Disse a ion p esen ed o ob ain he
Ph.D deg ee in Biology | Neu oscience
Ins i u o de Tecnologia Química e Biológica An ónio Xa ie | Uni e sidade No a de Lisboa
Oei as,
Decembe , 2016
Gonçalo C. Lopes
A Robus Role
o Mo o Co ex
Gonçalo C. Lopes
Disse a ion p esen ed o ob ain he
Ph.D deg ee in Biology | Neu oscience
Ins i u o de Tecnologia Química e Biológica An ónio Xa ie | Uni e sidade No a de Lisboa
Oei as, Decembe , 2016
A Robus Role
o Mo o Co ex
Resea ch wo k coo dina ed by:
A Robus Role o Mo o Co ex
Gonçalo C. Lopes
A Disse a ion
P esen ed o he Facul y
o Uni e sidade No a de Lisboa
in Candidacy o he Deg ee
o Doc o o Philosophy
Supe iso s:
Joseph J. Pa on
Adam R. Kamp
2016

In lo ing memo y o Kiba
Acknowledgmen s
An imp essi ely la ge numbe o people ha e accompanied me h ough-
ou his jou ney, and o all o hem I am deeply and p o oundly hank ul.
All in e ac ions ha e a meaning we can ne e ecognize and hei combined
impac ipples in o he u u e unaba ed. This is ue e en i I o ge o
publicly acknowledge some o hem, as I'm su e will be he case.
The Champalimaud Neu oscience P og amme has been o me a place
o deep pe sonal and in ellec ual ans o ma ion. I can ba ely ecognize
he pe son ha six yea s ago se ou on he jou ney om compu e science
o neu oscience, in igued by he mys e ies o he b ain. The eo s o
he ins i u e as a whole o b ing oge he people om widely die en cul-
u al and academic backg ounds has c ea ed a nexus in which in e es ing
pe sonali ies canno help bu be o ged and empe ed. I will ne e o ge
he unique oppo uni y I had o in e ac wi h such a la ge communi y o
scien is s om all o e he wo ld, and o discuss nea ly e e y possible opic
o my u mos sa is ac ion. These in e ac ions we e always done eely and
passiona ely, in he bes spi i o scien ic companionship, wi h no ega d
o hie a chy o ank.
My  s dedica ion goes o my companions o he In e na ional Neu -
oscience Doc o al P og amme, Ca olina Do an, Simone Lackne , Tiago
Ma ques, I o Ma celo, B uno Mi anda, Raimundo Leong, and Gus a o
Mo eno, who in he yea o 2010 emba ked wi h me on his ad en u e.
Toge he we ha e sha ed much mo e han jus ou ini ial aining in neu -
oscience. E en hough we all pa ed o pu sue ou own indi idual p ojec s,
he e we e many c ucial poin s du ing he jou ney whe e, e en se endip-
i ously, we we e s ill able o suppo each o he as pilla s o sani y in he
middle o uncon ollable u bulence. I was an honou and a pleasu e o
ha e me each one o you and I hope o ha e con ibu ed back a small
inkling o all he inspi a ion and admi a ion you ha e p o ided me.
ix
Financial Suppo
The esea ch leading o hese esul s has ecei ed unding om he
Eu opean Union's Se en h F amewo k P og amme (FP7/2007-2013) un-
de g an ag eemen no. 600925 and he Bial Founda ion (G an 190/12).
GL was suppo ed by he PhD S uden ship SFRH/BD/51714/2011 om
he Founda ion o Science and Technology, and a Visi ing Resea ch Fellow
s ipend om he Sainsbu y Wellcome Cen e o Neu al Ci cui s and Be-
ha iou a Uni e si y College London. The Champalimaud Neu oscience
P og amme is suppo ed by he Champalimaud Founda ion.
x i

Con en s
Acknowledgemen s
........................ ix
Tí ulo e Resumo
..........................xiii
Abs ac
............................... x
Financial Suppo
.........................x i
1 Towa ds a Teleology o Co ical Mo o Con ol 1
1.1 Chap e Summa y ....................... 2
1.2 A Dilemma o Co ical Mo o Con ol . . . . . . . . . . . . 3
1.2.1 Disco e y o he Mo o Co ex . . . . . . . . . . . . 3
1.2.2 The Gol z-Fe ie Deba es . . . . . . . . . . . . . . . 4
1.3 The Role o he Co icospinal T ac . . . . . . . . . . . . . . 8
1.3.1 A Func ional Theo y o he Mo o Co ex . . . . . . 9
1.3.2 The Eec s o Lesions in he Co icospinal T ac . . 13
1.4 An In eg a i e View o he Mo o Sys em . . . . . . . . . . 20
1.4.1 The Coo dina i e Role o Inhibi ion . . . . . . . . . 22
1.4.2 Hie a chical O ganiza ion o Mo o Beha iou . . . . 25
1.5 A S a egy o P obing Co ical Con ol . . . . . . . . . . . 30
1.5.1 Ou line o he Thesis . . . . . . . . . . . . . . . . . . 32
2 Rapid P o o yping Tools o he S udy o Beha iou 47
2.1 Chap e Summa y ....................... 48
2.2 In oduc ion........................... 49
x ii
2.3 Me hods............................. 51
2.3.1 The Modula Beha iou Box . . . . . . . . . . . . . 51
2.3.2 The Bonsai F amewo k . . . . . . . . . . . . . . . . . 55
2.4 Resul s.............................. 71
2.4.1 Ge ing S a ed wi h Bonsai . . . . . . . . . . . . . . 71
2.4.2 Unde he Hood . . . . . . . . . . . . . . . . . . . . 72
2.4.3 Applica ions . . . . . . . . . . . . . . . . . . . . . . . 74
2.5 Discussion............................ 79
2.6 Acknowledgemen s . . . . . . . . . . . . . . . . . . . . . . . 82
2.7 Au ho con ibu ions . . . . . . . . . . . . . . . . . . . . . . 83
3 Mo ing wi h and wi hou Mo o Co ex 87
3.1 Chap e Summa y ....................... 88
3.2 In oduc ion........................... 89
3.3 Me hods............................. 91
3.4 Resul s..............................103
3.5 Discussion............................121
3.6 Acknowledgemen s . . . . . . . . . . . . . . . . . . . . . . . 122
3.7 Au ho con ibu ions . . . . . . . . . . . . . . . . . . . . . . 123
4 Ex ended Discussion 129
4.1 Chap e Summa y .......................130
4.2 A challenge om obo ics . . . . . . . . . . . . . . . . . . . 131
4.3 A p imo dial ole o mo o co ex . . . . . . . . . . . . . . 133
4.4 Implica ions o non-p ima e mammals . . . . . . . . . . . . 133
4.5 Implica ions o p ima e s udies . . . . . . . . . . . . . . . . 134
4.6 Some specula ion on he ole o di ec co ical con ol . . . 135
4.7 Some p elimina y conclusions . . . . . . . . . . . . . . . . . 136
x iii
Chap e 1
Towa ds a Teleology o
Co ical Mo o Con ol
The inni e e ili y o he o ganism as a eld o adap ed
eac ions has become mo e appa en . The pu pose o a eex
seems as legi ima e and u gen an objec o na u al inqui y as
he pu pose o he colou ing o an insec o a blossom. And he
impo ance o physiology is, ha he eex eac ion canno be
eally in elligible o he physiologis un il he knows i s aim.
