.Open this axiomClose this axiom
The axiom in full

On one restrained strain-gauge pad, 24 skilled practitioners' roundhouse kicks read 1,547 ± 530 N (Taekwondo), 1,400 ± 419 N (Muay Thai) and 1,211 ± 219 N (Karate) with no significant between-discipline difference, and impulses of 38.8 ± 15.3, 38.1 ± 12.3 and 30.9 ± 6.9 N·s (S3, 2017), about 1.6 to 2.0 times the 19.57 N·s a dominant-fist punch delivered on a padded plate in 50 untrained adults (S1, 2021), on different rigs, so a direction and not a ratio. On a stiffer instrumented dummy the same technique reads 1,994 ± 537 N in experts against 1,477 ± 679 N in novices, maximum 3,482 N (S16, 2009); across rigs the review range for a roundhouse is 1,720 to 6,400 N with foot velocities up to 18.3 m/s (S28, 2024). Expertise shows in how force holds, not in its size: experts' impact force did not change with execution distance while novices' fell at the long distance and tracked body mass (S15, 2013; S16), all competitors having four or more years. The effective mass of a kick read as force over foot acceleration is the same definitional artefact as in punches: 73.09 percent of body mass for a side kick against 20.12 percent for a turning kick on a twofold acceleration denominator (S23, 2024). Populations: skilled Muay Thai, Karate and Taekwondo practitioners, expert and novice taekwondo competitors, Taekwon-Do practitioners; outcome: pad impact force and impulse by distance and level; class: CORRELATIONAL cross-sections. The S3 protocol allowed foot-to-shin contact and foot-marker velocity is not shin-contact velocity, so S3's numbers cannot be divided into a shin effective mass.

The physics underneath

The leg's segmental inertia is a multiple of the arm's and the shin's contact point moves faster than a fist, so both factors of (1 + e) μ u are larger against a heavy restrained pad (S27). For a rigid rotor, v_c = rω and m_eff = I/r² at the contact radius: moving contact farther from the pivot raises its speed at fixed ω and lowers its effective mass at fixed I, so the shin-versus-instep choice trades speed against coupled inertia and creates no energy (S26). Expert distance-insensitivity is the chain absorbing the spatial degree of freedom in timing rather than in delivered momentum; the novice body-mass correlation is bulk standing in for coordination.

The sources
← Previous axiom: 3.7, Torso loading is viscous, head loading is kinematic and surface loading is stress: three loading variables, three different optimal strikesNext axiom: 4.1, One defense is a margin on one clock, and a strike's measured duration depends on where the clock starts →Link to this axiom: #ax-3-8

What governs whether a defender sees and moves in time, given strike duration against reaction time, range, the cues a striker shows before a strike, and what a feint spends?

Defense as distance and time · eight axioms

The mechanism under test is defense as distance and time: one defensive opportunity, reaction time against strike duration, reading intention and feints. The easy answer this vector was told not to settle for: “Fighters have lightning reflexes; keep your hands up and your chin down”
4.1 One defense is a margin on one clock, and a strike's measured duration depends on where the clock startsVERIFIEDA defense succeeds only when the strike's arrival time, counted from the earliest readable body cue, exceeds the defender's latency plus the movement needed to clear the fist's path.Open this axiomClose this axiom
The axiom in full

A newly selected defensive movement protects the head only if the strike's arrival time, counted from the earliest usable cue, is at least the defender's cue-to-initiation latency plus the movement time needed to reach a protective state (T_A − t_C ≥ L + M). The strike's duration is not a property of the punch: in experienced boxers on a bag the fist's throw phase is about 111 to 117 ms (jab 117 ± 25, cross 111 ± 41 ms; Menzel & Potthast 2021, instrumented glove), the same punch counted from first detectable body motion to contact runs 357 ± 178 ms for straights and 477 ± 203 ms for hooks (Piorkowski 2011, amateur boxers, motion capture), and counted from a start signal 2 to 2.5 m away runs 512 ms for head punches and 610 to 664 ms for kicks (Rosalie & Müller 2013, karate, in situ). The ratio between the earliest and latest defensible start markers for one strike is about 3 to 4, and that ratio is the whole territory a defender's reading can win. Boundary: experienced and amateur boxers and karateka on bags, pads or scripted live attacks; no opened study jointly measured cue extraction, decision latency and geometric clearance in one unrestricted exchange; the inequality is a derived feasibility condition (MODEL), the durations are controlled physical measurements.

The physics underneath

Fist peak velocity of 6.6 to 12.5 m/s is reached only 8 to 21 ms before contact (Whiting 1988, four experienced boxers), so the most visible part of the strike sits far inside any visual response latency; usable time exists only in the body motion upstream of the fist in the kinetic chain, which is also upstream in time. Observation: the durations. Inference: the 3 to 4 ratio compares markers across populations and tasks.

The sources
← Previous axiom: 3.8, The roundhouse kick delivers more impulse than a punch on the same rig through larger segmental inertia and higher contact speed, but the discipline label does not change what the pad receives, and expertise shows as distance-insensitivity, not a bigger force numberNext axiom: 4.2, The visual response floor is about 200 ms of nerve and tendon that training barely moves; uncertainty is taxed in the brain, not the muscle →Link to this axiom: #ax-4-1
4.2 The visual response floor is about 200 ms of nerve and tendon that training barely moves; uncertainty is taxed in the brain, not the muscleVERIFIEDVisual-motor latency is fixed near 200 ms by conduction and tendon delay; the only trainable term is decision time, which uncertainty inflates by about 100 ms.Open this axiomClose this axiom
The axiom in full

A visually triggered defensive movement cannot begin in less than roughly 150 to 180 ms of neural processing plus about 26 to 41 ms of electromechanical delay (biceps 41 ± 13 ms, triceps 26 ± 11 ms in concentric contraction, not influenced by movement velocity; Norman & Komi 1979), so sub-100 ms responses in combat-sport protocols are discarded as anticipations by convention. Expertise compresses the neural term by only about 25 to 40 ms (athletes vs non-athletes: visuomotor reaction 243 vs 274 ms and 234.9 vs 260.3 ms, EMG onset 178 vs 217 ms; Hülsdünker cohorts, between-group). In an in-situ karate kumite decomposition, moving from one alternative to four alternatives plus an auditory distractor raised reaction time from 235.0 ± 35.0 to 351.1 ± 59.1 ms (+116 ms) while motor response time moved from 122.7 ± 27.7 to 131.7 ± 35.8 ms (+9 ms): uncertainty costs decision time almost entirely and movement time almost not at all. Generic choice RT did not predict level among 123 karate competitors including 11 world champions (Martínez de Quel 2015), and karate athletes showed faster choice responses to attack video but no simple-RT advantage (Mori 2002). Boundary: laboratory and in-situ scripted tasks on karateka, badminton players and healthy adults; the simple-vs-choice split is a within-person manipulation, the expert gaps are correlational, the electromechanical delay is a controlled physiological measurement.

The physics underneath

Motion-onset visual evoked potentials sit near 170 ms after stimulus onset and premotor activation near 150 ms; these are conduction and integration times with narrow trainable range. Electromechanical delay is tendon slack and calcium kinetics, untrainable. Choice reaction time grows with the number of alternatives (the Hick-Hyman relation, a log fit that degrades past about 3 bits). Inference: the decision term is the only latency lever training can reach.

The sources
← Previous axiom: 4.1, One defense is a margin on one clock, and a strike's measured duration depends on where the clock startsNext axiom: 4.3, Expertise relocates the start of the clock, it does not speed the clock: experts read above chance from stance and preparation, novices only once the attack is moving →Link to this axiom: #ax-4-2
4.3 Expertise relocates the start of the clock, it does not speed the clock: experts read above chance from stance and preparation, novices only once the attack is movingVERIFIEDBecause force is built in the trunk before the fist moves, a defender who reads trunk and stance gains hundreds of milliseconds of lead; faster nerves gain tens.Open this axiomClose this axiom
The axiom in full

In an in-situ temporal occlusion study of live karate attacks (Rosalie & Müller 2013; experts, near-experts and novices; attacker starting 2 to 2.5 m away; vision removed at a static stance, after preparatory head movement, or after attack initiation), only experts blocked above chance from the static on-guard stance; near-experts exceeded chance only at the two later points; novices only with the attack in motion. Against a strike whose action-to-contact time is 357 to 477 ms, the expert's gain is hundreds of milliseconds of earlier clock start, while the expert's gain in raw latency is tens of milliseconds (meta-analytic reaction-time SMDs fencing −1.07, karate −1.23, judo −0.54 are a secondary term). In live kumite (Petri, Droste & Witte 2020), recognition of arm attacks was most often inferred at early steps and reduction of interpersonal distance, an inferred interval, not a timestamped event. Expert gaze follows the kinetic chain: experienced boxers fixate head and trunk with fewer, longer fixations along the midline while novices chase fists and arms (Ripoll 1995 via review; 2021 boxer eye-tracking pilot); 11 elite boxers kept dodge and counter performance under gaze-contingent blur (Limballe 2024). Boundary: occlusion fixes what information sufficed by a cutoff, not when recognition occurred; static-stance information is probabilistic and its source unresolved; karateka, French and amateur boxers; correlational expert-novice contrasts with within-person occlusion.

The physics underneath

Force is produced proximally before it is expressed distally: weight transfer, pelvic rotation and trunk motion precede fist departure because the kinetic chain runs legs to trunk to arm, so the earliest information is in the trunk, not the fist. Gaze anchored centrally lets motion-sensitive peripheral vision pick up the limbs without saccade cost. Observation: the occlusion and gaze results. Inference: the proximal-to-distal explanation of why those cues exist.

