Symbio ic human-AI a chi ec u e o soma ic
sensing, symbolic easoning and me acogni i e
con ol†
Alexande Ma hiesen-Ohman1and Jacek Ma lecki2*
1AMOTHO Resea ch Ins i u e, Vallsj¨on 20, 780 00 R¨o b¨acksn¨as,
Sweden.
2Depa men o Ma hema ics, W oc law Uni e si y o Science and
Technology, Wyb ze˙ze Wyspia´nskiego 27, 50-370 W oc law, Poland.
*Co esponding au ho (s). E-mail(s): jacek.malecki@pw .edu.pl;
Abs ac
We p esen a ipa i e cogni i e a chi ec u e ha uni ies soma ic g ounding,
symbolic in e ence and me acogni i e con ol wi hin an ex ended SORK-N loop.
Cogni ion is cas as he in e ac ion o a Soma ic laye (biophysical and a ec-
i e inpu s), a Symbolic laye (linguis ic, logical, and ep esen a ional p ocesses),
and a Me acogni i e laye (global cohe ence es ima ion and policy adjus men ),
coo dina ed by a me hodological amewo k ha ime-locks and analyzes mul-
imodal physiological, linguis ic and sel - epo da a. Ma hema ical s uc u e is
p o ided by he Mi o ed P o ile G aph (MPG), an e idence-linked, hie a chical
s a e space, and Rogue Va iable (RV) analysis, which oge he localize s uc u al
sou ces o p edic ion-obse a ion gaps and suppo alsi iable es s. This ame-
wo k enables ep oducible es s o in ui ion, p e-e en cogni ion, and collec i e
cohe ence, while emaining compa ible wi h empi ical sc u iny. We u he dis-
cuss implica ions o symbio ic human-AI sys ems and a gue ha in en ional
co-e olu ion o biological and a i icial cogni ion o e s a p ac ical ou e owa d
obus , e lec i e in elligence.
Keywo ds: Cogni i e a chi ec u e; Human-AI symbiosis; Soma ic sensing; Symbolic
easoning; Me acogni i e con ol; SORK-N loop; Mi o ed P o ile G aph (MPG);
Rogue Va iables (RV); Minimal Unawa e Flip Se (MUFS); Coun e ac ual analysis;
Decision calib a ion; G aph-based easoning; Mul imodal signals; Hyb id in elligence;
H3lix sys em; LAIZA p o ocol; p ocess.
†Associa ed IP: U.S. P o isional U ili y Pa en Applica ion No. 63/910,500,
“H3LIX: AI–Human Symbio ic In eg a ion P ocess,” iled No embe 3, 2025 (p ocess pa en ).
1
1 In oduc ion: Towa d an In eg a i e Science o
Cogni ion
Con empo a y science app oaches in elligence om wo di e gen adi ions (see
[23,5]). The ma e ialis pa adigm explains cogni ion as an eme gen compu a ion o
neu al and algo i hmic p ocesses, while he noe ic adi ion ega ds consciousness as an
in insic, causally ac i e aspec o eali y. Bo h gene a e aluable insigh s, ye nei he
alone can accoun o he ull phenomenology o mind (see [52]). The o me s uggles
wi h he ha d p oblem o subjec i e expe ience (c . [20]), while he la e lacks a ep o-
ducible empi ical amewo k. This epis emic gap limi s p og ess in unde s anding and
enginee ing sys ems ha genuinely know ha hey know.
We p opose ha cogni ion can be modeled as a dual-aspec p ocess wi hin a single
in o ma ional subs a e ( ollowing [5]; c . [20]), in which mind and ma e ep esen
complemen a y exp essions o one con inuum (c . a ia ional accoun s unde he ee-
ene gy p inciple [28]). The H3LIX a chi ec u e ope a ionalizes his p emise h ough
h ee in e dependen laye s: Soma ic, Symbolic, and Noe ic (compa e o [36]). These
laye s o m a helical i-s uc u e whose dynamic in e ac ions a e go e ned by eedback
acco ding o he SORK-N model de i ed om beha io al science ([49]). Wi hin his
sys em, he LAIZA P o ocol se es as a esea ch me hodology designed o co ela e
subjec i e expe ience wi h objec i e measu emen , he eby ansla ing he s udy o
consciousness om me aphysical specula ion in o es able empi icism (see [91]).
The p o ocol collec s synch onized physiological, linguis ic, and phenomenological
da a s eams, allowing o he de ec ion o p econscious ma ke s o decision (see [84]),
in ui ion o nonlocal co ela ion (c . [59]; bu see [76] o c i ique). Analyses combine
in o ma ion- heo e ic summa ies (e.g. ans e en opy [75]) wi h causal easoning o e
s uc u ed ep esen a ions (see [68] and G ange -based e ec i e connec i i y [78]) o
suppo alsi iable claims. By combining Bayesian and in o ma ion- heo e ic analy-
ses wi h human- epo ed in ospec i e s a es [26,88], he model aims o de e mine
whe he speci ic con igu a ions o consciousness can measu ably in luence o an icipa e
physical and in o ma ional sys ems.
The ul ima e objec i e is no o eplace exis ing pa adigms, bu o in eg a e hem
by es ablishing a uni ied, alsi iable amewo k whe e biological and a i icial cogni-
ion co-e ol e. In doing so, he H3LIX sys em posi ions i sel as bo h a heo e ical
and p ac ical ins umen o explo ing he on ie be ween neu oscience, in o ma ion
heo y, and he eme ging ield o noe ic science.
2 Theo e ical Backg ound
Ou app oach is in en ionally ancho ed in well-es ablished scien i ic adi ions a he
han ad hoc specula ion. We syn hesize classical beha io ism (S-O-R-K), con em-
po a y p edic i e p ocessing and embodied cogni ion, and dual-aspec in o ma ion
pe spec i es in o a single, es able amewo k, and we ope a ionalize hese links wi h
igo ous ma hema ical ools (e idence-linked Mi o ed P o ile G aphs and Rogue
Va iable analysis). This g ounding ensu es ha each la e claim abou cohe ence,
2
in ui ion, and me a-con ol es s on ep oducible cons uc s and s anda d s a is ical
me hodology.
2.1 Beha io al Founda ions: The SORK Model and Adap i e
Feedback
The classical S-O-R-K amewo k (S imulus - O ganism - Response - Kon in-
genz/Consequence) expanded beha io is models by emphasizing o ganismic a iables
(pe cep ion, memo y, mo i a ion) and he ein o cemen con ingencies ha shape
u u e beha io (see [48]). We e e ence he SORK adi ion as a canonical o -
maliza ion o hese con ingencies in clinical and expe imen al se ings (c . [47,35]).
H3LIX ex ends his logic in o he digi al and cogni i e domain h ough SORK-N,
whe e N deno es Noe ic In eg a ion. This i h phase o malizes e lec i e model-
ing o he beha io al loop-suppo ing bo h i s -o de (beha io al) and second-o de
(me acogni i e) adap a ion ( ollowing [25]; see also [7]).
2.2 P edic i e P ocessing and Embodied Cogni ion
P edic i e p ocessing ames he b ain as a hie a chical in e ence engine minimizing
su p ise by con inuously upda ing p edic ions o senso y inpu (see [21,70]). This
aligns wi h he ee-ene gy o mula ion o co ical in e ence and ac i e sensing (c . [28,
67]). In pa allel, embodied cogni ion iews mind as he dynamic enac men o a body in
a wo ld (see [12]). Toge he hey imply ha any model app oaching genuine si ua ional
awa eness mus in eg a e mo o , a ec i e, and senso y eedback, he Soma ic Laye
o H3LIX (c . [77,1]).