Si Cha les S. She ing on
,
The In eg a i e Ac ion o he
Ne ous Sys em
(1906)
1
1.1 Chap e Summa y
Mo o co ex has 150 yea s o conic ing his o y. I was o iginally dened
as he pa o co ex whe e mo emen s can be e oked by low-cu en s im-
ula ion. S imula ed poin s ac oss he co ical su ace we e ound o be
o ganized in o a ough map o he body. A monosynap ic p ojec ion sys-
em, he co icospinal ac , was ound o di ec ly link mo o co ex o
neu ons in con ala e al spinal co d. The e is a ma ked dec ease in g anu-
la laye IV, he main sou ce o inpu s om senso y halamus, ac oss his
exci able zone. Lesions o he mo o co ex in humans can cause pa alysis
in any limb and pe manen ly dis up he execu ion o basic ac ions such
as s anding o walking. Reco dings o neu al ac i i y in mo o co ex co -
ela e wi h a ious mo emen pa ame e s. These lines o e idence suppo
he idea ha his pa o he b ain di ec ly con ols mo emen .
Howe e , lesions o he mo o co ex in non-human animals p ese e
mos o he animal's beha iou epe oi e. Sec ioning o he co icospinal
ac in p ima es is sucien o eins a e he p ima y eec s o mo o
co ical lesions, bu he e a e also as p ojec ions om mo o co ex o
o he co ical and sub-co ical a eas, including mul iple disynap ic pa allel
descending pa hways o spinal cen e s ia b ains em. In mos mammals,
he co icospinal ac does no a ge mo o neu ons in en al spinal co d,
as i does in mos p ima es, bu a he spinal in e neu ons.
I is clea ha his pa o he b ain is somehow in ol ed in mo emen ,
bu a la ge numbe o ques ions emain s angely unanswe ed. I mo o
co ex is a con olle , wha kind o mo emen s does i con ol? How does i
in e ac wi h o he exis ing b ain s uc u es o gene a e beha iou ? Why
do mo o co ical lesions p oduce such an appa en ly incomple e eec on
mo emen ? This chap e is an a emp o piece oge he all he agmen -
a y and con adic o y e idence on mo o co ical s uc u e and physiology
in o de o de i e a unied unc ional pic u e o co ical mo o con ol.
2
1.2 A Dilemma o Co ical Mo o Con ol
The in ol emen o he b ain and spinal co d in mo o con ol has been
ecognized since he ea lies known clinical eco ds on head and spinal in-
ju y, da ing back o ancien Egyp (Louis, 1994; an Middendo p, Sanchez
& Bu idge, 2010). Howe e , he ole o he ne ous sys em in gene a ing
beha iou was no ully app ecia ed un il Gal ani  s epo ed his amous
expe imen s on
animal elec ici y
(Gal ani, 1791). By isola ing he scia ic
ne e and gas ocnemius muscle in he og, Gal ani clea ly demons a ed
in a se ies o s imula ion expe imen s ha an elec ical p ocess, con ained
en i ely wi hin he biology o he og's leg, was esponsible o he spon-
aneous gene a ion o muscle con ac ions. This would lead o e he ol-
lowing cen u y o he disco e y and physiological cha ac e iza ion o he
ne e impulse, he ac ion po en ial, ha a els ac oss he ne e o ini i-
a e muscle mo emen (du Bois-Reymond, 1843; Be ns ein, 1868; Schue ze,
1983). The success o hese seminal expe imen s immedia ely aised a
undamen al ques ion ega ding ne e conduc ion: i spon aneous muscle
con ac ion is gene a ed by ne e impulses ansmi ed h oughou he
ne ous sys em, how is his ansmission coo dina ed in o de o gene a e
he complex pa e ns o muscle ac i i y obse ed in na u al beha iou ?