The sources
← Previous axiom: 4.2, The visual response floor is about 200 ms of nerve and tendon that training barely moves; uncertainty is taxed in the brain, not the muscleNext axiom: 4.4, The readability horizon →Link to this axiom: #ax-4-3
4.4 The readability horizonVERIFIEDInformation a strike emits after contact time minus the defender's loop is inert, so the fist itself is undefendable and straights are more readable than hooks.Open this axiomClose this axiom
The axiom in full

Define the horizon t_h = T_contact − T_D (the defender's total loop); any cue emitted after t_h is processed only after the fist has landed. Peak fist velocity of 6.6 to 12.5 m/s occurs 8 to 21 ms before contact in four experienced boxers (Whiting 1988), about twenty times inside a 350 to 480 ms loop, so the fist is the one segment whose motion is informative only when it is already arriving. Readability is a property of the pre-horizon interval: in a four-alternative video occlusion task with ten expert boxers (Limballe 2026, motion-capture animations), anticipation accuracy (the share of correct punch-type calls across trials) fell as occlusion moved earlier and straights were read far better than hooks (lead straight 0.744, rear straight 0.660, rear hook 0.455, lead hook 0.435). A karate punch's whole kinematic and EMG sequence sits within about 400 ms with arm activity before forearm extension (VencesBrito 2011, 18 team karateka vs 19 novices), and elite amateur boxers partially flex the elbow at punch onset, a tell the authors call intent advertisement (Cheraghi 2014, eight boxers). Boundary: experienced and elite amateur boxers and team karateka; correlational expert-novice and punch-type contrasts; controlled kinematics; the 2026 accuracies come from animations, not a live opponent; the 'twenty times' ratio divides intervals from different populations and is a scale illustration.

The physics underneath

The kinetic chain runs legs to trunk to arm, so pre-horizon information exists in trunk and shoulder while distal segments peak last. A curved trajectory shares its early posture with other strikes longer than a straight does, so its branch point comes later and the pre-horizon interval carries fewer bits. Observation: velocity timing, accuracies, the sequence. Inference: the horizon definition and the branch-point account of straights versus hooks.

The sources
← Previous axiom: 4.3, Expertise relocates the start of the clock, it does not speed the clock: experts read above chance from stance and preparation, novices only once the attack is movingNext axiom: 4.5, Distance buys time only through the strike's actual trajectory →Link to this axiom: #ax-4-4
4.5 Distance buys time only through the strike's actual trajectoryVERIFIEDCentimetres buy about a hundredth of a second each, while range beyond reach forces an approach phase that triples the budget.Open this axiomClose this axiom
The axiom in full

Remaining time is the integral of distance over closing speed, and for an otherwise unchanged strike the time added by extra range is about ΔD divided by closing speed: at 8 m/s, inside Whiting's measured contact-velocity band of 5.9 to 8.2 m/s, 10 cm buys about 12.5 ms (a constant-speed local approximation, MODEL). The large gain comes from another mechanism: out of reach the attacker must move first, and locomotion is a whole-body, large-amplitude, early cue. Atha's cross as reported by Piorkowski (2011) is about 100 ms of punch preceded by about 200 ms of advance; Rosalie & Müller's attacks from 2 to 2.5 m took 512 ms (head punch) to 664 ms (kick) from signal to nominal contact against an in-reach fist flight near 110 to 117 ms (Menzel & Potthast 2021). In live karate both distance and attacking technique significantly affected the defender's reaction (2016 in-situ study). Added range also shrinks the optical size of a lateral body movement roughly as 1/R, so distance improves the deadline without guaranteeing equally readable evidence. Boundary: the per-centimetre number is a straight-path constant-speed model; a lunge changes path length and target together; experienced boxers on a bag and karateka in scripted live attacks; the optical cost is an inference with no opened measurement.

The physics underneath

Time to contact is path length over speed along the actual path including the attacker's body translation; stepping forward adds a slow, high-mass phase before the fast, low-mass fist phase, and that approach phase is the one part of a strike that is both early and large. Observation: durations at range and the karate distance effect. Inference: the per-centimetre figure and the 1/R optics.

The sources
← Previous axiom: 4.4, The readability horizonNext axiom: 4.6, A feint spends the defender's commitment, not his eyesight →Link to this axiom: #ax-4-5
4.6 A feint spends the defender's commitment, not his eyesightVERIFIEDDisguise raises uncertainty, deception loads the wrong response and adds a reversal, and its value is set by how rarely it is thrown.Open this axiomClose this axiom
The axiom in full

Two operations must be separated (Panten et al.): disguise withholds cues and raises the number of live alternatives, costing the Hick term, which in the karate kumite decomposition is about +116 ms of decision time from one alternative to four plus a distractor; deception supplies misleading cues and installs a wrong prediction. In point-light soccer (PLOS ONE 2014, within-person video task) responses were faster when the cue matched the feint rather than the true direction, in displays reduced to dots with the head as two points, so the effect does not depend on gaze: the observer is not slowed, he is loaded in the wrong direction and must re-select and reverse a movement under way, with braking time at least u/a and extra distance at least u²/2a. Observer cost of a basketball head fake: 422 vs 492 ms and 501 vs 553 ms, about +50 to +70 ms plus more errors (laboratory choice tasks, healthy adults). The cost is base-rate priced: at 20 percent fake frequency 332 vs 349 ms, at 50 percent 340 vs 338 ms, nothing (within-person frequency manipulation). A feint can donate warning: in eye-tracked VR boxing with scripted sequences, first-punch interception probability was 0.353 after a foot feint, 0.197 after an arm feint and 0.252 with no feint, because every feint in that task preceded an attack (Limballe 2022). Boundary: magnitudes from basketball and soccer laboratory tasks on healthy adults and one VR boxing study; no opened study measured a feint's millisecond cost in live striking; the frequency result rests on one group's paradigm; within-person interventions and correlational designs; the braking bound is MODEL.

The physics underneath

A feint exploits a learned prior and consumes it as fuel; a feint thrown half the time conveys zero bits. A deceptive cue moves the mode of the defender's prediction to a false location, so the price is selection plus the mechanical reversal of displacement and momentum. Observation: the millisecond costs, the frequency effect, the VR probabilities. Inference: the bias-versus-variance framing and the braking algebra.

The sources
← Previous axiom: 4.5, Distance buys time only through the strike's actual trajectoryNext axiom: 4.7, The striker pays for a feint out of his own initiation time or out of visible preparation time, and the defender most exposed to it is the one best at reading early cues →Link to this axiom: #ax-4-6
4.7 The striker pays for a feint out of his own initiation time or out of visible preparation time, and the defender most exposed to it is the one best at reading early cuesVERIFIEDA fake costs its producer selection time unless prepared in advance, and it targets the early-cue reading that makes a skilled defender fast, so skill and feint exposure rise together.Open this axiomClose this axiom
The axiom in full

Producing a fake is slower than the true action: initiation 444 ms with a head fake vs 432 ms without, and in an ex-Gaussian re-analysis fake-production costs of 52.0 ms (effort instruction) and 39.7 ms (standard) at zero preparation, collapsing toward zero at 500 to 1000 ms of preparation; practice did not remove the zero-preparation cost (laboratory basketball-pass tasks, healthy adults, within-person). So an improvised feint buys the defender's 50 to 70 ms and pays 40 to 50 ms of its own initiation, near a wash, while a prepared feint pays nothing in initiation but spends at least half a second of preparation inside the defender's most valuable pre-horizon window. Skilled athletes read early kinematics better yet remain susceptible to deception, in some studies more than the less skilled: experienced players anticipated deceptive actions better but were still impaired and anticipated non-deceptive actions less accurately than novices; exaggerated movement shifts perceivers to a cue-based mode with overconfidence (soccer penalty kicks). In badminton, emphasising kinematic (low spatial frequency) information preserved skilled players' accuracy against deceptive shots while non-kinematic emphasis reduced it (Park et al., 12 vs 12). Boundary: producer costs are from basketball-pass laboratory tasks, not strikers; expert-vulnerability findings are from soccer and badminton video tasks; no opened combat-sport study measured a striker's own feint cost; within-person interventions and correlational contrasts; the ledger is MODEL.

The physics underneath

Response-response incompatibility in selection: planning a false and a true action together costs selection time unless preparation separates them. Anticipation and deception vulnerability are one mechanism read from two ends: the early pickup that gives the expert hundreds of milliseconds is the surface a feint writes on. Observation: production costs and expert impairments. Inference: the initiation-versus-legibility ledger.

The sources
← Previous axiom: 4.6, A feint spends the defender's commitment, not his eyesightNext axiom: 4.8, A defense completes to sufficiency, not to completion →Link to this axiom: #ax-4-7
4.8 A defense completes to sufficiency, not to completionVERIFIEDSmall displacements and an already-occupied guard are the only defenses that fit inside the budget, and the margin's spread, not its mean, decides.Open this axiomClose this axiom
The axiom in full

Feasibility is that the defender's reachable set intersects a geometrically safe region before contact, not that a technique finishes: the glove volume and the exposed head volume must never intersect through the dangerous part of the path, and from rest the minimum movement time for a clearance δ at useful acceleration a is √(2δ/a), 100 ms for 10 cm at 20 m/s² (MODEL). The fist travels about 0.65 m in about 0.3 s (Cheraghi 2014) while a head needs only glove radius plus half a head width, about 0.13 to 0.15 m (anthropometric model, no source opened), so a 4 to 5 to 1 advantage in required distance roughly cancels a 6 to 1 deficit in speed; large defenses fail categorically where a part-complete slip may still clear. A guard already in position costs no perception, decision, electromechanical delay or movement time; its price is geometric, a fixed prior over angles a striker re-aims around. International boxers reached peak force in standardized defensive movements in 85 ± 18 ms vs 112 ± 21 ms at national level, but time to peak force is not time to leave the collision corridor (Wang & Korobeynikov 2026). The margin is a distribution: pooling the straight-punch SD of 178 ms (Piorkowski) with the defender's SDs (35 and 28 ms, Petri) gives σ ≈ 184 ms on a mean margin near zero in the favourable cell, and MMA fighters after a Wingate test lengthened reaction time 1.5 percent while its consistency worsened 14.7 percent (within-person pre-post), so fatigue widens the tail rather than shifting the mean. Boundary: the clearance distance is unmeasured; the σ figure pools SDs across studies and punch modalities (the 178 ms SD pools modes and inflates spread); the fatigue datum is one MMA cohort; amateur and international boxers, MMA fighters; MODEL for sufficiency and variance, controlled measurement for the kinematics.