2.3 Dual-Aspec In o ma ion Monism
In e disciplina y wo k in philosophy, neu oscience, and in o ma ion heo y p oposes
ha ma e and mind a e complemen a y mani es a ions o a single in o ma ional
subs a e (see [89,65,4]). Con e gen s ands include in o ma ion- heo e ic models o
consciousness and empi ically ac able dual-aspec p oposals. H3LIX ope a ionalizes
his s ance by ea ing symbolic compu a ion (objec i e in o ma ion low) and noe ic
esonance (subjec i e co ela ion) as coupled domains o one p ocess, wi h he LAIZA
P o ocol measu ing c oss-co ela ions be ween hem.
2.4 Noe ic Resea ch and he Empi ical Challenge
Psi/noe ic esea ch has explo ed ex asenso y pe cep ion, elepa hy, and p ecogni-
ion (e.g., [14]). Repo ed e ec s a e ypically small and con es ed (see [73]; c .
[93]). Me a-analy ic e idence and high-powe ed eplica ion a emp s oge he mo i-
a e s ingen , p e egis e ed es s wi h balanced in e p e a ion (e.g. [59,19]; and
me hodological guidance in [64]). H3LIX does no assume a i ma i e answe s, bu i
p o ides a neu al empi ical amewo k o es ing hem, embedding noe ic hypo heses
in p edic i e-p ocessing and s anda d s a is ics.
3
2.5 Syn hesis
F om beha io ism, H3LIX inhe i s ein o cemen p ecision, om p edic i e neu o-
science, e o minimiza ion, om dual-aspec heo y, an on ological b idge, om noe ic
s udies, he cou age o es beyond cu en empi icism. Cogni ion becomes a ecu si e
in o ma ional p ocess. The LAIZA p o ocol is an expe imen al philosophy o mind
ende ed as me hod.
3 The H3LIX A chi ec u e
As we s a ed be o e, H3lix is a ipa i e cogni i e a chi ec u e (Soma ic, Symbolic,
Noe ic) in e ac ing ia a uni ying Mi o Co e based on Mi o P o ile G aph oge he
wi h AI agen ic sys em. The design cap u es cogni ion as a closed-loop in o ma ion
ecology. Laye s may be analyzed sepa a ely, bu hey ope a e as one p ocess.
3.1 The Soma ic Laye
The Soma ic Laye p o ides he biophysical g ounding o H3LIX. I agg ega es low-
la ency signals ha co- a y wi h a ousal, a ec , mo o p epa a ion, and au onomic
egula ion (see [2,22]; c . [81]), hen con e s hem in o a calib a ed s a e ec o
sui able o p edic i e coding and c oss-laye alignmen (compa e o [83]; c . [67]). The
gene al ope a ional scheme is as ollows
1. Signals and senso s. The ca alog o signals p ocessed by he sys em is no closed
and includes all echnically a ailable senso y da a comp ising, among o he s, elec-
ode mal ac i i y (EDA/GSR), ECG o PPG o HR/HRV ( ime/ equency and
non-linea indices), espi a ion, acial EM, oculome ics (blink, ixa ion, pupil),
speech p osody, pos u e/mic o-mo ion, EEG mic o-po en ials and na ow-band
hy hms.
2. Acquisi ion and synch oniza ion. Channels a e ime-locked o a mas e clock wi h
e en ma ke s o s imuli, esponses, and in ospec i e epo s. Sho , o e lapping
windows yield obus ea u es wi h pe -channel unce ain y.
3. S a e es ima ion. P ep ocessed s eams x( ) p oduce s anda dized ea u es z( )
and unce ain ies σ( ). A ligh Bayesian il e yields a smoo hed soma ic s a e
ˆ
s( ) wi h inno a ion ε( ) (Soma ic con ibu ion o p edic ion e o ). This ol-
lows a ligh weigh Kalman amily upda e, wi h physiological p io s ancho ed in
es ablished HRV and pupillome y ma ke s o a ousal and unce ain y [46,79,43].
4. An icipa o y ma ke s. E en -locked windows p eceding ac ion/ epo a e scanned
o eadiness-like ends and c oss-modal phase-locking; ma ke s a e o wa ded
o LAIZA o p econscious analyses.
5. Ou pu s. As he esul o he p ocess, we ob ain ime-aligned ec o s ˆ
s( ) wi h
gi en co a iance, e en -locked summa ies, change-poin s/anomaly lags, all con-
sumable by he sys em o be analyzed in he iew o Mi o ed P o ile G aph.
4
3.2 The Symbolic Laye
The Symbolic Laye is he a chi ec u e’s ep esen a ional and in e en ial wo kspace.
Ope a ionally, i is ealized by he LAIZA p o ocol, which coo dina es h ee hings:
he s o age and alignmen o mul iple da a s eams, he con e sion o language and
beha io in o s uc u ed ep esen a ions (see [10]), and a se o p obabilis ic ea-
soning ou ines ha use soma ic e idence wi h disc e e symbols o p oduce belie s,
explana ions, and policies (see [53,34]; c . [44]). This laye ansla es obse a ions
in o linguis ic, ma hema ical, and logical o ms, uns coun e ac ual simula ions and
“wha -i ” analyses ( ollowing [68]) and u ns conclusions in o candida e ac ions (com-
pa e o [29]). I p o ides op-down p edic ions ha guide pe cep ion and beha io (c .
[70]), and i su aces unce ain y so o he pa s o he sys em know when o ely on,
ques ion, o o e ide i s ou pu (see [51]).
1. LAIZA keeps ime-aligned eco ds o ex , speech, and beha io alongside he
soma ic summa ies coming om senso s. Language is pa sed in o en i ies, e en s,
and ela ions. Beha io s a e cap u ed as disc e e ac s and sequences and bo h a e
embedded in o a sha ed ep esen a ional space so ha bodily cues and symbolic
con en can be compa ed di ec ly. The laye hen main ains a unning “belie
s a e” o e his space-essen ially a compac snapsho o wha he sys em cu en ly
hinks is ue, ele an , o likely oge he wi h a ace o how con iden i is in
each pa o ha belie .
2. The Symbolic laye gene a es p edic ions abou upcoming wo ds, e en s and ou -
comes. I compa es hose p edic ions wi h wha ac ually happens and adjus s i s
in e nal ep esen a ions and weigh s o educe u u e mis akes. When e o s pe -
sis o con adic ions appea , i ma ks hose a eas as unce ain and in i es ei he
mo e e idence ( om he Soma ic Laye ) o s uc u al guidance ( om he Noe ic
Laye ).
3. Typical ou pu s include a machine- eadable accoun o he cu en si ua ion (en i-
ies, ela ions, in en s), p edic ions wi h calib a ed con idence, al e na i e plans
wi h expec ed u ili y and isk, and concise indica o s o whe e he model is unsu e
o misaligned wi h he da a.
3.3 The Noe ic Laye
This laye moni o s he whole sys em o signs o global cohe ence, i.e. hose momen s
when bodily hy hms, in e nal na a i es, and ex e nal ou comes line up. I de ec s
s a es commonly desc ibed as in ui ion, insigh , o an icipa o y awa eness by acking
how s able, synch onized, and mu ually ein o cing he signals a e ac oss laye s (c .
ime- and phase-based cohe ence measu es [54,92,32]). When such a s a e eme ges,
he laye ea s i as a sys em-le el con igu a ion a he han a single cue, and i
adjus s p io s, a en ion, and decision cadence acco dingly.