1.2.1 Disco e y o he Mo o Co ex
In sea ch o answe s o his ques ion, many esea che s looked a he b ain,
he sea o ana omical con e gence o he ne ous sys em, o such an in-
eg a i e ole. Following Gal ani's oo s eps, se e al a emp s we e made
o s imula e he ce eb al co ex elec ically, bu wi h li le success (G oss,
2007). I wasn' un il he 1870s ha he  s indica ions o a di ec in-
ol emen o he co ex in he p oduc ion o mo emen came o ligh ,
a ound he ime when Hughlings Jackson unde wen his s udies on epi-
lep ic con ulsions (Jackson, 1870). He obse ed ha in some pa ien s he
3

 s would s a by a delibe a e spasm on one side o he body, and ha
die en body pa s would become sys ema ically aec ed one a e he
o he . He connec ed he o de ly ma ch o hese spasms o he exis ence o
localized lesions in he
pos -mo em
b ain o his pa ien s and hypo hesized
ha he o igin o hese  s was uncon olled exci a ion caused by local
changes in co ical
g ey ma e
(Jackson, 1870). In ha same yea , F i sch
and Hi zig published hei amous s udy demons a ing ha i is possible
o elici mo emen s by di ec s imula ion o he co ex in dogs (F i sch
& Hi zig, 1870). Fu he mo e, s imula ion o die en pa s o he co ex
p oduced mo emen in die en pa s o he body (F i sch & Hi zig, 1870).
I appea ed ha he causal mechanism o epilep ic con ulsions p edic ed
by Hughlings Jackson had been ound, and wi h i a possible explana ion
o how he no mal b ain migh con ol mo emen . The ce eb al co ex
was al eady conside ed a he ime o be he sea o easoning and sensa-
ion, so i ac i i y o e his so-called
mo o co ex
was able o exe di ec
con ol o e he whole muscula u e o he body, hen i migh ep esen
in he no mal b ain he a ea ha connec s oli ion o muscles (F i sch &
Hi zig, 1870).
1.2.2 The Gol z-Fe ie Deba es
Da id Fe ie , a Sco ish neu ologis deeply imp essed by he ideas o
Hughlings Jackson and by he posi i e esul s o F i sch and Hi zig's ex-
pe imen s, p oceeded o ep oduce and expand on hei obse a ions wi h
comp ehensi e s imula ion s udies showing how ac i i y in he mo o co -
ex was sucien o p oduce a la ge a ie y o mo emen s ac oss a wide
ange o mammalian species (Fe ie , 1873). Meanwhile, o he esea ch-
e s ac oss Eu ope such as Gol z and Ch is iani we e acing a dilemma:
in many o he so-called lowe mammals massi e lesions o he ce eb al
co ex ailed o demons a e any isible long- e m impai men s in he mo-
o beha iou o animals (James, 1885; Gol z, 1888). These wo lines o
4
inqui y  s clashed a he se en h In e na ional Medical Cong ess held in
London in Augus 1881, whe e Gol z o S assbu g and Fe ie o London
p esen ed hei esul s in a se ies o deba es on he localiza ion o unc ion
in he ce eb al co ex (Phillips, Zeki & Ba low, 1984; Tyle & Malessa,
2000).
Gol z assumed a clea an i-localiza ionis posi ion. He ad anced ha
i was impossible o p oduce a comple e pa esis o any muscle, o comple e
dys unc ion o any pe cep ion, by des uc ion o any pa o he ce eb al
co ex, and ha he ound mos ly deci s o gene al in elligence in his dogs
(Tyle & Malessa, 2000). Following Gol z's p esen a ion, Fe ie emphas-
ized he dange o gene alizing om he dog o animals o o he o de s (e.g.
man and monkey). He hen p oceeded o exhibi his own lesion esul s by
means o an isep ic su ge y in he monkey, desc ibing how a ci cumsc ibed
unila e al lesion o he mo o co ex p oduced comple e con ala e al pa a-
lysis o he leg. He also p oduced a s iking se ies o mic oscopic sec ions o
Walle ian degene a ion (Walle , 1850) o he mo o pa h om he co ex
o he con ala e al spinal co d, he c ossed descending p ojec ions o ming
he py amidal co icospinal ac (Tyle & Malessa, 2000).
The deba es concluded wi h he public demons a ion o li e specimens:
a dog wi h la ge lesions o he pa ie al and pos e io lobes om Gol z; and
om Fe ie , a hemiplegic monkey wi h a unila e al lesion o he mo o
co ex o he con ala e al side. As p edic ed, Gol z's dog showed a clea
abili y o locomo e and a oid obs acles and o make use o i s o he basic
senses, while displaying peculia deci s o in elligence such as ailing o
espond wi h ea o he c acking o a whip o igno ing obacco smoke
blown o i s ace. On he o he hand, Fe ie 's monkey showed up se e ely
hemiplegic, in a condi ion simila o human s oke pa ien s. A e he
demons a ions, he animals we e killed and hei b ains emo ed. P e-
limina y obse a ions e ealed ha he lesions in Gol z's dog we e less
ex ensi e han expec ed, pa icula ly on he le hemisphe e. Fe ie 's le-
5
sions on he o he hand we e p ecisely ci cumsc ibed o he con ala e al
mo o co ex. These demons a ions secu ed he iumph o Fe ie , who
wen on o  mly es ablish he localiza ionis app oach o neu ology and
he idea o a soma o opic a angemen o e he mo o co ex.
The Gol z-Fe ie deba es had a - eaching implica ions h oughou he
en i e esea ch communi y o he ime, and he basic dilemma ha was
p esen ed has spa ked con o e sy and con usion o o e a hund ed yea s
since (Phillips e al., 1984; Lashley, 1924; de Ba enne, 1933; Tyle &
Malessa, 2000; G oss, 2007). In he mean ime, iews o mo o co ex ha e
e ol ed o sugges i plays a ole in unde s anding he mo emen s o
o he s (Rizzola i & C aighe o, 2004), imagining one's own mo emen s
(Po o e al., 1996), o in lea ning new mo emen s (Kawai e al., 2015),
bu whe e a e we oday ega ding i s sugges ed p ima y ole in di ec ly
con olling mo emen ?