The physics underneath

Collision avoidance is a volumes-and-trajectories problem, so partial credit exists for displacement and none for a technique's completeness; position substitutes geometry for time. A near-zero mean margin with a wide spread makes success a coin flip whose odds move with variance. Observation: the Cheraghi displacement, the Wang peak-force times, the MMA fatigue percentages. Inference: the clearance anthropometry, the σ arithmetic, the sufficiency law.

The sources
← Previous axiom: 4.7, The striker pays for a feint out of his own initiation time or out of visible preparation time, and the defender most exposed to it is the one best at reading early cuesNext axiom: 5.1, Forward pressure and precise placement are two operators on one quantity, the defender's set of exits still executable before interception: pressure closes spatial and temporal exits by geometry, placement closes postural exits by making a specific response punishable →Link to this axiom: #ax-4-8

What governs how forward pressure and precise placement each constrain an opponent's choices across exchanges, and how well does the Rodtang against Tawanchai contrast describe recorded fights?

Pressure against precision · eight axioms

The mechanism under test is pressure against precision: how positioning, initiative and placement constrain an opponent's choices across exchanges and force errors. The easy answer this vector was told not to settle for: “Rodtang pressures and Tawanchai is technical, so the contrast explains their fights”
5.1 Forward pressure and precise placement are two operators on one quantity, the defender's set of exits still executable before interception: pressure closes spatial and temporal exits by geometry, placement closes postural exits by making a specific response punishableVERIFIEDPressure removes a defender's exits by closing space and time; placement removes exits by punishing a chosen response; the shrinking exit set, not output, is what forces errors.Open this axiomClose this axiom
The axiom in full

Define the defender's exit set as every trajectory or defensive action that can be completed before the attacker's earliest interception, bounded by boundary geometry, by the defender's momentum and support leg, and by the time margin between the attacker's arrival and the defender's clearance. Pressure shrinks that set spatially and temporally; placement prunes it posturally by punishing one chosen response until the defender stops choosing it. The one opened intervention on geometry: in 20 male regional and national kickboxers performing nine randomized 2-minute combats, shrinking the area from 6 m x 6 m or 4 m x 4 m to 2 m x 2 m raised block-and-parry counts (d 0.47 and 0.66) and backwards-lean actions and lowered foot defenses (d 1.04 and 0.63) (S3, INTERVENTION, within-person). In 108 international elite amateur boxing bouts ring-centre control stayed independently associated with winning after adjustment (OR 1.07, conditional logistic model, S2, CORRELATIONAL). Being backed up is not being controlled: blocking, checking, clinching and countering stay open when escape closes, and the 2 m x 2 m condition shows exactly that reallocation. Boundary: the area study is sparring in kickboxers, not a professional bout; the exit-set model is a MODEL, not a fitted index; no validated pressure score exists.

The physics underneath

A body cannot occupy every exit before one deadline; movement from rest covers a distance proportional to the square of the time left after reaction, so halving the time left quarters the reachable distance, and a step that covers the defender's likely path removes an exit without a strike. A smaller enclosure reallocates defense from the feet to the arms and torso because foot escapes need distance and time that the enclosure no longer supplies.

The sources
← Previous axiom: 4.8, A defense completes to sufficiency, not to completionNext axiom: 5.2, Output, accuracy and constraint are three different quantities →Link to this axiom: #ax-5-1
5.2 Output, accuracy and constraint are three different quantitiesVERIFIEDIn elite amateur boxing winners differ from losers in accuracy, defensive success and ring-centre control while total punches barely differ, and a fighter can win the volume rounds and lose the constraint battle.Open this axiomClose this axiom
The axiom in full

In 26 elite male amateur boxers at a national championship, winners landed 33 percent of punches against losers' 23 percent and threw 17 percent air punches against 27 percent, while total punches, defensive actions and interaction time were similar (S1, CORRELATIONAL, 2017). In 108 international elite amateur bouts from 2019 to 2025 (216 boxer observations, matched winner-loser pairs), punching accuracy (OR 1.26), defensive success rate (OR 1.18) and ring-centre control (OR 1.07) were each independently associated with winning after adjustment, total punches thrown adding only OR 1.06 per unit (S2, CORRELATIONAL, 2026). On the professional record, Rodtang stormed back and tagged Takeru at will in rounds three and four of their April 2026 kickboxing bout and was dropped four times and stopped at 2:22 of round five (S15, S16, bout record). Population: elite amateur boxers for the measurements, two professional strikers for the case. Boundary: the odds ratios are per-unit and model-dependent; no Muay Thai dataset with a position variable was opened; the case is one bout.

The physics underneath

Volume is a cost line, paid in energy and exposure; it becomes revenue only when it lands on a target or removes an exit. Accuracy and position are the variables that convert work into scoring events, so two fighters with equal counts can sit on opposite sides of the result.

The sources
← Previous axiom: 5.1, Forward pressure and precise placement are two operators on one quantity, the defender's set of exits still executable before interception: pressure closes spatial and temporal exits by geometry, placement closes postural exits by making a specific response punishableNext axiom: 5.3, Pressure forces errors through the clock, not the count →Link to this axiom: #ax-5-2
5.3 Pressure forces errors through the clock, not the countVERIFIEDA defender's read rests on cues before the strike starts, deciding adds time whatever is chosen, and fatigue widens the spread of reaction time more than it lengthens it.Open this axiomClose this axiom
The axiom in full

In 20 national first-class boxers and 20 novices, occluding an attack-or-defense video at 80 ms before action onset lowered recognition accuracy and reaction time relative to 40 ms before and to action start, and mental fatigue from a 45-minute Stroop task lengthened reaction time without changing accuracy (S5, INTERVENTION, mixed design). In 47 Polish K-1 national team kickboxers, having to decide whether to start or change an action in reaction to a stimulus significantly (p < 0.001) extended total task time, and the time spent deciding did not differ by which action was chosen (S6, INTERVENTION, within-person). In 45 MMA fighters an upper-body Wingate test lengthened simple visual reaction time by 1.5 percent and worsened its consistency by 14.7 percent (S4, INTERVENTION, within-person). Together: a pressurer who re-enters before the defender's stance is restored forces the read to begin after initiation, where accuracy is lower, while the decision itself costs time and fatigue makes the response time unreliable. Population: trained boxers, kickboxers and MMA fighters in laboratory tasks; outcome: reaction time, decision time and recognition accuracy; boundary: no study measured a defender under live pressure, and the 14.7 percent is a change in reaction-time variability after upper-body fatigue, not a fight statistic.

The physics underneath

Defensive accuracy depends on information available before the attacker's limb accelerates; shortening the gap moves the defender's sampling window past that point. Decision is a serial cost paid in milliseconds regardless of option, and fatigue adds variance to motor timing, so a required response that must fit a fixed arrival window fails more often even when the mean reaction time barely moves.

The sources
← Previous axiom: 5.2, Output, accuracy and constraint are three different quantitiesNext axiom: 5.4, Fatigue narrows the arsenal before it narrows the count →Link to this axiom: #ax-5-3
5.4 Fatigue narrows the arsenal before it narrows the countVERIFIEDLower-body and trunk fatigue cuts punch force on every punch and delays force development on all but the jab, so a tired fighter keeps throwing while the rotation-dependent weapons fade first.Open this axiomClose this axiom
The axiom in full

In 28 highly-trained male amateur boxers, nine 1-minute rowing bouts significantly reduced peak punch force for jab, cross and both hooks and delayed the time to specific force levels in every punch except the jab, with the larger effect in crosses and hooks that depend on trunk rotation (S7, INTERVENTION, within-subject). In elite boxers under a punching-fatigue protocol, anterior-posterior punch acceleration fell visibly from the sixth burst and jerk from the seventh, consistent with a lower rate of force development (S8, INTERVENTION). In nine professional Muay Thai fighters, 20 maximal roundhouse kicks at one per second or one per three seconds produced no difference in overall neuromuscular fatigue, only a different time course (S9, randomized cross-over). In 26 elite amateur boxers, clinch time and guard drops rose and bouncing fell across rounds (S1, CORRELATIONAL). Population: trained boxers and professional Thai fighters; outcome: punch force, rate of force development, countermovement jump, observed behaviours. Boundary: the fatigue in S7 was induced by rowing, not by an opponent's pressure, and none of these studies measured a fighter who was being pressured.

The physics underneath

Punch force rides ground reaction force up a kinetic chain; when the legs and trunk tire the chain delivers less impulse to the hand, and the punches that need the most rotation lose the most. The count can stay high because the arm-extension jab survives, so output persists while danger falls.