The laye ope a ionalizes in ospec ion by u ning “ el cohe ence” in o es able,
sys em-wide indica o s. We summa ize mul is eam egula i ies using complex-
i y/en opy ools sui ed o sho windows and non-Gaussian dynamics (see [11,72,
5
94]). The noe ic laye asks whe he pa icula men al o emo ional con igu a ions
co espond o measu able educ ions in noise o p edic ion e o and e alua es his
claim epea edly ac oss asks and con ex s. In doing so, i e ames “noe ic” phenom-
ena as pa e ns o non-local in o ma ion esonance: dis ibu ed alignmen s ha a ise
ac oss senso s, symbols and ou comes. Ins ead o aking in ui ion a ace alue, i es s
whe he cohe en s a es eliably p ecede mo e accu a e judgmen s, as e con e gence,
o sa e ac ions and unde which bounda y condi ions hey do no (see condi ions o
alid in ui ion [45]). The ou pu s o he laye can be desc ibed as
1. Co ela ion ma ices: compac summa ies o how s ongly key signals co- a y
(e.g., soma ic ma ke s wi h na a i e consis ency and pe o mance), used o
iden i y s able alignmen s e sus b i le coincidences.
2. En opy-change coe icien s: indica o s o whe he he sys em is mo ing owa d
o de o diso de , compu ed o e ele an s eams (physiology, language, policy
choices) o lag eme ging s abili y o agmen a ion (c . [11,72,94]).
3. Cohe ence spec a: ime- and equency-awa e po ai s o alignmen ha e eal
when and a wha scales cohe ence appea s (e.g., sus ained, ansien , o
oscilla o y episodes) (see [54,32,92]).
4. P obabilis ic measu es o in ui i e accu acy: calib a ed es ima es o how likely a
de ec ed cohe en s a e will ansla e in o be e ou comes o he cu en ask,
de i ed om he sys em’s own his o y a he han a ixed p io (assessed wi h
p ope sco ing ules [31]).
Toge he , hese ou pu s le he a chi ec u e use “in ui ion” as a disciplined con ol
signal: de ec able, audi able, and ac ionable.
3.4 Mi o Co e and Feedback Dynamics
The Mi o Co e is he coo dina ion hub ha ins an ia es SORK-N a un ime. Build-
ing on S-O-R-K (see [48]), i adds Noe ic In eg a ion (N) o u n each beha io al
loop in o a e lec i e one, whe e we combine he i s -o de ac ion and second-o de
o pe o m me acogni i e adjus men ( ollowing [25]). Consequen ly, he s uc u e o
he cycle is
S (S imulus): inges and ime-align soma ic and con ex ual signals.
O (O ganism): upda e he Symbolic Laye ’s belie s, expec a ions, and candida e
ac ions.
R (Response): execu e/schedule ac ions and eco d p ecise ou comes (c . [87]).
K (Kon ingenz): a ach consequences and ex e nal eedback o he same clock
(see [87]).
6
N (Noe ic): assess global cohe ence, a ibu e anomalies, and se me a-adjus men s
(p io s, h esholds, a en ion) (c . adap i e unce ain y/a en ion con ol [95];
lea ning- a e/p io s adap a ion [60,13]).
S’ (W i e-back): apply hose adjus men s o he nex s imulus epoch (adjus ing
decision h esholds/cadence as needed [71]).
The ollowing pa ame e s a e subjec o egula ion: a en ion and ga ing o inpu s
(c . [95]), decision h esholds and cadence (see [71]), lea ning a es and p io s (c .
[60,13]), and b ie p obes o inhibi ions du ing pe iods o low cohe ence. In his
manne , he Noe ic Laye (N) con inuously condi ions S’, he eby closing he loop and
enabling bo h beha io al and me acogni i e adap a ion.
4 The LAIZA P o ocol
The LAIZA p o ocol is a me hod used o s udy he dependencies be ween biological,
symbolic, and noe ic p ocesses, based on AI models and ma hema ical modeling. How-
e e , his model is no igidly de ined. Ins ead, i is c ea ed by a sys em o coope a ing
a i icial in elligence ools which, wi hin he amewo k o he p oposed ma hema ical
and algo i hmic mechanisms, o m a s uc u e e lec ing he use ’s cogni i e cons uc
(see [9,41]). In addi ion, he sys em is in a con inuous p ocess o c ea ion and ans-
o ma ion, which i sel is also a subjec o analysis (c . con inual lea ning [66]). A e y
impo an elemen o he sys em is a as s eam o inpu da a encompassing all a ail-
able da a ela ed o he use , which ensu es he possibili y o in-dep h analysis and
he c ea ion o a model ha akes in o accoun all equi ed aspec s (see digi al pheno-
yping and sa e y conside a ions [40]). The immense p og ess in he ield o a i icial
in elligence algo i hms and he inc eased a ailabili y o echnological ools ela ed o
da a collec ion and p ocessing inally enables he LAIZA p o ocol o become an empi -
ically alid me hod ha was no p e iously a ailable (c . wea able sensing ad ances
[3] and ounda ion models [57]). We begin wi h p esen ing wo main ma hema ical
models used in he p o ocol, which a e Mi o ed P o ile G aph (MPG) and Rogue
Va iable (RV) analysis.
4.1 Mi o ed P o ile G aph (MPG): A compac o malism
MPG is an e idence-linked, hie a chical g aph model o a pe son’s p o ile ha plugs
in o he LAIZA p o ocol. I u ns da a and mul imodal aces in o a s uc u ed
objec ha suppo s in e ence, eedback, and symbio ic AI co-adap a ion (see knowl-
edge g aph o malisms [38,42] and LLM-powe ed cons uc ion [17]). I complemen s
he Soma ic–Symbolic–Noe ic loop and SORK-N eedback by p o iding a pe sonal,
audi able s a e space o e which hose dynamics can ope a e (c . p o enance s anda ds
[58]). Conc e ely, each node cap u es psychologically meaning ul cons uc s ( ai s,
alues, coping ou ines, cues, pi o al e en s) wi h p o enance, while yped, di ec ed
edges encode ela ions such as causes, igge s,bu e s, and mode a o s, yielding a
compac causal-con ex ual sca old (c . [86,69]).
7
Beyond he la iew, MPG is hie a chical: any node may deno e a segmen (a sub-
g aph) wi h i s own e idence, me ics, and in e nal edges. Edges can a ge segmen s
as wholes o p ojec o speci ic subnodes ia bounda y in e aces, enabling ho izon al
pa hs wi hin a le el and e ical pa hs ha a e se le els (node ↔segmen ; segmen
↔segmen ). T ea ing segmen s as nodes a a highe le el induces a ecu si e li o
me a-g aphs o “ hough s abou hough s,” whe e segmen -le el impo ance and con-
idence a e olled up om hei con en s (see mul i-scale/ hie a chical o ganiza ion
pa allels [16,8]).
Node/edge weigh s s ill sepa a e impo ance (beha io al cen ali y) om con i-
dence (e idence suppo ), enabling p incipled upda es as new aces a i e a any
laye (e.g. p obabilis ic weigh ing [6], ne wo k cen ali y ounda ions [61]). The esul
is a human-in e p e able, machine-legible p o ile: small edi s o nodes, segmen s,
edges, o weigh s p opaga e anspa en ly h ough p edic i e models and AI-assis ance
s a egies ac oss le els (c . ela ional/lea ning me hods on KGs [63]).
We begin he desc ip ion o he ma hema ical model by in oducing he laye - ype
g aph and i s s uc u e, which hen will be p opaga ed in o highe le els by he li
ope a o .
De ini ion 1 ALaye -Type G aph is a yped, di ec ed mul ig aph equipped wi h a ibu e
maps:
G = V, E, ΣE;τ, m, w, E,Con ,Imp, R,
whe e
•Nodes and edges. Vis he se o nodes; Eisamul ise o di ec ed edges (pa allel
edges allowed). Thus, o u, ∈V he e may exis se e al dis inc edges ek∈E(e.g.,
di e en ela ion ypes).