S imula ing mo o co ex causes mo emen ; mo o co ex is ac -
i e du ing mo emen
Mo o co ex is s ill b oadly dened as he egion o he ce eb al hemi-
sphe es om which mo emen s can be e oked by low-cu en s imula-
ion, ollowing F i sch and Hi zig's o iginal expe imen s in 1870 (F i sch &
Hi zig, 1870). S imula ing die en pa s o he mo o co ex elici s mo e-
men in die en pa s o he body, and sys ema ic s imula ion su eys
ha e e ealed a opog aphical ep esen a ion o he en i e skele al muscu-
la u e ac oss he co ical su ace (Ley on & She ing on, 1917; Peneld
& Bold ey, 1937; Nea sey e al., 1986). Elec ophysiological eco dings in
mo o co ex ha e ou inely ound co ela ions be ween neu al ac i i y and
many die en mo emen pa ame e s, such as muscle o ce (E a s, 1968),
mo emen di ec ion (Geo gopoulos, Schwa z & Ke ne , 1986), speed
(Schwa z, 1993), o e en aniso opic limb mechanics (Sco , G ibble, G a-
ham & Cabel, 2001) a he le el o bo h single neu ons (E a s, 1968;
6
Chu chland & Shenoy, 2007) and popula ions (Geo gopoulos e al., 1986;
Chu chland e al., 2012). De e mining wha exac ly his ac i i y in mo o
co ex con ols (Todo o , 2000) has been u he complica ed by s ud-
ies using long s imula ion du a ions in which con inuous s imula ion a a
single loca ion in mo o co ex e okes complex, mul i-muscle mo emen s
(G aziano, Taylo & Moo e, 2002; Aalo & G aziano, 2006). Howe e , as
a whole, hese obse a ions all suppo he long s anding iew ha ac i i y
in mo o co ex is in ol ed in he di ec con ol o mo emen .
Mo o co ex lesions p oduce die en deci s in die en species
Wha ypes o mo emen equi e mo o co ex? In humans, a mo o co -
ical lesion is de as a ing. Pe manen inju y o he on al lobes o he
b ain by s oke o mechanical means is o en ollowed by weakness o
pa alysis o he limbs in he side o he body opposi e o he lesion (Louis,
1994). Al hough he pa e ic symp oms ha e a endency o eco e pa -
ially by hemsel es, especially wi h aining and ehabili a ion, pe man-
en mo emen deci s and loss o muscle con ol in he aec ed limbs is
he common p ognosis; mo emen is pe manen ly and ob iously impai ed
(Laplane, Talai ach, Meininge , Bancaud & Boucha eine, 1977; Kwakkel,
Kollen, an de G ond & P e o, 2003). In non-human p ima es, sim-
ila g oss mo emen deci s a e obse ed a e lesions, albei ansien ly
(Ley on & She ing on, 1917; T a is, 1955). The longes las ing eec o a
mo o co ical lesion is he dec eased mo ili y o dis al o elimbs, especially
in he con ol o indi idual nge mo emen s equi ed o p ecision skills
(Ley on & She ing on, 1917; Da ling, Pizzimen i & Mo ec a , 2011). Bu
equally imp essi e is he ex en o which o he mo emen s ully eco e ,
including he abili y o si , s and, walk, climb and e en each o g asp, as
long as p ecise nge mo emen s a e no equi ed (Ley on & She ing on,
1917; Da ling e al., 2011; Zaaimi, Edgley, So e opoulos & Bake , 2012).
In non-p ima e mammals, he absence o las ing deci s ollowing mo o
7
Tha he py amidal ac s a e in he dog equisi e o oli-
ional impulses o each limbs and body seems nega i ed by
he ac ha he animal can un, leap, u n o ei he side, use
neck and jaws, &c. wi h ease and success a e nea ly, i no
wholly, comple e degene a ion o hese ac s on bo h sides.
Fu he , a e comple e degene a ion o one py amid, he e is
in he dog no ob ious die ence be ween he mo emen s o he
igh and le sides. (She ing on, 1885, p.189)
In e es ingly, he does no e ha `de ec o mo ion is obse able only as
a clumsiness in execu ion o ne mo emen s' (She ing on, 1885). These
obse a ions once again s ood ou in s a k con as wi h lesion expe imen s
epo ed by Fe ie in he monkey, whe e cau e iza ion o specic mo o
co ical a eas p oduced comple e and pe sis en pa alysis o he co es-
ponding body pa s (Fe ie & Yeo, 1884).
Yea s la e , She ing on would come back o he mo o co ex wi h a
new se o landma k s udies on s imula ion and abla ion o he p ecen al
egion (G ünbaum & She ing on, 1903; G aham B own & She ing on,
1913; Ley on & She ing on, 1917). In hese s udies oge he wi h G ün-
baum, She ing on a ge ed mo o co ical lesions o he exci able a ea
o he a m o he leg and acked he eco e y o he animals o e ime.
Following he ini ial pa esis and loss o muscle con ol hey obse ed d a-
ma ic eco e y o mos skilled mo o ac s, such as peeling open a banana
o climbing cages (Ley on & She ing on, 1917). In o de o es whe he
he eco e y p ocess was due o co ical eo ganiza ion, hey sys ema ic-
ally s imula ed he a eas adjacen o he lesion as well as he mo o co ex
o he opposi e hemisphe e, bu ailed o e oke mo emen s in he aec ed
limb (Ley on & She ing on, 1917), as would be expec ed i commands
we e a eling down he co icospinal ac in spa ed egions. Fu he mo e,
subsequen abla ion o hose a eas ailed o p oduce any new impai men s
14

in he eco e ed limb, lea ing She ing on and his colleagues a a loss o
nd he locus o eco e y (Ley on & She ing on, 1917).