The sources
← Previous axiom: 5.3, Pressure forces errors through the clock, not the countNext axiom: 5.5, The ruleset is the payoff function →Link to this axiom: #ax-5-4
5.5 The ruleset is the payoff functionVERIFIEDUnder ONE and ABC unified Muay Thai scoring a knockdown or damage advantage outranks clean strikes and aggression ranks last, so pressure has no scoring value of its own and must convert within the round that pays for it.Open this axiomClose this axiom
The axiom in full

ONE judges score round by round on the 10-Point-Must system and, when the tally is equal, apply in order knockdowns, damage, number of clean strikes, then aggression and ring generalship; Muay Thai bouts use 4-ounce gloves and clinching, kickboxing 8- or 10-ounce gloves by weight class (S10, ARGUMENT, rules text). The ABC Unified Rules of Muay Thai (July 2024) state that a fighter with an advantage in the number or quality of knockdowns in a round shall never lose that round, then cumulative damage decides, then number and variety of clean techniques, and aggressiveness, control or effective defense only if those fail (S11, ARGUMENT, rules text). The instantiating record: on September 22 2023 Superlek, who missed weight for a 140-pound three-round catchweight, drove Rodtang to the ropes, landed a knee and then an elbow for an eight-count and won a unanimous decision, Rodtang's first striking loss in ONE (S13, bout record). Population: professional strikers under ONE and ABC rules; outcome: round and bout result; boundary: the rules texts are the payoff definition, not a measurement, and one bout instantiates rather than tests the law.

The physics underneath

A scoring system is the function that maps constraint and damage to a result; when knockdowns and damage sit above activity, territory and workload are instrumental and a single conversion event erases rounds of unconverted pressure. Glove mass and clinch legality change which exits exist, so the same pressure has a different payoff under Muay Thai and kickboxing rules.

The sources
← Previous axiom: 5.4, Fatigue narrows the arsenal before it narrows the countNext axiom: 5.6, The Rodtang against Tawanchai contrast describes zero recorded bouts, and each man's official record bounds his label: Rodtang's pressure converted in 80 seconds once and lost twice to a counter-geometry, Tawanchai's precision finished through leg attrition and a ten-strike flurry and was stopped by a pocket counter and a leg kick →Link to this axiom: #ax-5-5
5.6 The Rodtang against Tawanchai contrast describes zero recorded bouts, and each man's official record bounds his label: Rodtang's pressure converted in 80 seconds once and lost twice to a counter-geometry, Tawanchai's precision finished through leg attrition and a ten-strike flurry and was stopped by a pocket counter and a leg kickVERIFIEDA style label is a recurrent opening preference with a counter-geometry on the record, so it predicts the first exchange, not the result; the two named men never met.Open this axiomClose this axiom
The axiom in full

ONE's athlete records list Rodtang at 168 cm with a 134.9 lb limit and Tawanchai at 180 cm with a 154.1 lb limit, two divisions apart, and neither opponent column names the other (S23, S24, official records opened 2026-10-04). Rodtang: a left hook to the liver and a cornered flurry stopped Haggerty at 2:39 of round three in January 2020 (S22); Superlek drove him to the ropes, landed a knee and an elbow knockdown and won a three-round decision in September 2023 (S13); he walked Takeru down and finished him in 80 seconds in March 2025 (S14), against an opponent who had entered with broken ribs and a broken sternum (S16); in April 2026 Takeru blocked, absorbed and countered, dropped him twice with left hooks in round two, took his rounds three and four surge and stopped him at 2:22 of round five after four knockdowns, the first stoppage of Rodtang in 13 years (S15, S16). Tawanchai: a title decision in which Petchmorakot held the clinch midway and Tawanchai surged in round five (S25); repeated kicks to Kiria's blocking arm until it failed 29 seconds into round three (S20); a majority decision over Nattawut, whose low kicks to the lead leg set up straight rights (S21); a 10-piece combination for a third knockdown and a TKO at 1:10 of round two over Superbon (S17); a short left hook and follow-up that stopped him at 1:55 of round three after his teeps had controlled two rounds (S18); four inside leg kicks to the same spot as he closed distance, stopping him in 52 seconds in December 2025 (S19). Evidence class: official bout reports, CORRELATIONAL case records with no comparative coding of output, accuracy or position for either man; the per-minute strike figures circulating for both fighters carry no denominator and are not carried.

The physics underneath

A style label records a recurrent preference in sequence selection; a result records who removed whose exits first under one ruleset against one opponent. Each preferred mechanism has a counter-geometry (a long guard and clinch against entries, a leg kick against the planted lead leg that distance control depends on, a pocket counter when the teep stops), so the label predicts the first exchange, not the outcome.

The sources
← Previous axiom: 5.5, The ruleset is the payoff functionNext axiom: 5.7, Ring position corrupts the record as well as the fight →Link to this axiom: #ax-5-6
5.7 Ring position corrupts the record as well as the fightVERIFIEDJudges' scores drift by ring zone under traditional ropes and rise 1.2 to 1.5 points per zone with transparent ropes, so pressure that pins an opponent in a badly seen zone inflates its own scorecard and the manual counts built on it.Open this axiomClose this axiom
The axiom in full

Twenty-five certified international boxing referees (20 scoring judges and 5 supervisory officials) judged the same athletes in traditional and transparent roped rings across five predefined ring zones; with transparent ropes they awarded 1.2 to 1.5 more points per scoring zone and scored more consistently in every zone (p < .05, d 0.36 to 0.50), while under traditional ropes one zone's scores diverged from multi-angle video review (S12, INTERVENTION on the viewing environment, 2025). The winner-loser datasets that carry accuracy and ring-centre control are hand-coded from competition video (S1, S2), so a position variable and a scoring variable share the same sightline. Population: international boxing officials; outcome: points awarded per zone; boundary: one study, boxing not Muay Thai, no fighter-level test of whether cornering wins rounds through judges' eyes rather than through damage.

The physics underneath

A judge scores what the sightline delivers; ropes and corners occlude the pinned fighter's work more than the pinning fighter's, so position enters the result twice, once through the exits it removes and once through the record of who landed.

The sources
← Previous axiom: 5.6, The Rodtang against Tawanchai contrast describes zero recorded bouts, and each man's official record bounds his label: Rodtang's pressure converted in 80 seconds once and lost twice to a counter-geometry, Tawanchai's precision finished through leg attrition and a ten-strike flurry and was stopped by a pocket counter and a leg kickNext axiom: 5.8, Composure under pressure is a defender's instrument, measurable as defensive success and held exits, not a pressurer's trait: defensive success rate predicts winning independently of output, and on the record the fighter receiving pressure kept his exits open and converted →Link to this axiom: #ax-5-7
5.8 Composure under pressure is a defender's instrument, measurable as defensive success and held exits, not a pressurer's trait: defensive success rate predicts winning independently of output, and on the record the fighter receiving pressure kept his exits open and convertedVERIFIEDA defender who keeps vision, base and exit timing holds his exit set open while the attacker spends exposure to close it, so the attacker's commitment becomes the defender's counter.Open this axiomClose this axiom
The axiom in full

In 108 international elite amateur bouts, defensive success rate was independently associated with winning (OR 1.18) alongside accuracy and ring-centre control, after adjustment for total punches (S2, CORRELATIONAL, 2026). In 20 elite boxers and 20 novices, recognition accuracy depended on seeing cues before action onset while fatigue moved reaction time, not accuracy (S5, INTERVENTION), so the operational content of composure is preserved information pickup under a shortened window. On the record, Takeru blocked, absorbed and countered Rodtang's entries, dropped him twice in round two, took a two-round surge and stopped him in round five (S15, S16), and Superlek gave ground to the ropes and converted a knee and an elbow into the knockdown that won a decision (S13). Population: elite amateur boxers for the measurement, two professional strikers for the cases; outcome: defensive success rate, recognition accuracy, bout result; boundary: no study measured 'composure' as a construct, the OR is per-unit of a coded rate, and the case attribution rests on two bouts.

The physics underneath

A defender who keeps vision, balance, guard and exit timing holds his exit set open while the attacker spends energy and exposure to close it; the attacker's commitments are themselves cues, so composure converts the pressurer's loan into the defender's counter. The quantity that scores is the defensive action that succeeds and the counter it enables, never the posture of looking calm.

The sources
← Previous axiom: 5.7, Ring position corrupts the record as well as the fightNext axiom: 6.1, Sparring is the high-count, low-magnitude arm of head impact exposure and competition the low-count, high-magnitude arm; the dose is a right-skewed distribution whose thin tail, not its mean, reaches the magnitudes tied to acute injury →Link to this axiom: #ax-5-8

How does head impact exposure in sparring and competition become acute injury, recovery and longer-term brain change, and what can current measurements establish about cumulative risk?

The brain under head impact · eight axioms

The mechanism under test is the brain under repeated head impact: exposure in sparring and competition, acute concussion, recovery and what cumulative load does over time. The easy answer this vector was told not to settle for: “A few sparring rounds a week are harmless if you wear headgear”
6.1 Sparring is the high-count, low-magnitude arm of head impact exposure and competition the low-count, high-magnitude arm; the dose is a right-skewed distribution whose thin tail, not its mean, reaches the magnitudes tied to acute injuryVERIFIEDRepeated low-energy momentum transfers in sparring accumulate the exposure integral while rarer full-intent competition blows produce the diagnosed injuries, through a right-skewed magnitude tail.Open this axiomClose this axiom
The axiom in full

Population: amateur boxers and MMA athletes (23 athletes, 896 video-verified events above 10 g), 13 MMA athletes, 22 novice headgeared military cadets, one elite Muay Thai athlete. Outcome: verified head impact counts per session and peak linear and angular acceleration from instrumented mouthguards. Evidence class: CORRELATIONAL field measurement. Male boxers took 18.4 +/- 13.1 verified events per sparring session with median peaks 19.7 g and 1,534 rad/s2 against 12.3 +/- 10.4 per competition bout at 17.1 g and 1,642 rad/s2; male MMA athletes 19.0 +/- 20.5 per sparring session at 17.5 g and 1,766 rad/s2 against 10.0 +/- 4.2 per bout at 37.9 g and 3,773 rad/s2 (medians of event peaks). In 13 MMA athletes sparring produced 298 events over 19 sessions (median 28.4 g) and competition 153 events over 11 bouts (36.8 g); 5 of 11 bouts ended in diagnosed concussion, none of 19 sparring sessions did. Headgeared cadets averaged 16.5 g with 640 of 809 events (79.1 percent) between 10 and 20 g and a 70.8 g maximum; the Muay Thai case logged 590 events, nearly all below 66 g, with maxima of 162.2 g and 26,850 rad/s2. Boundary: amateur and novice populations, unstructured 5 to 20 minute sparring, competition samples of 2 to 11 bouts, no professional Thai-stadium data, the Muay Thai case unverified by video and not poolable with 10 g-thresholded studies.