•Laye s. Lis an open label se , sys em-gene a ed amily o laye s’ names (e.g., Psy-
chological,Social,His o ical,Family,P o essional,Hobby) in e ed om da a. The
node- o-laye map is λ:V→ P(L) (mul i-labels allowed).
•Edge ypes. ΣEis an ex ensible alphabe o ela ion ypes, ini ially seeded wi h causes,
igge s,ampli ies,bu e s,mode a es,enables,aligns,is pa o ,con adic s,co ela es,
e c. The ype map is τ:E→ΣE.
•Node me ics. m:V→[−1,1] ×[0,1]3 e u ns ( alence,in ensi y, ecency,s a-
bili y), cap u ing a ec i e sign/magni ude,s eng h/ equency, eshness, and empo al
consis ency.
•Edge s eng h. w:E→[0,1] gi es he e ec s eng h o each di ec ed ela ion.
•E idence. E: (V∪E)→2Massigns each node/edge a se o e idence i ems ei∈
M. Each eis o es: a sho desc ip ion and p o enance class (selec s quali y mul iplie
qi o con idence), a esol able poin e /iden i ie , a sho e ba im o ligh ly quo ed
snippe suppo ing he claim, and a imes amp ( o imeliness ac o i). Fou nume ical
pa ame e s accompany each i em: i em-le el suppo ci∈[0,1] (how s ongly his i em
suppo s he speci ic node/edge, independen o qi), sou ce-quali y mul iplie qi>0,
di e si y bonus ui∈[1,1.15], and imeliness ac o i∈(0,1].
8
•Con idence. Con : (V∪E)→[0,1] is compu ed om e idence ia
S(x) = X
ei∈E(x)
ciqiui i,Con (x) = 1 −exp
−αS(x),(1)
wi h adap able calib a ion pa ame e α∈(0,∞).
•Impo ance. Imp : V→[0,1] e u ns he impo ance o e e y node based on i e ac-
o s: alence,in ensi y, ecency,s abili y, and cen ali y. Each ac o comes om simple
inpu s and he cu en g aph s uc u e. All a e scaled o [0,1] o easy compa ison,
excep alence, which is bipola [−1,1]. In pa icula , alence comes om pleasan ness
a ings, emo ion/a ec labels, and sen imen o he da a; in ensi y om s eng h a -
ings and sho -window equency coun s; ecency om he age o he la es suppo ing
episode o a i ac ; s abili y om empo al consis ency and sou ce ag eemen ; and cen-
ali y om he node’s posi ion in he cu en g aph. The inal compu a ion uses a
lea nable linea model.
Node impo ance Imp and Impu oge he wi h edge s eng h a e hen used o measu e
he con ex ual edge p io i y o he edge u→ , which is one o he key elemen s in he
sys em.
•Reasoning p o enance. R: (V∪E)→Tex s o es a sho , human- eadable jus-
i ica ion o why each node o edge was c ea ed (e.g., he ule-o - humb, abduc i e
s ep, o mapping om an in e iew answe ), dis inc om aw e idence. This enables
anspa en audi s and model c i ique.
We deno e by GLT he amily o all Laye -Type G aphs.
The nex cons uc ion s ep is o desc ibe he segmen a ion mechanism and he
p ocess o building he highe -le el g aph s uc u e. The p oposed p ocedu e is uni-
e sal and is an algo i hmic me hod o mo ing om a gi en le el o a highe le el
(c . mul ile el pa i ioning and hie a chical coa sening [50,18]; see also low-based
segmen a ion [74] and spec al me hods [62]). The s a ing poin is a g aph wi h he
s uc u e desc ibed abo e. The cha ac e is ics o he indi idual e ices and edges
allow o he iden i ica ion o segmen s, which a e subg aphs consis ing o a se o e -
ices and edges wi h speci ic p ope ies ( opological, seman ic, unc ional, hema ic,
e c.) (see [27,15]).
De ini ion 2 Gi en a le el-kg aph o Laye -Type G aph
G(k)= (V(k), E(k),ΣE;τ(k), m(k), w(k),E(k),Con (k),Imp(k), R(k))
asegmen is any subg aph S= (VS, ES) wi h VS⊆V(k),ES⊆E(k). Le S(k)=
{S1,...,SNk}deno e a chosen (possibly o e lapping) amily o segmen s a le el k, whe e
Nkis he numbe o selec ed segmen s o gi en le el k∈ {0,1,2, . . .}.
The selec ed segmen s become he e ices o he g aph a he highe le el. No e ha
a selec ed segmen migh be jus one node, which allows us o conside also segmen
o node and node o segmen ela ions. Addi ionally, each segmen has a speci ied
bounda y in e ace, which a e g oups o e ices ha connec wi h o he segmen s
9
•Lea ning and e-cohe ence. Ou comes om in e en ions upda e edge
s eng hs, impo ances, and in e ace selec ions; summa ies oll upwa d ac oss
le els and con i med highe -le el ela ions p ojec back down so cohe ence is
es o ed as p edic ion e o subsides.
•Go e nance and impac accoun ing. All changes a e alida ed, e sioned,
and na a ed; he sys em acks which e en s o chains mos shaped s uc u e
and policy so u u e decisions become as e , sa e , and mo e explainable.
5 MPG in ui ion model
Many concep s included in he b oadly concei ed noe ic g asp a e conside ed unscien-
i ic p ima ily because hey lack a well-es ablished de ini ion and a se o ope a ional
guidelines wi hin which his de ini ion can be applied and es ed expe imen ally. The
ma hema ical models p esen ed in he p e ious chap e s allow o a e y speci ic
desc ip ion o phenomena ha ha e p e iously been comple ely elusi e. In his chap-
e , we will p esen a s ic ly ma hema ical de ini ion o in ui ion wi hin he MPG
model (MPG-In ui ion), which, hanks o digi al echnology and ad anced a i icial
in elligence models, can be es ed expe imen ally.
We will begin by de ining in ui ion wi hin he MPG model as a p ocess occu ing
in ou mind (modeled by he sys em), whe e he e is a di e ence be ween all he da a
and s uc u es u ilized by he b ain (sys em) du ing in e ence and p edic ion, and
hose ha a e consciously pe cei ed (c .[37], [82]). This di e ence mus be su icien ly
signi ican , which is also o mally de ined, so as o cause a change in he inal decision
made.
Se up. Ou s a ing poin is he Mi o ed P o ile G aph
MPG = {G(0),G(1) . . . , G(N)}
Addi ionally, we ha e a de ined se o decisions ha mus be aken, which we deno e
by H0. This could be a simple bina y decision (H0={0,1}), a classi ica ion p oblem
(H0={h1, . . . , hn}), o any con inuous p oblem (H0⊆Rd). Fu he mo e, we ha e
a se o obse a ions, which we deno e by O={o1, . . . , om}. Ou goal is o make a
decision based on he obse ed da a Oand he MPG g aph, a p ocess which we model
wi h he decision ule ˆ
h=ˆ
h(O, MPG, R(H)) and u ili y a es R:H→Rd, which in
gene al migh be mul idimensional.
Unawa eness ac o s. As we ha e al eady o mula ed, he key concep in he de -
ini ion o in ui ion is he no ion o a lack o ull awa eness. We p opose wo ypes o
unawa eness ha need o be conside ed.
•Inpu Unawa eness (IU). The i s is he simples and conce ns he ac ha
we lack ull awa eness ega ding he obse ed da a ela ed o he decision being
made. This means ha ou awa eness pe ains o he se ˜
O, which is di e en
om he se O; speci ically, ˜
O⊊O. Consequen ly, we de ine he se o unawa e
inpu s as Uin =O ˜
O.