Con used by hese esul s, which hey hough caused conce n o, s u-
den s o ce eb al physiology, Glees and Cole in oduced a se o mo e
quan i a i e beha iou al assays in he hope o acking in de ail he e-
co e y o mo o con ol (Glees & Cole, 1950; Cole, 1952). They s udied
he beha iou o monkeys sol ing a ious puzzle boxes ollowing successi e
ci cumsc ibed lesions o he humb, index and a m a eas o he mo o co -
ex. As She ing on epo ed, he e was a quick eco e y a e an ini ial
pe iod o pa alysis and loss o mo o con ol. Howe e , e en hough he
monkeys ully eco e ed hei abili y o skill ully open he puzzle box, some
sub le mo emen deci s and pa esis in he con ol o ne mo emen s o
he digi s was epo ed o pe sis (Glees & Cole, 1950). When s imula -
ing mo o co ical a eas su ounding he ci cumsc ibed lesions, hey we e
able o e oke mo emen s in he impac ed digi s and eins a e he pa e ic
symp oms a e u he abla ion (Glees & Cole, 1950). This sugges ed
he hypo hesis ha su ounding a eas o he mo o co ex could unde go
eo ganiza ion ollowing he lesion. Howe e , an impo an die ence o
emphasize be ween hese expe imen s and hose o She ing on is he ac
ha only ela i ely ci cumsc ibed mo o co ical egions we e emo ed in
each su ge y, whe eas in he o iginal She ing on s udy he en i e elbow,
w is , index, humb and emaining digi mo o a eas we e excised a once
(Ley on & She ing on, 1917), mos likely causing degene a ion o he en-
i e co icospinal pa hway o he aec ed limb. The p esence o an in ac
co icospinal ac , exci abili y o mo emen s o low-cu en s imula ion
and ansien pa e ic symp oms ollowing abla ion hus seem o go hand
in hand.
In he hopes o cla i ying he con usion o which exac mo emen s we e
con olled by co ex, o he s udies ocused on lesions es ic ed o he co -
icospinal ac , using bo h unila e al and bila e al sec ion a he le el
15
o he medulla y py amids (Towe , 1940; Law ence & Kuype s, 1968a;
Law ence & Kuype s, 1968b). The goal was o isola e he eec s o all he
indi idual descending pa hways o he spinal co d and esol e once and
o all he ques ion o whe he he co icospinal ac o he mo o co ex
was he sou ce o all  olun a y mo emen s. Sa ah Towe was he  s o
desc ibe in de ail he esul s o unila e al and bila e al py amido omy in
p ima es, wi h and wi hou lesion o he mo o co ex (Towe , 1940). She
summa ized he condi ion as hypo onic pa esis, cha ac e ized by a loss o
skele al muscle one and dep ession o he asomo o sys em, along wi h
gene al weakening o he eexes in ol ing he aec ed limb segmen s. Al-
hough all disc e e usage o he hand and digi s was elimina ed, she did
emphasize he clea p esence o olun a y mo emen s in he a ious pu -
pose ul compensa ions p oduced by he animals o deal wi h he aic ion.
Towe a ibu ed hese compensa ions o he p ese ed capaci ies o b ain-
s em ci cui s.
A mo e deni i e s udy o dissocia e he eec s o di ec co icospinal
and indi ec b ains em descending pa hways was conduc ed by Law ence
and Kuype s, and p esen ed in hei now classical publica ions (Law ence
& Kuype s, 1968a; Law ence & Kuype s, 1968b). Using he Klü e boa d,
a ask whe e monkeys ha e o pick mo sels o ood om die en ly sized
ound holes, hey obse ed ha while no mal monkeys ou inely pick up
he ood by pinching indi idual bi s wi h hei nge s, monkeys wi h bila -
e al co icospinal lesions we e mos ly unable o pe o m his p ecise pince
mo emen , and ins ead employed coa se compensa o y clasping s a egies
o e ie e he ood (Law ence & Kuype s, 1968a). In addi ion, lesioned
monkeys we e consis en ly epo ed o be somewha slowe and less agile
han no mal animals. Howe e , mos o hei o e all mo emen epe oi e
was su p isingly p ese ed. Thei nal conclusions  ema kably well
wi h he ini ial obse a ions o She ing on in he dog, sugges ing ha
he co icospinal pa hways supe impose speed and agili y on subco ical
16
mechanisms, and p o ide he capaci y o ac iona ion o mo emen s such
as independen nge mo emen s (Law ence & Kuype s, 1968a). These
obse a ions ecapi ula e he eec s o mo o co ical lesions epo ed by
She ing on, bu emain a odds wi h he p ima y s a ed ole in ol ing
mo o co ex, and he di ec co icospinal ac , wi h he con ol o all
olun a y mo emen s.
The e a e ana omical die ences in co icospinal p ojec ions
be ween p ima es and o he mammals
In p ima es, he conspicuous eec s o mo o co ical lesion can also be
induced by sec ioning he co icospinal ac , he di ec monosynap ic p o-
jec ion ha connec s mo o co ex, and o he co ical egions, o he spinal
co d (Towe , 1940; Law ence & Kuype s, 1968a). In monkeys, and sim-
ila ly in humans, his pa hway has been ound o di ec ly e mina e on
spinal mo o neu ons esponsible o he con ol o dis al muscles (Ley on
& She ing on, 1917; Be nha d & Bohm, 1954) and is also hough o sup-
po he low-cu en mo emen esponses e oked by elec ical s imula ion
o he co ex, as e idenced by he inc eased dicul y in ob aining a s imu-
la ion esponse ollowing sec ion a he le el o he medulla (Woolsey e al.,
1972).