The physics underneath

Each contact transfers momentum and angular momentum to the head; sparring repeats many low-energy transfers while competition delivers fewer transfers at full intent, so the exposure integral is dominated by training and the recorded injuries by competition. The event-magnitude distribution is right-skewed with a thin tail, so the mean peak describes almost none of the loading that matters.

The sources
← Previous axiom: 5.8, Composure under pressure is a defender's instrument, measurable as defensive success and held exits, not a pressurer's trait: defensive success rate predicts winning independently of output, and on the record the fighter receiving pressure kept his exits open and convertedNext axiom: 6.2, A head impact count is an instrument-and-threshold-defined event rate, not a physical quantity: trigger level, jaw coupling, quality filtering and device transfer function set the number, and the filtering error runs one way, discarding the largest events →Link to this axiom: #ax-6-1
6.2 A head impact count is an instrument-and-threshold-defined event rate, not a physical quantity: trigger level, jaw coupling, quality filtering and device transfer function set the number, and the filtering error runs one way, discarding the largest eventsVERIFIEDMouthguard counts depend on trigger, jaw coupling and quality filters, and the worst-coupled recordings are the biggest blows, so measured exposure is systematically undercounted at the tail.Open this axiomClose this axiom
The axiom in full

Population: amateur boxers wearing mouthguard, skin-patch and headgear-patch sensors at once; cadaver heads under direct jaw impacts of the boxing kind; the Muay Thai single-athlete case; the combat-sport device literature. Outcome: share of recordings meeting quality criteria, validation slopes against a reference sensor, false-positive and false-negative mechanisms. Evidence class: INTERVENTION for laboratory validations, CORRELATIONAL for the field quality study, ARGUMENT for the conceptual paper, review for the 22-study synthesis. In boxing sparring 53 percent of mouthguard recordings, 20 percent of skin-patch and 26 percent of headgear-patch recordings met quality criteria and the low-quality recordings carried higher peak kinematics (p < 0.001). Under direct jaw impacts one mouthguard underpredicted (slopes 0.72 to 1.04) while another overpredicted every channel (slopes 1.4 to 8.7). Triggers of 5, 8 and 10 g across studies make counts incomparable; the Muay Thai case had no raw signals and no video verification. A sensor acceleration event approximates an in-vivo head acceleration event; false positives are mostly removable, false negatives are not, so research likely undercounts. Boundary: one sparring protocol for the quality shares; cadaver, not live, slopes; the undercount direction argued from mechanism, not measured in strikers.

The physics underneath

A jaw-mounted sensor measures its own acceleration, which equals skull acceleration only when coupling holds; mandible motion, clenching, trigger thresholds and proprietary filters all enter the count, and because the biggest blows most disturb coupling they are the ones most often discarded, biasing exposure downward and dose-response slopes toward null.

The sources
← Previous axiom: 6.1, Sparring is the high-count, low-magnitude arm of head impact exposure and competition the low-count, high-magnitude arm; the dose is a right-skewed distribution whose thin tail, not its mean, reaches the magnitudes tied to acute injuryNext axiom: 6.3, One ordinary sparring session changes the brain below the symptom floor within an hour, and one bout raises axonal and glial injury markers in the spinal fluid that remain elevated past a 14-day rest, in boxers with no concussion →Link to this axiom: #ax-6-2
6.3 One ordinary sparring session changes the brain below the symptom floor within an hour, and one bout raises axonal and glial injury markers in the spinal fluid that remain elevated past a 14-day rest, in boxers with no concussionVERIFIEDRotational shear stretches axons and astrocytes so protein leakage and cortical inhibition follow a sparring session for hours to weeks without any symptom, measured in amateur boxers.Open this axiomClose this axiom
The axiom in full

Population: 20 amateur boxers against 20 mock-sparring controls; 30 Olympic-style amateur boxers against 25 controls; 14 amateur boxers; 4 elite amateur Muay Thai athletes. Outcome: corticomotor inhibition and motor-unit recruitment (TMS and EMG), memory score, CSF and serum neurofilament light and GFAP. Evidence class: INTERVENTION (sparring versus mock sparring), CORRELATIONAL within-person repeats, a four-athlete feasibility series, cell and animal mechanism. One hour after three 3-minute sparring rounds the boxers showed increased corticomotor inhibition, altered motor-unit recruitment and poorer memory, back at baseline at 24 hours in that protocol. 1 to 6 days after a bout CSF NfL averaged 532 ng/L against 135 ng/L in controls and GFAP 496 against 247 ng/L; 77 percent of boxers had elevated NfL and nearly half remained elevated after at least 14 days of rest; none had concussion symptoms. Serum NfL was elevated 7 to 10 days after a bout against the three-month rested state, higher after more than 15 head hits. In four Muay Thai athletes the group-level GFAP-to-impacts association was negative. Mechanism: stretched axons break microtubules and degenerate while straightening back to normal appearance; injured tissue runs ionic flux and hyperglycolysis then depressed metabolism. Inference carried as inference: because NfL peaks over days and clears over weeks, a fighter sparring two or three times a week is sampled faster than the marker clears and the marker becomes a rolling integral. Boundary: amateur boxers, single sessions or bouts; the 24-hour return is for three rounds only; marker elevation is a response compatible with injury, not a neuron count or a disease.

The physics underneath

Shear strain from head rotation stretches axons and disturbs astrocytes; the mechanical event lasts milliseconds, the cellular response (microtubule failure, ionic flux, metabolic depression, protein leakage) runs for days to weeks, and none of it requires a symptom to occur.

The sources
← Previous axiom: 6.2, A head impact count is an instrument-and-threshold-defined event rate, not a physical quantity: trigger level, jaw coupling, quality filtering and device transfer function set the number, and the filtering error runs one way, discarding the largest eventsNext axiom: 6.4, Symptom reports are not a dose meter →Link to this axiom: #ax-6-3
6.4 Symptom reports are not a dose meterVERIFIEDAcross 16 boxing sparring sessions symptom change did not correlate with measured head kinematics, the group mean did not move, symptoms were lower 48 hours after sparring than before, and a quarter of sessions still produced a large individual rise.Open this axiomClose this axiom
The axiom in full

Population: 7 competitive boxers over 16 sparring sessions with instrumented mouthguards and SCAT5 scales; 30 amateur boxers with symptom-free biomarker elevation; 631 American football brain donors as the shape of the law only. Outcome: total symptom severity change against session exposure; CTE presence against reported concussions versus estimated cumulative impacts. Evidence class: CORRELATIONAL (within-person repeated field sessions; brain-bank cohort). No group pre-to-post symptom difference (median change -1, p = 0.85); symptoms lower at 48 hours than before sparring; 4 of 16 sessions (25 percent) from 3 participants rose above 5 points (rises +5 to +17, mean 12 +/- 5); no correlation between symptom change and kinematics across the 16 athlete-sessions. In football donors reported symptomatic concussions were not associated with CTE presence or severity while years played and acceleration-weighted cumulative exposure were. Boundary: seven boxers and hypothesis-generating by the authors' own words; the donor series is selected and helmeted and transfers only its shape; athlete symptom reports carry a documented return-to-sport incentive.

The physics underneath

Symptoms are a reported, incentive-laden output of a system whose biological state changes without them; averaging across sessions cancels large individual rises, and practice and reporting effects can move the score in the protective direction after exposure, so symptom change is decoupled from mechanical dose at the session level.

The sources
← Previous axiom: 6.3, One ordinary sparring session changes the brain below the symptom floor within an hour, and one bout raises axonal and glial injury markers in the spinal fluid that remain elevated past a 14-day rest, in boxers with no concussionNext axiom: 6.5, No single-impact concussion threshold exists for strikers →Link to this axiom: #ax-6-4
6.5 No single-impact concussion threshold exists for strikersVERIFIEDA diagnosed-concussion bout's largest event peaked at 50.5 g while a non-concussed bout's reached 105.7 g, published 50 percent-risk values span 65 to 192 g by dataset, and sub-failure axonal strain accumulates across impacts.Open this axiomClose this axiom
The axiom in full

Population: 13 MMA athletes and the 5 concussive events simulated in finite-element brain models; football impact reconstructions and sensor datasets re-analysed in a 2025 preprint; a 2026 axonal strain model against above-8-g field impacts; the 22-study combat-sport review. Outcome: peak linear and angular acceleration at diagnosed concussion, modelled corpus-callosum strain and shear stress, 50 percent risk thresholds, modelled fibre strain at failure. Evidence class: CORRELATIONAL for bout-level kinematics, MODEL for the simulations. A concussed bout's highest-angular event had a linear peak of 50.5 g; a non-concussed bout contained a 105.7 g event; the injuring impact could not be identified. Concussion was associated with greater modelled callosal strain across 5 events. The 50 percent risk for peak linear acceleration was estimated at about 100 g (95 percent CI 89 to 110) against earlier 85 g (31 NFL reconstructions), 65 g (40 Australian football) and 192 g (helmet sensors). Axonal failure is modelled from about 6 percent fibre strain, with about 31.4 percent of above-trigger field impacts modelled to cause no callosal damage and the balance some sub-failure damage. The review states thresholds must not be used without clinical assessment. Boundary: finite-element and strain results are models; the 2025 and 2026 items were read at abstract level; threshold datasets are football, cited only for the spread.