16
•P ocess Unawa eness (PU). The second ype o unawa eness conce ns he
lack o cogni i e knowledge ega ding ce ain pa s o he MPG g aph ha a e
signi ican in he decision-making p ocess we a e discussing. Simila ly as be o e,
we de ine Up oc ⊆MPG as unawa e pa o he g aph.
The nex s ep is o indica e a measu e ha will de e mine whe he he le el o
unawa eness is la ge enough o ha e a signi ican impac on he decision-making p o-
cess. We p opose ha his be done by poin ing o he exis ence o a se o unawa e
s a es o unawa e objec i e goals ha lip o change he decision. Acco dingly, we
de ine he o al se o unawa e s a es as U=Uin ∪ Up oc.
De ini ion 7 A se U⊂ U is called Minimal Unawa e Flip Se (MUFS) i emo ing all
elemen s in U lips o changes ˆ
h, while emo ing any p ope subse o Udoes no .
Based on ha de ini ion we can p oceed di ec ly o he de ini ion o in ui ion, mo e
p ecisely in ui i e decision in he MPG based sys em.
De ini ion 8 (MPG - In ui ion) A decision is said o be MPG-In ui i e i he e exis s a
leas one nonemp y Minimal Unawa e Flip Se U⊆ U.
This is no he only possible de ini ion ha can be conside ed in he con ex o
a ma hema ical model; howe e , i se es as a s a ing poin and is su icien ly com-
p ehensi e ye ma hema ically p ecise o o m he basis o expe imen al esea ch.
Subsequen possible gene aliza ions and applica ions o he en i e sys em, including
he ma hema ical model, a e no only ela ed o mo e sub le de ini ions o in ui ion-
which include, among o he hings, he ac ual di e ence be ween he es ablished
p edic ion a ge and he ue p edic ion a ge , o a di e ence in he loss unc ion
o he es ima o i sel . The ma hema ical ounda ion o he sys em also allows us o
p o ide s ic ly o mal de ini ions o o he phenomena ela ed o cogni ion and me a-
cogni ion, which a e collec i ely e med noe ic (such as insigh , o example). All hese
concep s will be he subjec o u he esea ch.
6 Expe imen al S a egy o Tes ing MPG - In ui ion
We p opose a compac p og am o expe imen s ha p obes MPG - In ui ion a h ee
le els: sys em-only alida ion inside H3LIX, human decision eplica ion wi h he sys-
em unning in pa allel, and a link o sel - epo ed in ui ion. In all cases, he co e
manipula ion con as s a ull awa eness condi ion, whe e all inpu s and model p o-
cesses a e a ailable wi h a es ic ed awa eness condi ion, c ea ed ei he by masking
inpu s (Inpu Unawa eness, IU) o by abla ing limi ed g aph s uc u e in he MPG
(P ocess Unawa eness, PU). The same asks a e used ac oss le els wi h di icul y
adap ed o main ain compa able pe o mance en elopes.
17
Sys em-only (in e nal alida ion). Wi hin H3LIX we un ma ched blocks unde
ull and es ic ed awa eness. IU is induced by wi hholding sub le cues and mic o-
e en s om he Symbolic/Noe ic laye s while e aining hei soma ic e ec s. PU is
induced by empo a ily disabling selec ed MPG segmen s o c oss-le el pa hways iden-
i ied a p io i by RV po ency. Fo each ial LAIZA sea ches coun e ac ually o a
Minimal Unawa e Flip Se (MUFS), he smalles se o masked inpu s o abla ed
s uc u es, whose es o a ion e e ses he decision. P ima y eadou s a e: he a e a
which a nonemp y MUFS exis s, he p obabili y o decision lip ela i e o ull, and
ask accu acy/calib a ion as a unc ion o Noe ic cohe ence. MPG - In ui ion is sup-
po ed i es ic ed blocks yield mo e MUFS and mo e lips han ull, ye ials ma ked
as highly cohe en p ese e (o e en imp o e) accu acy and con idence calib a ion
despi e es ic ion.
Human decisions (ex e nal con i ma ion). The e y same s imuli and iming
a e p esen ed o human pa icipan s while he sys em ope a es in pa allel. Human IU
is p oduced wi h s anda d psychophysical me hods (b ie masking, pe iphe al mic o-
cues) e i ied by awa eness checks; PU emains a silen , sys em-in e nal abla ion so
ha he in o ma ion a ailable o he pa icipan is unchanged. We compa e human
choices o sys em choices ac oss ull and es ic ed blocks, ocusing on he incidence
o lips be ween block ypes and on hei alignmen wi h ials in which he sys em
de ec ed a MUFS. Con e gen e idence is ob ained i human decisions lip mo e o en
unde es ic ed han ull, and i hese lips co-occu wi h sys em MUFS ials on he
same i ems, indica ing sha ed dependence on minimal unawa e ac o s.
In ui ion sel - epo linkage. Immedia ely a e each es ic ed ial, pa ici-
pan s p o ide a b ie in ospec i e a ing accompanied by con idence. These epo s
a e ela ed o con empo aneous Noe ic measu es, o MUFS p esence and size
and o beha io al ou comes (accu acy, calib a ion, and la ency). The key p edic-
ion is ha sel -labeled in ui i e ials exhibi highe Noe ic cohe ence, a g ea e
likelihood o con aining a MUFS, and supe io calib a ion/accu acy ela i e o non-
in ui i e ials-speci ically unde es ic ed awa eness, whe e delibe a ion is limi ed
and me a-cogni i e in eg a ion should ma e mos .
7 Conclusions
We ha e p esen ed H3LIX, a ipa i e cogni i e a chi ec u e ha uni ies soma ic
g ounding, symbolic in e ence, and noe ic me acogni ion wi hin a single, ime-
synch onized con ol loop (SORK-N). Coupled wi h he LAIZA p o ocol, he sys em
ansla es adi ionally elusi e phenomena-in ui ion, insigh , an icipa o y awa e-
ness—in o ope a ional quan i ies ha can be measu ed, analyzed, and used o
closed-loop con ol. A he cen e o his amewo k s ands he Mi o ed P o ile
G aph (MPG), an e idence-linked, hie a chical s a e space ha suppo s anspa en
upda es, c oss-le el easoning, and human-audi able p o enance.
A cen al con ibu ion o he wo k is a o maliza ion o su p ise and de ia ion
ia Rogue Va iables (RVs), which iden i y s uc u al pa e ns-segmen s and in e -
le el pa hways wi h ou sized explana o y powe o p edic ion-obse a ion gaps. This
p o ides a p incipled way o p io i ize in e en ions and ela e model dynamics o
18
in e p e able mechanisms. Building on his subs a e, we de ined MPG - In ui ion as
he exis ence o a nonemp y Minimal Unawa e Flip Se (MUFS): a minimal se o
inpu s o g aph s uc u es ou side conscious access whose emo al lips a decision.
This de ini ion shi s “in ui ion” om anecdo e o c i e ion, linking me a-cogni i e
cohe ence o conc e e, es able coun e ac uals.
Me hodologically, we ou lined an expe imen al p og am ha e alua es MPG -
In ui ion unde ull e sus es ic ed awa eness, bo h wi hin he sys em and wi h
human pa icipan s, including b ie sel - epo s o in ui i e expe ience. The p oposed
eadou s-cohe ence spec a, en opy-change coe icien s, MUFS incidence, calib a ion,
and accu acy-allow us o es whe he cohe en s a es p ese e o imp o e decision
quali y when delibe a ion is limi ed, and whe he human “in ui i e” choices align
wi h sys em-de ec ed MUFS on he same i ems. Toge he , hese s udies close he loop
be ween heo y, ma hema ical o malism, and ep oducible e idence.