Howe e , he co icospinal ac is by no means he only pa hway om
co ex o mo emen (Figu e 1.2). Mo o co ex a ge s many o he b ain
egions ha can hemsel es gene a e mo emen . In ac , his specialized
connec ion om elencephalon o spinal co d appea ed only ecen ly in
e eb a e e olu ion ( en Donkelaa , 2009), and was u he elabo a ed o
include a di ec connec ion om co ex o mo o neu ons only in some
p ima e species and o he highly manipula i e mammals such as accoons
(Hene & Mas e on, 1983). In all o he mammals, including ca s and
a s, he e mina ion pa e n o he co icospinal ac la gely a oids he
mo o neu on pools in en al spinal co d and concen a es ins ead on
17
in e media e zone in e neu ons and do sal senso y neu ons (Kuype s, 1981;
Yang & Lemon, 2003). Why hen is he e such a la ge dependency on his
ac o human mo o con ol? One possibili y is ha he ub ospinal
ac a descending pa hway o igina ing in he b ains em and e mina ing
in he in e media e zoneis degene a ed in humans compa ed o o he
p ima es and mammals (Na han & Smi h, 1955, 1982), and is hough
o play a ole in compensa ing o he loss o he co icospinal ac in
non-human species (Law ence & Kuype s, 1968b; Zaaimi e al., 2012).
I hus seems likely ha mos mammals ely on indi ec  pa hways o
con ey co ical mo o commands o muscles. These die ences in ana omy
migh explain he lack o conspicuous, las ing mo emen deci s ollow-
ing mo o co ical lesion in non-p ima es, bu lea es behind a signican
ques ion: wha is he mo o co ex ac ually con olling in all hese o he
mammals?
Wha is he ole o mo o co ex in non-p ima e mammals?
In he a , a la ge po ion o co ex is conside ed mo o  based on ana-
omical (Donoghue & Wise, 1982), s imula ion (Donoghue & Wise, 1982;
Nea sey e al., 1986) and elec ophysiological e idence (Hyland, 1998).
Howe e , he mos consis en ly obse ed long- e m mo o con ol deci
ollowing mo o co ical lesion has been an impai men in supina ion o he
w is and indi idua ion o digi s du ing g asping, which in u n impai s
eaching o ood pelle s h ough a na ow e ical sli (Whishaw e al.,
1991; Ala e dash ili & Whishaw, 2008). Despi e he ac ha ac i i y in
oden mo o co ex has been co ela ed wi h mo emen s in e e y pa o
he body (no jus dis al limbs) (Hill, Cu is, Moo e & Klein eld, 2011;
E lich, Bialek & B ody, 2011), i would appea we a e led o conclude
ha his la ge high-le el mo o s uc u e, wi h dense ee en p ojec ions
o mo o a eas in he spinal co d (Kuype s, 1981), basal ganglia (Tu ne
& DeLong, 2000; Wu, Co win & Reep, 2009), halamus (Lee, Ca ell &
18
~ 110BmillionByea s
~ 560BmillionByea s
Human
Fo eb ain
In e b ain
Midb ain
SpinalBco d
Hindb ain
Mo o
Pallium
S ia um
Thalamus
Tec um
MLR
Ce ebellum
B ains em
Do sal
Ven al
MM
Roden
Mo o
Pallium
S ia um
Thalamus
Tec um
MLR
Ce ebellum
B ains em
Do sal
Ven al
MM
Lamp ey
Mo o
Pallium
S ia um
Thalamus
Tec um
MLR
Ce ebellum
B ains em
?
Do sal
Ven al
MM
Figu e 1.2.
Fo eb ain mo o con ol pa hways ac oss die en e eb a e
axa. The molecula di e gence imes be ween human (p ima e), oden
and lamp ey g oups (Kuma & Hedges, 1998) a e no ed abo e a schema ic
iew o he majo di isions in he e eb a e b ain. A ows indica e he
descending monosynap ic p ojec ions iden ied in each g oup om mo o
egions o he o eb ain pallium o lowe mo o cen es. No e he special-
ized monosynap ic p ojec ion di ec ly a ge ing spinal mo o neu ons in
human. MLR, Mesencephalic Locomo o Region; M, Mo o Neu ons.
19

Simons, 2008), ce ebellum (Bake , Ja id & Edgley, 2001) and b ains em
(Ja a & Hyland, 1999), as well as o mos p ima y senso y a eas (Pe -
eanu e al., 2012; Schneide , Nelson & Mooney, 2014), e ol ed simply o
acili a e mo e p ecise w is o a ions and g asping ges u es. Maybe we
a e missing some hing. Migh he e be o he p oblems in mo emen con-
ol ha mo o co ex is sol ing, bu ha we may be o e looking wi h ou
cu en assays?
1.4 An In eg a i e View o he Mo o Sys em
A die en app oach o he p oblems o mo o con ol de eloped ini ially
om s udies on he in eg a ion o spinal eexes conduc ed by he She ing-
on school. While many esea che s con inued o look o he in eg a ion o
complex mo emen s in highe b ain s uc u es like he mo o co ex, She -
ing on u ned ins ead o sys ema ically cha ac e izing ana omically and
physiologically he dis ibu ion o ee en (She ing on, 1892) and ae -
en (She ing on, 1893a) ne e oo s in he spinal co d o mul iple species.
His goal was o shed ligh on he so-called
eex a c
, he ne e pa hways
in ol ed in muscula eac ions like he knee-je k whe eby simple senso y
s imuli elici an immedia e, au oma ic esponse om he animal, e en in
he absence o highe b ain inpu (She ing on, 1893b).
She ing on and his con empo a ies s udied in de ail a numbe o long
and sho spinal eexes
3
in a a ie y o model o ganisms unde die en
le els o anes hesia, pha macological manipula ions and spinal anssec ion
(She ing on & Lasle , 1903). This sys ema ic app oach made abundan ly
clea a numbe o ac s abou how he ne ous sys em o ganizes mo o
beha iou .
3
A eex ac ion in which a s imulus applied o one egion elici s a esponse in ano he
egion is e med a
long spinal
eex, whe eas a eex eac ion whe e he muscula
esponse happens in he same egion as he s imulus is e med a
sho spinal
eex.