The physics underneath

Injury depends on the whole pulse (rotation, direction, duration) and on tissue state set by prior impacts, so peak linear acceleration is a weak summary and any g-threshold is a property of the dataset and instrument that produced it; sub-failure strain leaves damage that lowers the next impact's margin.

The sources
← Previous axiom: 6.4, Symptom reports are not a dose meterNext axiom: 6.6, Over years of continued fighting the thalamus and corpus callosum shrink within the same person faster than in controls, cumulative exposure tracks slower processing speed, and sparring dose has a measured cognitive gradient in professional boxers, while a surrogate change is not a demonstrated impairment and a null on a crude exposure proxy is a statement about the proxy →Link to this axiom: #ax-6-5
6.6 Over years of continued fighting the thalamus and corpus callosum shrink within the same person faster than in controls, cumulative exposure tracks slower processing speed, and sparring dose has a measured cognitive gradient in professional boxers, while a surrogate change is not a demonstrated impairment and a null on a crude exposure proxy is a statement about the proxyVERIFIEDRepeated rotational shear at the midline and deep nuclei drives within-person thalamic and callosal volume loss and slower processing speed over years of fighting, bounded by surrogate-versus-impairment and survivor-selection limits.Open this axiomClose this axiom
The axiom in full

Population: the Professional Fighters Brain Health Study (overlapping papers, never summed): 224 professional combatants and 22 controls cross-sectionally; 204 participants followed about 2.6 years; 140 active boxers, 211 active MMA fighters, 69 retired boxers, 52 controls; 94 active MMA fighters in a sparring-rounds cross-section. Outside it: 237 professional boxers at commission exams; 9 former professional boxers against 15 Alzheimer's and 14 subjective-cognitive-decline patients. Outcome: regional MRI volumes, computerised cognitive scores (processing speed, Trail-Making), plasma NfL and GFAP, Symbol Digit Modalities and balance. Evidence class: CORRELATIONAL (longitudinal within-person; cross-sectional; small comparative series). Active boxers lost left thalamus at 102.3 mm3 per year (p = 0.0004), mid-anterior corpus callosum at 10.2 (p = 0.018) and central corpus callosum at 16.5 mm3 per year (p < 0.0001) versus controls; higher baseline NfL predicted decline; caudate decliners worsened on Trail-Making. More fights, years or Fight Exposure Score tracked lower thalamic and caudate volume and slower processing speed, about 0.8 percent lower caudate volume per score point. Rising GFAP tracked volume loss and worsening cognition, chiefly in retired boxers. Self-reported cumulative sparring tracked worse Symbol Digit and balance scores in 237 professional boxers with a spline cutpoint near a Composite Sparring Index of 150. Nine former boxers had higher GFAP, p-tau181 and p-tau217 and lower volumes in every prespecified region with no difference in neuropsychological scores. In 94 active MMA fighters self-reported weekly sparring rounds were not associated with smaller volumes and were associated with a larger caudate after adjustment. Boundary: professional boxers and MMA fighters with careers underway at baseline, convenience recruitment, 22 controls in the cross-section, overlapping participants, no clinicopathological confirmation, no professional Thai fighters; exposure measured by fight counts and recalled sparring, never instrumented career integrals.

The physics underneath

Rotational shear concentrates at the midline callosum and deep grey nuclei, so repeated exposure drives measurable within-person volume loss there and slows the relay functions those structures serve; but imaging and plasma markers are surrogates, skill training enlarges the same motor-learning structures injury shrinks, and active-fighter cross-sections are survivor-selected, so marker change and null proxies each bound what the gradient can claim.

The sources
← Previous axiom: 6.5, No single-impact concussion threshold exists for strikersNext axiom: 6.7, Stopping both fighting and sparring for at least two years is followed by improved cognitive scores and falling NfL against continued decline in matched active fighters, while GFAP keeps rising in retired boxers and tracks continued volume loss, so the exposure-associated state has a reversible component and a component that does not reverse on cessation alone →Link to this axiom: #ax-6-6
6.7 Stopping both fighting and sparring for at least two years is followed by improved cognitive scores and falling NfL against continued decline in matched active fighters, while GFAP keeps rising in retired boxers and tracks continued volume loss, so the exposure-associated state has a reversible component and a component that does not reverse on cessation aloneVERIFIEDAxonal marker NfL falls and cognition improves when fighting and sparring stop, while astroglial GFAP and limbic volume loss continue in older retired boxers, separating reversible from persisting processes.Open this axiomClose this axiom
The axiom in full

Population: 45 male professional fighters who stopped sanctioned fighting and sparring for at least two years against 45 matched active fighters; 69 retired boxers in the 2023 PFBHS biomarker analysis; 9 former professional boxers with 5 followed about seven years. Outcome: computerised cognitive domains, plasma NfL and GFAP, regional volumes. Evidence class: CORRELATIONAL (non-randomised longitudinal; the transition was the fighter's own). The transitioned group improved on cognitive domains with decreasing NfL; the active group declined with stable NfL. Baseline GFAP was highest in retired boxers (mean age 49) and its rise tracked regional volume loss and worsening cognition in that group. In 5 former boxers plasma NfL and p-tau181 rose over about seven years without evident clinical decline. Boundary: stopping was self-selected and may travel with other health changes; the transitioned and retired groups differ in age, era and prior history, so they are two populations, not a contradiction; cessation is not shown to repair prior injury or remove later risk; the 'self-sustaining neuroinflammatory cascade' is an explanation, not a finding.

The physics underneath

The axonal-injury signal (NfL) follows ongoing mechanical exposure and falls when exposure stops, while an astroglial signal (GFAP) and limbic volume loss in older retired boxers continue, consistent with an exposure-driven process that halts and a slower process set in motion earlier that does not.

The sources
← Previous axiom: 6.6, Over years of continued fighting the thalamus and corpus callosum shrink within the same person faster than in controls, cumulative exposure tracks slower processing speed, and sparring dose has a measured cognitive gradient in professional boxers, while a surrogate change is not a demonstrated impairment and a null on a crude exposure proxy is a statement about the proxyNext axiom: 6.8, Headgear attenuates headform acceleration in the laboratory and does not establish a safe sparring dose: headgeared novices still produced football-band impacts, the field evidence against headgear rests on a referee-stoppage surrogate from an unpublished internal study, and no study measures headgear's effect on cumulative sparring exposure →Link to this axiom: #ax-6-7
6.8 Headgear attenuates headform acceleration in the laboratory and does not establish a safe sparring dose: headgeared novices still produced football-band impacts, the field evidence against headgear rests on a referee-stoppage surrogate from an unpublished internal study, and no study measures headgear's effect on cumulative sparring exposureVERIFIEDPadding lengthens a blow's pulse and lowers its peak without zeroing momentum transfer, so headgear attenuates laboratory loading while its effect on cumulative sparring exposure stays unmeasured.Open this axiomClose this axiom
The axiom in full

Population: boxing headguards on instrumented headforms in punch-machine tests; AIBA tournament rounds with and without head guards; headgeared, 16 to 20 oz-gloved military cadets; the 22-study combat-sport review. Outcome: peak linear and angular headform acceleration; referee stoppages for head blows per round and facial cuts; mouthguard peaks in novice sparring. Evidence class: INTERVENTION (laboratory), CORRELATIONAL (observational tournaments and field sparring), review. Tested headguards reduced both linear and angular headform acceleration in the laboratory, so the claim that padding cannot touch rotation is false there. In AIBA tournaments removing head guards was associated with fewer referee stoppages for head blows (the authors' surrogate for concussion) and more facial cuts; the quoted 0.38 percent of 7,352 rounds with guards against 0.17 percent of 7,545 without came from an unpublished internal study, and the earlier mandated-headgear era bundled computer scoring and rule changes. Headgeared novices averaged 16.5 g with 79.1 percent of events between 10 and 20 g and a 70.8 g maximum, compared by their authors to Service Academy football. The review calls the role of headgear in striking sports contentious. Boundary: punch machines, one tournament series, novice sparring; no opened study measures headgear against mouthguard-recorded cumulative exposure across a training block.

The physics underneath

Foam lengthens the deceleration pulse and lowers peak acceleration of a given blow without zeroing the momentum transferred; a competition stoppage count is an officiating surrogate measured where headgear is least worn, and if protection changes what a fighter accepts, equipment and tactics are confounded in the same endpoint.

The sources
← Previous axiom: 6.7, Stopping both fighting and sparring for at least two years is followed by improved cognitive scores and falling NfL against continued decline in matched active fighters, while GFAP keeps rising in retired boxers and tracks continued volume loss, so the exposure-associated state has a reversible component and a component that does not reverse on cessation aloneNext axiom: 7.1, A learned striking skill has no single decay constant →Link to this axiom: #ax-6-8

What governs which parts of a striking skill decay under reduced practice or a full interruption, which persist, and how fast each returns with renewed practice?