Concep ually, H3LIX in eg a es mul iple adi ions wi hou collapsing hei dis-
inc ions. F om beha io ism i inhe i s p ecise ein o cemen con ingencies, om
p edic i e p ocessing, e o minimiza ion and unce ain y managemen , om dual-
aspec in o ma ion monism, an on ological b idge be ween ma e and mind and
om noe ic s udies, a willingness o es hypo heses abou non-local cohe ence unde
alsi iable designs. The Mi o Co e ope a ionalizes his syn hesis by w i ing noe ic
me a- eedback back in o s imulus p ocessing (N→S′), enabling bo h i s -o de
(beha io al) and second-o de (me a-cogni i e) adap a ion.
The e a e, howe e , clea limi a ions. Ou o maliza ions do no en ail claims abou
he me aphysical s a us o consciousness; hey speci y decision-le el c i e ia and con-
ol signals wi hin a bounded a chi ec u e. Empi ical e ec s ela ed o p e-e en o
non-local cohe ence, i p esen , a e likely small and con ex -dependen ; igo ous p e-
egis a ion, de ice-le el con ols, and mul i-lab eplica ions will be c ucial. Finally,
e hical conside a ions-p i acy o mul imodal aces, in e p e abili y o s uc u al edi s
in he MPG, and human o e sigh o me a-policy changes and mus accompany any
deploymen beyond he lab.
Despi e hese ca ea s, he amewo k o e s a ac able ou e owa d in eg a i e
cogni ion. By binding physiological signals, symbolic s uc u e, and me acogni i e
cohe ence in o a single, audi able loop, H3LIX p o ides bo h a scien i ic ins umen
o s udying mind and a p ac ical bluep in o human-AI symbiosis. We expec ha
e ining RV po ency measu es, expanding he amily o cohe ence indica o s, and scal-
ing he expe imen al p og am o collabo a i e se ings (mul i-agen MPG echoes) will
u he cla i y when and how “in ui ion” unc ions as a eliable con ol signal. In his
sense, he pa h o wa d is i e a i e: ins umen , es , e ise, and e-ins umen -allowing
biological and a i icial cogni ion o co-e ol e unde a sha ed, alsi iable me hodology.
Re e ences
[1] Misha Ahissa and Elad Assa. Pe cep ion as a closed-loop con e gence p ocess.
Cu en Opinion in Neu obiology, 40:58–65, 2016.
[2] Ga y As on-Jones and Jona han D. Cohen. An in eg a i e heo y o locus
coe uleus–no epineph ine unc ion: Adap i e gain and op imal pe o mance.
19
Annual Re iew o Neu oscience, 28:403–450, 2005.
[3] Hace C. A es, Bade Alizadeh, Ka in Dince , Aysu M. Elma, Elias M. W¨u hne ,
Can Dince , Ge ald A. U ban, and Can Dince . End- o-end design o wea able
senso s. Na u e Re iews Ma e ials, 7(10):887–907, 2022.
[4] Ha ald A manspache . Dual-aspec monism `a la pauli and jung. In Paul Da ies
and Niels Hen ik G ege sen, edi o s, In o ma ion and he Na u e o Reali y: F om
Physics o Me aphysics, pages 260–274. Camb idge Uni e si y P ess, Camb idge,
2010.
[5] Ha ald A manspache . A b ie p ime o dual-aspec monism. Syn hese,
203(9):1–24, 2024.
[6] S ephen H. Bach, Ma hias B oechele , Be Huang, and Lise Ge oo . Hinge-loss
ma ko andom ields and p obabilis ic so logic. Jou nal o Machine Lea ning
Resea ch, 18(109):1–67, 2017.
[7] Da id Bad e. Cogni i e con ol. Annual Re iew o Psychology, 75:1–28, 2024.
[8] Da id Bad e and De ek E. Nee. F on al co ex and he hie a chical con ol o
beha io . Neu on, 99(5):993–1004, 2018.
[9] Tadas Bal uˇsai is, Chai anya Ahuja, and Louis-Philippe Mo ency. Mul imodal
machine lea ning: A su ey and axonomy. IEEE T ansac ions on Pa e n
Analysis and Machine In elligence, 41(2):423–443, 2019.
[10] Lau a Bana escu, Clai e Bonial, Shu Cai, Madalina Geo gescu, Ki a G i i , Ul
He mjakob, Ke in Knigh , Philipp Koehn, Ma ha Palme , and Na han Schnei-
de . Abs ac meaning ep esen a ion o sembanking. In P oceedings o he
7 h Linguis ic Anno a ion Wo kshop and In e ope abili y wi h Discou se, pages
178–186, 2013.
[11] Ch is oph Band and Be nd Pompe. Pe mu a ion en opy: A na u al complexi y
measu e o ime se ies. Physical Re iew Le e s, 88(17):174102, 2002.
[12] Law ence W. Ba salou. G ounded cogni ion. Annual Re iew o Psychology,
59:617–645, 2008.
[13] Timo hy E. Beh ens, Ma k W. Wool ich, Ma k E. Wal on, and Ma hew F. S.
Rushwo h. Lea ning he alue o in o ma ion in an unce ain wo ld. Na u e
Neu oscience, 10(9):1214–1221, 2007.
[14] Da yl J. Bem. Feeling he u u e: Expe imen al e idence o anomalous e oac-
i e in luences on cogni ion and a ec . Jou nal o Pe sonali y and Social
Psychology, 100(3):407–425, 2011.
[15] Aus in R. Benson, Da id F. Gleich, and Ju e Lesko ec. Highe -o de o ganiza ion
o complex ne wo ks. Science, 353(6295):163–166, 2016.
[16] Richa d F. Be zel and Danielle S. Basse . Mul i-scale b ain ne wo ks. T ends in
Cogni i e Sciences, 21(8):589–602, 2017.
[17] Haonan Bian. Llm-empowe ed knowledge g aph cons uc ion: A su ey. a Xi ,
2025.
[18] Vincen D. Blondel, Jean-Loup Guillaume, Renaud Lambio e, and E ienne
Le eb e. Fas un olding o communi ies in la ge ne wo ks. Jou nal o S a is ical
Mechanics: Theo y and Expe imen , (10):P10008, 2008.
[19] E zel Ca de˜na. The expe imen al e idence o pa apsychological phenomena: A
e iew. Ame ican Psychologis , 73(5):663–677, 2018.
20
[20] Da id J. Chalme s. Facing up o he p oblem o consciousness. Jou nal o
Consciousness S udies, 2(3):200–219, 1995.
[21] Andy Cla k. Wha e e nex ? p edic i e b ains, si ua ed agen s, and he u u e
o cogni i e science. Beha io al and B ain Sciences, 36(3):181–204, 2013.
[22] Hugo D. C i chley and Neil A. Ha ison. Visce al in luences on b ain and
beha io . Na u e Re iews Neu oscience, 14(6):394–406, 2013.
[23] S anislas Dehaene, Hakwan Lau, and Sid Kouide . Wha is consciousness, and
could machines ha e i ? Science, 358(6362):486–492, 2017.
[24] Jean-Cha les Del enne, Sophia N. Yali aki, and Mau icio Ba ahona. S abili y o
g aph communi ies ac oss ime scales. P oceedings o he Na ional Academy o
Sciences, 107(29):12755–12760, 2010.
[25] S ephen M. Fleming and Na haniel D. Daw. Sel -e alua ion o decision-making: A
gene al bayesian amewo k o me acogni i e compu a ion. Psychological Re iew,
124(1):91–114, 2017.
[26] S ephen M. Fleming and Hakwan C. Lau. How o measu e me acogni ion.
F on ie s in Human Neu oscience, 8:443, 2014.