20
The  s one, and pe haps he mos s iking, is ha complex mo o
esponses can be in eg a ed and coo dina ed e en in he comple e absence
o he b ain (She ing on, 1906). While he exis ence o au oma isms and
xed ac ion pa e ns had been ecognized since an iqui y, sys ema ic s im-
ula ion s udies in dece eb a e animals quickly e ealed ha he eex was
a om being a igid and xed en i y, bu was a he adap i e and dy-
namic. In pa icula , eex ci cui s e ealed a much wide ange o e-
sponse cha ac e is ics han ne e be s, which we e well known since he
ime o Gal ani o exhibi comple e s e eo ypy in hei esponse o a s im-
ulus unde a ious condi ions
4
.
Indeed, he mo o ou pu p oduced by he massi ely simplied spinal
ci cui s was ema kably o ganized and displayed clea e hological mean-
ing: adap i e beha iou s such as sc a ching (She ing on & Lasle , 1903),
shaking (Gol z & Ewald, 1896; She ing on & Lasle , 1903) o eex s ep-
ping and s anding (She ing on, 1910; She ing on, 1915) we e all a ailable
o be elici ed om s imula ion o he isola ed spinal sys em. S ikingly,
hese eexes we e also shown o be deployed and modula ed app op i-
a ely o specic s imuli. The sc a ch eex, o example, ca ies he oo
oughly o he place o s imula ion (She ing on, 1904), and in eex s ep-
ping he animal can main ain a hy hmic ma ch h ough all phases o
locomo ion o e unobs uc ed su aces (She ing on, 1910). In eg a ion o
hese eexes wi h inpu om he ele ecep o s is ob iously en i ely ab-
sen , bu hese obse a ions cla ied, beyond any easonable doub , ha
spinal co d ci cui s alone a e sucien o p oduce and sus ain en i e beha-
iou sequences unde he igh condi ions. Fu he mo e, deae en a ion
expe imen s showed ha aspec s o hese hy hmic ne wo k mo i s pe sis
4
Some unique esponse cha ac e is ics o eex a c conduc ion include i e e sibili y
o he di ec ion o conduc ion; a igabili y and e ac o y pe iod; g ea e a iabili y
o h eshold; empo al acili a ion wi h successi e s imuli; a weake co espondence o
end-eec wi h in ensi y and equency o he s imulus; and a g ea e suscep ibili y o
me abolic and pha macological manipula ions (She ing on, 1906, p.14)
21
e en in he absence o senso y inpu (G aham B own, 1911). Many o
hese eex ci cui s we e la e e med
cen al pa e n gene a o s
, o CPGs
(G illne & Zangge , 1975; G illne , 1981), and ound o be p esen ac oss
bo h e eb a e and in e eb a e species (O lo sky, Deliagina & G illne ,
1999; Sel e s on, 2010).
1.4.1 The Coo dina i e Role o Inhibi ion
One o he aspec s o spinal eexes ha mos deeply imp essed She ing-
on was he gene al capaci y o eex ci cui s o ini ia e and swi ch be ween
concu en esponses despi e he exis ence o a
nal common pa h
om
he ne ous sys em o muscles (She ing on, 1904). Mo o neu ons in he
spinal co d send hei axons h ough he en al oo s o spinal segmen s
o synapse di ec ly on muscle b es. F om his expe imen s, She ing on
showed ha i was common o nd mul iple mo o neu ons pa icipa ing
syne gis ically o an agonis ically in a single coo dina ed eex esponse.
Mo e impo an ly, he e ealed ha he same mo o neu ons we e ac ually
sha ed among mul iple, po en ially conic ing, eex a cs. She ing on
was ascina ed by he ac ha hese an agonis ic eexes, ini ia ed sim-
ul aneously om dis inc senso y ecep o s, we e s ill ound o be able
o coo dina e hei inuence despi e sha ing his nal common pa h o
muscles. Tha such coo dina ion exis ed was made clea by s imula ion
expe imen s whe e wo o mo e eexes we e elici ed a he same ime,
gene a ing muscle esponses o he combined s imula ion ha we e no
a simple summa ion o linea combina ion o he esponses ob ained by
s imula ion deli e ed in isola ion. She ing on desc ibes he concep ion
clea ly:
Take he p ima y e inal eex, which mo es he eye so as
o b ing he o ea o he si ua ion o he s imula ing image.
F om all he ecep o s in each la e al e inal hal ise eex
22
a cs wi h a nal common pa h in he ne e o he opposi e
ec us la e alis
. Suppose simul aneous s imula ion o wo o
hese e inal poin s, one nea e o, one a he om, he o ea.
I he a cs o bo h poin s pou hei impulses in o he nal
common pa h oge he , he eec mus be a esul an o he
wo discha ges. I hese sum, he sho ening o he muscle
will be oo g ea and he o ea swing oo a o ei he poin .
I he esul an be a comp omise be ween he wo indi idual
poin s, he o ea will come o lie be ween he wo poin s o
s imula ion. In bo h cases he esul ob ained would be useless
o he pu poses o ei he ....
When wo s imuli a e applied simul aneously which would
e oke eex ac ions ha employ he same nal common pa h
in die en ways, in my expe ience one eex appea s wi hou
he o he . The esul is his eex o ha eex, bu no he
wo oge he . (She ing on, 1904, p.461)
In She ing on's ime he exis ence o such common pa hs was a p ob-
lem o he classic iew o eex con ol, whe e he unc ion o he ne ous
sys em was concei ed in e ms o ne e conduc ion o exci a o y impulses.
The exis ence o he nal common pa h media ing mul iple eexes made
i necessa y o speak openly o he p oblem o how o coo dina e die en
ci cui elemen s and o desc ibe mechanisms ha would allow he same
neu ons o ake on con ex -dependen oles in gene a ing mo o esponses.