Skill decay and return · eight axioms

The mechanism under test is decay and return of a combat skill under reduced practice or complete interruption: what stays, what goes and how fast it comes back. The easy answer this vector was told not to settle for: “It is like riding a bike; you never forget”
7.1 A learned striking skill has no single decay constantVERIFIEDCoordination structure persists for years while output magnitude, conditioning and calibration drift on separate clocks.Open this axiomClose this axiom
The axiom in full

In adults who learned a complex coordination to a stable pattern (S3: 5 adults, 390 minutes of ski-simulator practice; S4: 16 adults, 225 minutes of bimanual 3:2 practice; S5: a polyrhythm case series), the learned coordination mode was expressed on the first retention trial after 3 months (S4), 8 years (S5) and about 10 years (S3), while in the same tests movement amplitude and frequency had declined (S3) and the slowest-learned variable, intermanual crosstalk suppression, was the only one not maintained at 8 years (S5). Over the same kind of interval, supporting conditioning in elite taekwondo fell (S7: estimated aerobic power down about 2.43 percent of baseline in 16 black belts after 8 weeks of cessation; S6: endurance fell more in the cessation arm than the ~7 h/week arm over 12 weeks) and tendon stiffness and muscle cross-section returned to pre-training values at one and two months of detraining while strength and neural activation held (S13). Boundary: the retention studies are laboratory coordination tasks with no opponent, threat or impact; S3's cohort skied occasionally and had a 5-month retention test, so '10 years' is not zero related exposure; the slowest-learned-goes-first finding rests on one case series (S5) and is carried as a bound, not a law. Evidence classes: INTERVENTION within-person (S3, S4, S5, S6, S7, S13).

The physics underneath

The components rest on tissues with different turnover: mitochondrial enzymes and plasma volume turn over in days to weeks (S19), tendon collagen in months (S13), and the cortical synaptic structure encoding a practiced movement is selectively stabilized and persists after training stops (S35, mice; S36, songbirds). The observed skill is the envelope of these curves, so time away cannot be read as one number; what stopped and what continued must be named per component.

The sources
← Previous axiom: 6.8, Headgear attenuates headform acceleration in the laboratory and does not establish a safe sparring dose: headgeared novices still produced football-band impacts, the field evidence against headgear rests on a referee-stoppage surrogate from an unpublished internal study, and no study measures headgear's effect on cumulative sparring exposureNext axiom: 7.2, Procedural skill loss grows with the nonuse interval and is steepest for accuracy-based, cognitive and artificial tasks; speed-based, physical and natural tasks lose least, and a retrieval context matching training reduces apparent loss →Link to this axiom: #ax-7-1
7.2 Procedural skill loss grows with the nonuse interval and is steepest for accuracy-based, cognitive and artificial tasks; speed-based, physical and natural tasks lose least, and a retrieval context matching training reduces apparent lossVERIFIEDNonuse erodes accuracy-based procedural performance about one-third faster per month than speed-based performance because accuracy rides on volatile calibration while speed rides on stabilized coordination.Open this axiomClose this axiom
The axiom in full

Across 1,344 effect sizes from 457 reports of procedural skills with motor components in adults (S1, random-effects metaregression with retention interval as a continuous predictor), the acquisition-to-retention performance difference grew by 0.08 standard deviations per month of nonuse for accuracy measures and 0.06 SD per month for speed and mixed measures, so half of the initial acquisition gain was lost after about 6.5 months (accuracy), 13 months (speed) and 11 months (mixed). Across 189 data points from 53 articles (S2), loss ran from d = -0.01 immediately after training to d = -1.4 after more than 365 days of nonuse, physical, natural and speed-based tasks were less susceptible than cognitive, artificial and accuracy-based tasks, and retention tests run under conditions matching training showed less loss. Boundary: both estimates pool heterogeneous laboratory and occupational tasks, not striking; the denominators are the initial acquisition gain (S1) and the acquisition-to-retention difference in SD units (S2), never a share of an expert's total ability; S1's authors name relearning contamination of retention scores and speed-accuracy trade-offs as live threats. Evidence class: CORRELATIONAL (meta-analyses of retention studies).

The physics underneath

Accuracy demands a current calibration of magnitudes and timings against a target, an error-driven and volatile state; speed on a natural continuous task reflects a stabilized coordination structure (A1). Retrieval is context-gated: the stored skill includes the cues present at training, so a changed environment adds a retrieval failure that reads as decay (S2, converging with S28).

The sources
← Previous axiom: 7.1, A learned striking skill has no single decay constantNext axiom: 7.3, Reduced practice and complete cessation are different exposures →Link to this axiom: #ax-7-2
7.3 Reduced practice and complete cessation are different exposuresVERIFIEDMagnitude adaptations survive on one-third to one-ninth of the building dose when intensity is held, while rate-type adaptations (ballistic power) fall at the same reduced frequencies.Open this axiomClose this axiom
The axiom in full

In 70 younger and older adults who first trained resistance exercise three days a week for 16 weeks (S8, randomized), 32 weeks at one-third or one-ninth of that weekly dose with loading intensity preserved maintained hypertrophy in the younger adults and not the older, and maximal strength was largely retained even in the cessation arm; in trained adults (S9) isometric lumbar strength held over 12 weeks at a frequency as low as once every four weeks with per-session intensity and volume maintained, while the detraining group lost 55 percent of its gain; after 12 weeks of concurrent training (S10) one session every two weeks for three months preserved 90 to 95 percent of strength and aerobic-power gains and 72 percent of the muscle-mass gain. In the companion ballistic design (S11) both reduced frequencies produced significant losses of power performance and one session a week was needed to hold it. In elite taekwondo (S6, 11 athletes) about 7 h/week of isolation training held a light-cued kicking response over 12 weeks while endurance fell more in the no-training arm. Boundary: resistance- and aerobic-trained adults plus one small elite taekwondo cohort; no opened study measures the minimum live-partner dose that holds opponent timing, distance calibration or composure (UNVERIFIED after search by two legs); S6's response measure bundles reaction and movement time and keeps the best of three kicks, so it is not cold first-trial retention, and small groups mean non-significance is not equivalence. Evidence class: INTERVENTION (randomized dose arms).

The physics underneath

A magnitude adaptation (how much force, how much oxygen) is a tissue stock a periodic high-intensity signal can keep topped up; a rate adaptation (force per unit time) depends on neural timing and synchronous motor-unit recruitment that decays between exposures and needs frequent re-tuning. A strike is impulse delivered into a closing window, a rate quantity, so the components reduced practice protects best are the ones a striker needs least and the one it fails to hold at low frequency, power, is the one the strike is made of.

The sources
← Previous axiom: 7.2, Procedural skill loss grows with the nonuse interval and is steepest for accuracy-based, cognitive and artificial tasks; speed-based, physical and natural tasks lose least, and a retrieval context matching training reduces apparent lossNext axiom: 7.4, Under full cessation, repeat-effort capacity and the cardiovascular stock fall first (days to three weeks) while single-effort strength, speed and even power hold or rise for about four weeks, then fall by eight to twelve weeks →Link to this axiom: #ax-7-3
7.4 Under full cessation, repeat-effort capacity and the cardiovascular stock fall first (days to three weeks) while single-effort strength, speed and even power hold or rise for about four weeks, then fall by eight to twelve weeksVERIFIEDPlasma volume and mitochondrial enzymes turn over in days, so repeat-effort capacity drains within a week of cessation while neural drive and contractile protein hold single-effort force for a month.Open this axiomClose this axiom
The axiom in full

In 11 well-trained soccer players after 7 days of cessation (S15, within-person), repeated-sprint total time worsened from 45.7 to 48.0 s and the fatigue index from 5.8 to 7.8 percent while 5 m and 10 m sprint times improved (1.05 to 1.02 s; 1.79 to 1.74 s) and agility, coordination, Yo-Yo IR2 and knee peak torque did not change. In 18 professional players after 4 weeks (S16), jump height and sprint times were unchanged and half-squat peak force and mean power at 50 and 100 percent body mass rose (effect sizes 0.55 to 0.82). In endurance-trained adults (S19, tabulated cessation studies), VO2max fell about 7 percent of baseline by 12 to 21 days, driven largely by plasma-volume and stroke-volume loss, and to about 16 to 20 percent by 56 to 84 days. By 8 weeks elite taekwondo black belts lost about 2.43 percent of estimated aerobic power (S7); by 12 weeks 14 elite swimmers lost swimming speed and power, lactate-threshold speed, jump height and upper-limb power with no change in body weight or composition (S17); after 24 weeks 38 older adults lost 15.5 to 17.8 percent of trained power and 16.5 to 17.1 percent of strength, regained with comparable accrual in 12 weeks of retraining (S18); after a 10-week interruption, strength and size losses were regained mostly within the first five weeks of retraining (S12). Boundary: soccer, swimming, endurance and older-adult populations plus one elite taekwondo cohort; the early rise in single-effort output is an end-of-season fatigue release and does not transfer to a mid-year layoff without the same accumulated fatigue; body composition is not a readiness proxy (S17). Evidence class: INTERVENTION within-person, plus a CORRELATIONAL review table (S19).

The physics underneath

Plasma volume and mitochondrial enzyme activity turn over in days (S19), so the capacity to repeat efforts and recover between them is the first stock to drain; maximal strength and single-effort speed rest on neural drive and contractile protein that hold for weeks, and a rested nervous system can briefly express more of them. Over two to three months contractile and elastic capacity follow, without visible atrophy.

The sources
← Previous axiom: 7.3, Reduced practice and complete cessation are different exposuresNext axiom: 7.5, Force-producing capacity outlasts and returns ahead of force-absorbing tissue: strength and neural activation hold through detraining while tendon stiffness and muscle cross-section return to baseline at one and two months →Link to this axiom: #ax-7-4
7.5 Force-producing capacity outlasts and returns ahead of force-absorbing tissue: strength and neural activation hold through detraining while tendon stiffness and muscle cross-section return to baseline at one and two monthsVERIFIEDTendon's slow turnover makes stiffness fall within two months of detraining while strength and neural drive hold, so force production returns ahead of the tissue that must absorb it.Open this axiomClose this axiom
The axiom in full

In young adults who trained isometrically for 3 months and then stopped for 3 months (S13, within-person), maximal strength and neural activation did not change during detraining while muscle cross-sectional area and tendon stiffness fell back to pre-training values at one and two months respectively; the authors conclude tendon and muscle morphology adapt more slowly to training and de-adapt faster than muscle function. In collegiate overhead athletes through the COVID cessation (S14, natural experiment), shoulder and hip range of motion changed more than strength, in the injury-associated direction, and resumed team training reversed some but not all of it. In 20 elite Greco-Roman wrestlers (S38), 14 weeks of restricted activity was followed by 8 weeks of retraining in which the Specific Wrestling Fitness Test rose from 31.40 to 37.40 while heart-rate variability, depressed in cessation, recovered through retraining. Retraining records converge on weeks for the force side: strength and size regained mostly within five weeks after a 10-week interruption (S12), and support staff of elite Japanese team and combat athletes reported about two months to regain strength and power after the pandemic interruption (S37, interview data, ARGUMENT class, weak signal only). Boundary: S13 is isometric knee-extension training in young adults, not impact loading; no opened study measures hand, wrist, shin or neck tissue tolerance across a striker's layoff (UNVERIFIED); S38 measures a wrestling fitness test, not tissue. Evidence class: INTERVENTION within-person (S13, S38) and CORRELATIONAL natural experiment (S14).