[27] San o Fo una o and Da ko H ic. Communi y de ec ion in ne wo ks: A use guide.
Physics Repo s, 659:1–44, 2016.
[28] Ka l J. F is on. The ee-ene gy p inciple: a uni ied b ain heo y? Na u e Re iews
Neu oscience, 11:127–138, 2010.
[29] Malik Ghallab, Dana Nau, and Paolo T a e so. Au oma ed Planning: Theo y and
P ac ice. Mo gan Kau mann, San F ancisco, 2004.
[30] John C. Gi ins, Ke in Glazeb ook, and Richa d Webe . Mul i-A med Bandi
Alloca ion Indices. Wiley, Chiches e , 2 edi ion, 2011.
[31] Tilmann Gnei ing and Ad ian E. Ra e y. S ic ly p ope sco ing ules, p edic ion,
and es ima ion. Jou nal o he Royal S a is ical Socie y: Se ies B, 69(2):243–268,
2007.
[32] Aslak G ins ed, John C. Moo e, and S e lana Je eje a. Applica ion o he c oss
wa ele ans o m and wa ele cohe ence o geophysical ime se ies. Nonlinea
P ocesses in Geophysics, 11(5/6):561–566, 2004.
[33] Adi ya G o e and Ju e Lesko ec. node2 ec: Scalable ea u e lea ning o ne -
wo ks. In P oceedings o he 22nd ACM SIGKDD In e na ional Con e ence on
Knowledge Disco e y and Da a Mining, pages 855–864, 2016.
[34] Da id L. Hall and John Llinas. An in oduc ion o mul isenso da a usion.
P oceedings o he IEEE, 85(1):6–23, 1997.
[35] G ego y P. Hanley, B ian A. Iwa a, and B andon E. McCo d. Func ional analysis
o p oblem beha io : A e iew. Jou nal o Applied Beha io Analysis, 36(2):147–
185, 2003.
[36] S e an Ha nad. The symbol g ounding p oblem. Physica D: Nonlinea
Phenomena, 42(1–3):335–346, 1990.
[37] A iad Hei e z, Ma in Meie , and Bu kha d C. Schippe . In e ac i e unawa eness.
Econome ica, 74(6):1525–1556, 2006.
[38] Aidan Hogan, E a Blomq is , Michael Cochez, Claudia d’Ama o, Ge a d de Melo,
Claudio Gu i´e ez, Jos´e Emilio Lab a Gayo, Sab ina Ki ane, Sebas ian Neu-
maie , Axel Polle es, Robe o Na igli, Axel-Cy ille Ngonga Ngomo, Sebas ian
21
Rudolph, Anna To dai, and Rapha¨el T oncy. Knowledge g aphs. ACM Compu ing
Su eys, 54(4):1–37, 2021.
[39] Pe e Holme and Ja i Sa am¨aki. Tempo al ne wo ks. Physics Repo s, 519(3):97–
125, 2012.
[40] Ki Huck ale, S e ha Venka esh, and Helen Ch is ensen. Towa d clinical digi al
pheno yping: a imely oppo uni y o conside pu pose, quali y, and sa e y. npj
Digi al Medicine, 2:88, 2019.
[41] F ank Hu e , La s Ko ho , and Joaquin Vanscho en, edi o s. Au oma ed
Machine Lea ning: Me hods, Sys ems, Challenges. Sp inge , Cham, 2019.
[42] Shaoxiong Ji, Shi ui Pan, E ik Camb ia, Pekka Ma inen, and S Yu Philip. A
su ey on knowledge g aphs: Rep esen a ion, acquisi ion, and applica ions. IEEE
T ansac ions on Knowledge and Da a Enginee ing, 33(5):312–333, 2021.
[43] Saga Joshi and Joshua I. Gold. Pupil size as a window on neu al subs a es o
cogni ion. T ends in Cogni i e Sciences, 24(6):466–480, 2020.
[44] Leslie Pack Kaelbling, Michael L. Li man, and An hony R. Cassand a. Plan-
ning and ac ing in pa ially obse able s ochas ic domains. A i icial In elligence,
101(1–2):99–134, 1998.
[45] Daniel Kahneman and Ga y Klein. Condi ions o in ui i e expe ise: a ailu e o
disag ee. Ame ican Psychologis , 64(6):515–526, 2009.
[46] Rudol E. Kalman. A new app oach o linea il e ing and p edic ion p oblems.
Jou nal o Basic Enginee ing, 82(1):35–45, 1960.
[47] F ede ick H. Kan e , Hans Reinecke , and Die e Schmelze . Selbs managemen -
The apie: Ein Leh buch ¨u die klinische P axis. Sp inge , Be lin, 5 edi ion,
2012.
[48] F ede ick H. Kan e and Go don Saslow. Beha io al analysis: An al e -
na i e o diagnos ic classi ica ion. In P. H. Hoch and J. Zubin, edi-
o s, The Applica ion o Lea ning P inciples o Clinical P oblems, pages
211–238. G une & S a on, New Yo k, 1965. SORCK model (S imu-
lus–O ganism–Response–Con ingency–Kon ingenz).
[49] F ede ick H. Kan e and Lau ence G. Saslow. Beha io al analysis. Beha io
Modi ica ion, 1(2):159–191, 1977.
[50] Geo ge Ka ypis and Vipin Kuma . A as and high quali y mul ile el scheme o
pa i ioning i egula g aphs. SIAM Jou nal on Scien i ic Compu ing, 20(1):359–
392, 1998.
[51] Da id C. Knill and Alexand e Pouge . The bayesian b ain: he ole o unce ain y
in neu al coding and compu a ion. T ends in Neu osciences, 27(12):712–719,
2004.
[52] Ch is o Koch, Ma cello Massimini, Melanie Boly, and Giulio Tononi. Neu al
co ela es o consciousness: p og ess and p oblems. Na u e Re iews Neu oscience,
17:307–321, 2016.
[53] Daphne Kolle and Ni F iedman. P obabilis ic G aphical Models: P inciples and
Techniques. MIT P ess, Camb idge, MA, 2009.
[54] Jean-Philippe Lachaux, Eugenio Rod iguez, Jacques Ma ine ie, and F ancisco J.
Va ela. Measu ing phase synch ony in b ain signals. Human B ain Mapping,
8(4):194–208, 1999.
22
[55] Sco M. Lundbe g and Su-In Lee. A uni ied app oach o in e p e ing model
p edic ions. In Ad ances in Neu al In o ma ion P ocessing Sys ems 30, pages
4765–4774, 2017.
[56] Ron Milo, Shale I zko i z, Nada Kash an, Reu en Le i , Shai Shen-O , Ilan
Ayzensh a , Michal She e , and U i Alon. Ne wo k mo i s: simple building blocks
o complex ne wo ks. Science, 298(5594):824–827, 2002.
[57] Michael Moo , P´e e Riba, Max Welling, e al. Founda ion models o gene alis
medical a i icial in elligence. Na u e, 616:259–265, 2023.
[58] Luc Mo eau, Paolo Missie , and he W3C PROV Wo king G oup. P o -dm: The
p o da a model. W3C Recommenda ion, 2013. h ps://www.w3.o g/TR/p o -
dm/.
[59] Julia Mossb idge, Pa izio E. T essoldi, and Jessica U s. P edic i e physiological
an icipa ion p eceding seemingly unp edic able s imuli: a me a-analysis. F on ie s
in Psychology, 3:390, 2012.
[60] Ma hew R. Nassa , Ke in T. Wilson, William F. Heasly, and Joshua I. Gold. An
app oxima ely bayesian del a- ule model explains he dynamics o belie upda ing
in a changing en i onmen . P oceedings o he Na ional Academy o Sciences,
107(37):15990–15995, 2010.