The physics underneath

Tendon has a far lower cell-to-dry-mass ratio, vascularity and turnover than muscle, so its stiffness lags muscle function on the way up and falls faster on the way down (S13). The force a strike delivers is also the reaction force the hand, shin and neck absorb, so a returning striker can regain the capacity to generate impulse before the tissue that receives it has re-stiffened; the hazard window is the gap between the two clocks, a structural consequence of differential time constants.

The sources
← Previous axiom: 7.4, Under full cessation, repeat-effort capacity and the cardiovascular stock fall first (days to three weeks) while single-effort strength, speed and even power hold or rise for about four weeks, then fall by eight to twelve weeksNext axiom: 7.6, Savings on relearning is faster re-access of a strategy plus heightened sensitivity to familiar errors, dissociable from long-term retention; it requires prior error and depends on how practice stopped and how exposure was structured, not on total minutes →Link to this axiom: #ax-7-5
7.6 Savings on relearning is faster re-access of a strategy plus heightened sensitivity to familiar errors, dissociable from long-term retention; it requires prior error and depends on how practice stopped and how exposure was structured, not on total minutesVERIFIEDPrior error history leaves a fast re-engaging adaptive process and a recalled strategy, so relearning is quicker than first learning even when the tuned output itself was not retained.Open this axiomClose this axiom
The axiom in full

In adults adapting to a visuomotor rotation (S20, five experiments) savings reflected improved action selection, recalling an appropriate aiming strategy, not faster implicit adaptation; savings required a prior history of error rather than repetition of successful actions (S21), and experiencing an error durably raised sensitivity to that same error (S22). Savings and long-term memory doubly dissociated (S23): the fast, temporally volatile component drove faster relearning but not next-day retention, and the slow component drove retention but not savings, while a history of errors modulated both processes in force-field reaching (S27). How adaptation was removed decided whether savings survived: washout by removed feedback or error-clamp trials left none while gradual reversion did (S24). In 40 adults on a split-belt treadmill (S25, randomized four-arm) four 7.5-minute exposures on one day produced day-5 savings comparable to repeated 30-minute exposures across days at one quarter the adaptation time, and consolidation of a new internal model required practice beyond the performance plateau with a longer inter-session interval making it harder to reactivate (S26). Boundary: laboratory reaching, tracking and locomotor adaptation in adults over minutes to days, never months of natural nonuse; no study measures savings in a striking skill after a documented layoff; the '80 percent fewer trials' figure in one leg is a within-paradigm count from a fitted two-state model (MODEL class) and is not a striking number. Evidence class: INTERVENTION plus MODEL.

The physics underneath

Two processes with different time constants adapt to error (S23, S27): a fast one that learns and forgets quickly and a slow one that learns and forgets slowly. Savings arises when the fast process is re-engaged by a recognized perturbation and by raised error sensitivity, so the returning performer re-tunes quickly without the tuned output having been stored. The engine of re-tuning is the error signal itself; in striking the calibration error of distance and timing is being hit, so the savings signal is paid in head impacts (an inference carried to the brain vector).

The sources
← Previous axiom: 7.5, Force-producing capacity outlasts and returns ahead of force-absorbing tissue: strength and neural activation hold through detraining while tendon stiffness and muscle cross-section return to baseline at one and two monthsNext axiom: 7.7, Composure under incoming strikes is extinction learning, not a motor memory: it is outcompeted by time (spontaneous recovery), by context change (renewal) and by one aversive event (reinstatement), fear reacquires faster than safety, and a threat-trained performance benefit held four months in police trainees with anxiety still elevated →Link to this axiom: #ax-7-6
7.7 Composure under incoming strikes is extinction learning, not a motor memory: it is outcompeted by time (spontaneous recovery), by context change (renewal) and by one aversive event (reinstatement), fear reacquires faster than safety, and a threat-trained performance benefit held four months in police trainees with anxiety still elevatedVERIFIEDComposure is a context-specific safety memory competing with a generalizing fear memory, so time, a new context or one aversive event lets the fear response return.Open this axiomClose this axiom
The axiom in full

In human and animal fear conditioning (S28, S29, S31) extinction does not erase the fear association; it lays down a second, context-specific inhibitory memory that competes with it, so extinguished responding returns with the passage of time (spontaneous recovery, a change of temporal context), with a change of background context (renewal), after a single unpaired aversive event (reinstatement), and the original response reacquires faster than it was extinguished; unrelated stress outside the learning context strengthens spontaneous recovery (S30). In 27 police officers (S32, randomized two-arm, 4 one-hour sessions, 48 training shots, 4-month retention) the group that shot against an opponent who sometimes fired painful marking cartridges kept its accuracy under the anxiety manipulation at retention while the control group's accuracy fell under anxiety, and reported anxiety stayed higher in the threatening condition for both groups: regulation meant performing despite anxiety, not removing it, and the control group also improved by retention, so the exclusive advantage is bounded. Boundary: fear-conditioning laboratories and one police-trainee intervention; no opened study measures composure under strikes across a documented layoff in fighters (UNVERIFIED after search by three legs); the Cooper colour-code heart-rate bands in one leg are ARGUMENT and carry no measured threshold here. Evidence class: INTERVENTION (S32, conditioning experiments) and CORRELATIONAL review (S29).

The physics underneath

Threat learning generalizes and safety learning is tied to the context in which it was learned (S28), so composure built in one room against known partners is the component most exposed to a layoff: time alone shifts the temporal context that gated the safety memory, a new room or partner removes the cues, and one bad exchange on return reinstates the old response. This makes composure the one component that can be worse after an interruption than a disuse model predicts, and it cannot be kept by solo practice, which contains no incoming threat to extinguish. Adjacency named: The Physics of Flow States (February 2026) treats arousal regulation in non-combat performers and is silent on retention; this axiom bounds that frame by name for the combat case, where the regulated state is an actively maintained, context-gated inhibitory memory, not a stable trait.

The sources
← Previous axiom: 7.6, Savings on relearning is faster re-access of a strategy plus heightened sensitivity to familiar errors, dissociable from long-term retention; it requires prior error and depends on how practice stopped and how exposure was structured, not on total minutesNext axiom: 7.8, Expert anticipation in striking lives in the coupling of perception to a live, moving opponent, so practice that removes the opponent cannot by construction maintain it; its decay across a layoff has never been measured →Link to this axiom: #ax-7-7
7.8 Expert anticipation in striking lives in the coupling of perception to a live, moving opponent, so practice that removes the opponent cannot by construction maintain it; its decay across a layoff has never been measuredVERIFIEDAnticipation is extracted from a live opponent's early kinematics in depth, so any practice without that moving body cannot maintain the component, whatever else it maintains.Open this axiomClose this axiom
The axiom in full

In karate athletes tested in situ with temporal occlusion (S33, cross-sectional expert, near-expert, novice) experts anticipated above chance at every occlusion point, near-experts only at the two latest points and novices only at the last; expert anticipation did not differ from near-experts at the earliest occlusion point but did at all later ones. Methodologically (S34) expertise effects are larger when the response is a coupled movement than an uncoupled verbal or button prediction, and occlusion methods that preserve perception-action coupling and binocular depth cues are what reveal expert performance. From these follows a bound on reduced practice: film study, shadow work, bag work and compliant pad work decouple perception from the opponent's kinematics and therefore maintain coordination (A1) and conditioning (A3, A4) but not the anticipation component. Boundary: the combat-sport anticipation literature is cross-sectional expert-versus-novice and training-intervention work; no opened study measures the decay of combat anticipation across a documented layoff (three legs searched and found none), so the rate of loss is UNRESOLVED and the 'two to four weeks' figure in one leg has no primary source; expert-to-novice contrasts describe a population difference, never an individual's present read. Evidence class: CORRELATIONAL (S33) and ARGUMENT grounded in intervention methods (S34).

The physics underneath

Anticipation is early pickup of kinematic information from a specific moving body in depth, mapped to a response (S33, S34); it is not a stored file retrievable without the information field. Removing the opponent removes the input the skill is made of, so the exposure that preserves it is live, coupled work and no other exposure substitutes. By A2's task-type ordering it sits in the cognitive, accuracy class that loses most with nonuse, an inference from pooled tasks, not a combat measurement.

The sources
← Previous axiom: 7.7, Composure under incoming strikes is extinction learning, not a motor memory: it is outcompeted by time (spontaneous recovery), by context change (renewal) and by one aversive event (reinstatement), fear reacquires faster than safety, and a threat-trained performance benefit held four months in police trainees with anxiety still elevatedThis is the last axiom on the page.Link to this axiom: #ax-7-8

The easy answers this run refused

Each vector was told the common saying it must not settle for, and told to rebuild the answer from first principles instead.

Composure under attack runs on mechanisms, and every mechanism on this page can be checked again.