[61] M. E. J. Newman. Ne wo ks: An In oduc ion. Ox o d Uni e si y P ess, Ox o d,
2010.
[62] And ew Y. Ng, Michael I. Jo dan, and Yai Weiss. On spec al clus e ing: Anal-
ysis and an algo i hm. In Ad ances in Neu al In o ma ion P ocessing Sys ems,
olume 14, pages 849–856, 2002.
[63] Maximilian Nickel, Ke in Mu phy, Volke T esp, and E geniy Gab ilo ich. A
e iew o ela ional machine lea ning o knowledge g aphs. P oceedings o he
IEEE, 104(1):11–33, 2016.
[64] B ian A. Nosek, Cha les R. Ebe sole, Alexande C. DeHa en, and Da id T. Mel-
lo . The p e egis a ion e olu ion. P oceedings o he Na ional Academy o
Sciences, 115(11):2600–2606, 2018.
[65] Masa umi Oizumi, La issa Alban akis, and Giulio Tononi. F om he phenomenol-
ogy o he mechanisms o consciousness: In eg a ed in o ma ion heo y 3.0. PLOS
Compu a ional Biology, 10(5):e1003588, 2014.
[66] Ge man I. Pa isi, Ronald Kemke , Jose L. Pa , Ch is ophe Kanan, and S e an
We m e . Con inual li elong lea ning wi h neu al ne wo ks: A e iew. Neu al
Ne wo ks, 113:54–71, 2019.
[67] Thomas Pa , Gio anni Pezzulo, and Ka l J. F is on. Ac i e in e ence: wen y
yea s on. Na u e Re iews Neu oscience, 2022.
[68] Judea Pea l. Causali y: Models, Reasoning, and In e ence. Camb idge Uni e si y
P ess, Camb idge, 2 edi ion, 2009.
[69] Jonas Pe e s, Dominik Janzing, and Be nha d Sch¨olkop . Elemen s o Causal
In e ence: Founda ions and Lea ning Algo i hms. MIT P ess, Camb idge, MA,
2017.
[70] Rajesh P. N. Rao and Dana H. Balla d. P edic i e coding in he isual co ex:
a unc ional in e p e a ion o some ex a-classical ecep i e- ield e ec s. Na u e
Neu oscience, 2(1):79–87, 1999.
23
[71] Roge Ra cli and Gail McKoon. The di usion decision model: heo y and da a
o wo-choice decision asks. Neu al Compu a ion, 20(4):873–922, 2008.
[72] Joshua S. Richman and J. Randall Moo man. Physiological ime-se ies analysis
using app oxima e en opy and sample en opy. Ame ican Jou nal o Physiology-
Hea and Ci cula o y Physiology, 278(6):H2039–H2049, 2000.
[73] S ua J. Ri chie, Richa d Wiseman, and Ch is ophe C. F ench. Failing he
u u e: Th ee unsuccess ul a emp s o eplica e bem’s ’ e oac i e acili a ion o
ecall’ e ec . PLOS ONE, 7(3):e33423, 2012.
[74] Ma in Ros all and Ca l T. Be gs om. Maps o andom walks on complex
ne wo ks e eal communi y s uc u e. P oceedings o he Na ional Academy o
Sciences, 105(4):1118–1123, 2008.
[75] Thomas Sch eibe . Measu ing in o ma ion ans e . Physical Re iew Le e s,
85(2):461–464, 2000.
[76] D. Samuel Schwa zkop . We should ha e seen his coming. F on ie s in Human
Neu oscience, 8:332, 2014.
[77] Anil K. Se h. In e ocep i e in e ence, emo ion, and he embodied sel . T ends in
Cogni i e Sciences, 17(11):565–573, 2013.
[78] Anil K. Se h, Adam B. Ba e , and Lionel Ba ne . G ange causali y analysis in
neu oscience and neu oimaging. The Jou nal o Neu oscience, 35(8):3293–3297,
2015.
[79] F ed Sha e and J. P. Ginsbe g. An o e iew o hea a e a iabili y me ics
and no ms. F on ie s in Public Heal h, 5:258, 2017.
[80] Lloyd S. Shapley. A alue o n-pe son games. Con ibu ions o he Theo y o
Games, 2:307–317, 1953.
[81] Hi oshi Shibasaki and Ma k Halle . Wha is he be ei scha spo en ial? Clinical
Neu ophysiology, 117(11):2341–2356, 2006.
[82] Ch is ophe A. Sims. Implica ions o a ional ina en ion. Jou nal o Mone a y
Economics, 50(3):665–690, 2003.
[83] Anne C. Smi h and Eme y N. B own. Es ima ing a s a e-space model om poin
p ocess obse a ions. Neu al Compu a ion, 15(5):965–991, 2003.
[84] Chun Siong Soon, Ma cel B ass, Hans-Jochen Heinze, and John-Dylan Haynes.
Unconscious de e minan s o ee decisions in he human b ain. Na u e Neu o-
science, 11(5):543–545, 2008.
[85] E ik ˇ
S umbelj and Igo Kononenko. Explaining p edic ion models and indi id-
ual p edic ions wi h ea u e con ibu ions. Knowledge and In o ma ion Sys ems,
41(3):647–665, 2014.
[86] Yizhou Sun and Jiawei Han. Mining he e ogeneous in o ma ion ne wo ks: A
s uc u al analysis. IEEE T ansac ions on Knowledge and Da a Enginee ing,
24(12):2507–2530, 2012.
[87] Richa d S. Su on and And ew G. Ba o. Rein o cemen Lea ning: An In oduc-
ion. MIT P ess, Camb idge, MA, 2 edi ion, 2018.
[88] Ch is ophe Timme mann, Leo Roseman, Robin L. Ca ha -Ha is, e al. A neu-
ophenomenological app oach o non-o dina y s a es o consciousness: In eg a ing
i s -pe son examina ions and neu ophysiology. T ends in Cogni i e Sciences,
26(12):1078–1091, 2022.
24
[89] Giulio Tononi, Melanie Boly, Ma cello Massimini, and Ch is o Koch. In eg a ed
in o ma ion heo y: F om consciousness o i s physical subs a e. Na u e Re iews
Neu oscience, 17(7):450–461, 2016.
[90] Cha les T uong, Lau en Oud e, and Nicolas Vaya is. Selec i e e iew o o line
change poin de ec ion me hods. Signal P ocessing, 167:107299, 2020.
[91] F ancisco J. Va ela. Neu ophenomenology: A me hodological emedy o he ha d
p oblem. Jou nal o Consciousness S udies, 3(4):330–349, 1996.
[92] Ma in Vinck, Robe Oos en eld, Ma ijn an Winge den, F anscesco Ba aglia,
and Cy iel M. A. Penna z. An imp o ed index o phase-synch oniza ion o
elec ophysiological da a in he p esence o olume-conduc ion, noise and sample-
size bias. Neu oImage, 55(4):1548–1565, 2011.
[93] E ic-Jan Wagenmake s, Ruud We zels, Denny Bo sboom, and Han L. J. an de
Maas. Why psychologis s mus change he way hey analyze hei da a: The case
o psi (commen on bem, 2011). Jou nal o Pe sonali y and Social Psychology,
100(3):426–432, 2011.
[94] Shuen-De Wu, Chia-Chi Wu, Kuan-Lin Lin, and Chan-Cheng Wang. Time
se ies analysis using e ined composi e mul iscale en opy. Physica A: S a is ical
Mechanics and i s Applica ions, 392(24):5865–5873, 2013.
[95] Angela J. Yu and Pe e Dayan. Unce ain y, neu omodula ion, and a en ion.
Neu on, 46(4):681–692, 2005.
25