I was du ing he hun o such a mechanism o eex a c coo dina ion
ha She ing on hi upon he undamen al ole o inhibi ion in he o gan-
iza ion o neu al unc ion. Inhibi ion had always been a complica ed opic
o physiologis s, bu ollowing he demons a ion o ca diac muscle inhibi-
ion by he agus ne e (Webe , 1846), and Secheno 's g and p oposal o a
cen al o igin o eex inhibi ion (Secheno , 1863), She ing on was able o
23
Kably & La oie, 1996). These obse a ions sugges ha mo o co ex
neu ons a e necessa y o p ecise s epping and adjus men o ongoing loco-
mo ion o changing condi ions. Howe e , long- e m eec s seem o equi e
comple e lesion o
bo h
he co icospinal and ub ospinal ac s (D ew e
al., 2002). E en in hese animals, he olun a y ac o s epping o e an
obs acle does no disappea en i ely, and mo eo e , hey can adap o
changes in he heigh o he obs acles (D ew e al., 2002). Specically,
e en hough hese animals ne e egain he abili y o g ace ully clea an
obs acle, when aced wi h a highe obs acle, hey a e able o adjus hei
s epping heigh in such a way ha would ha e allowed hem o com o -
ably clea he lowe obs acle (D ew e al., 2002). Fu he mo e, deci s
caused by lesions es ic ed o he py amidal ac seem o disappea o e
ime (Liddell & Phillips, 1944), and a e mos clea ly isible only he  s
ime an animal encoun e s a new obs acle (Liddell & Phillips, 1944).
The iew ha mo o co ex in non-p ima e mammals is p incipally
esponsible o adjus ing ongoing mo emen pa e ns gene a ed by lowe
b ain s uc u es is appealing. Wha is his modula ion good o ? Wha
does i allow an animal o achie e? How can we assay i s necessi y?
1.5 A S a egy o P obing Co ical Con ol
I should now be clea ha he in ol emen o mo o co ex in he di ec
con ol o all  olun a y mo emen  is human-specic. The e is a ole o
mo o co ex ac oss mammals in he con ol o p ecise mo emen s o he
ex emi ies, especially hose equi ing indi idual mo emen s o he nge s,
bu hese eec s a e sub le in non-p ima e mammals. Fu he mo e, wha
would be a de as a ing impai men o humans may no be so se e e o
mammals ha do no depend on p ecision nge mo emen s o su i al.
The e o e, gene alizing his specic ole o mo o co ex om humans o all
o he mammals would be misleading. We could be missing ano he , mo e
30

p imo dial ole o his s uc u e ha p edomina es in o he mammals,
and by doing so, we may also be missing an impo an ole in humans.
The p oposal ha mo o co ex induces modica ions o ongoing mo e-
men syne gies, p omp ed by he elec ophysiological s udies o ca loco-
mo ion, deni ely poin s o a ole consis en wi h he esul s o a ious
lesion s udies. Howe e , in assays used, he abili y o modi y ongoing
mo emen gene ally eco e s a e a mo o co ical lesion. Wha a e he
en i onmen al si ua ions in which mo o co ical modula ion is mos use-
ul?
Co ex has long been p oposed o be he s uc u e esponsible o in-
eg a ing a ep esen a ion o he wo ld and imp o ing he p edic i e powe
o his ep esen a ion wi h expe ience (Ba low, 1985; Doya, 1999). I mo-
o co ex is he means by which hese ep esen a ions can gain inuence
o e he body, howe e sub le and modula o y, can we nd si ua ions
(i.e. asks) in which his co ical con ol is equi ed?
The necessi y o co ex o a ious beha iou al asks has been ac i ely
in es iga ed in expe imen al psychology o o e a cen u y, including he
ounda ional wo k o Ka l Lashley and his s uden s (Lashley, 1921, 1950).
In he a , la ge co ical lesions we e ound o p oduce li le o no impai -
men in mo emen con ol, and e en deci s in lea ning and decision mak-
ing abili ies we e dicul o demons a e consis en ly o e epea ed i-
als. Howe e , Lashley did no ice some e idence ha co ical con ol may
be in ol ed in pos u al adap a ions o unexpec ed pe u ba ions (Lashley,
1921). These s udies once again seem o ecapi ula e he wo mos consis -
en obse a ions ound ac oss he en i e mo o co ical lesion li e a u e in
non-p ima e mammals since Hi zig (F i sch & Hi zig, 1870), Gol z (Gol z,
1888), She ing on (She ing on, 1885) and o he s (Oakley, 1979; Te y e
al., 1989). One, di ec olun a y con ol o e mo emen is mos deni ely
no abolished h ough lesion; and wo, ce ain aspec s o some mo emen s
a e deni ely impai ed, bu only unde ce ain challenging si ua ions. The
31
la e a e o en epo ed only anecdo ally. I was his collec ion o in-
iguing obse a ions in animals wi h mo o co ical lesions ha p omp ed
us o expand he scope o s anda d labo a o y asks o include a b oade
ange o mo o con ol challenges ha b ains encoun e in hei na u al
en i onmen s.
1.5.1 Ou line o he Thesis
In his wo k, an a emp o delinea e a new ole o mo o co ex is epo -
ed. As many p e ious eo s, i s a s wi h beha iou , and he ealiza ion
ha con olled exposu e o animals o a wide ange o en i onmen s is o
absolu e necessi y o gain insigh in o he eleology o he sys em. To
his end, we ha e de eloped new ools o make i easie o su ey a la ge
ange o en i onmen s while eco ding as many ne scale measu es o be-
ha iou and physiology as possible. These echnical and me hodological
de elopmen s a e desc ibed in Chap e 2.
In Chap e 3, a se o beha iou and lesion s udies is epo ed in he
a . These s udies had he goal o p obing he limi s o eco e y ollow-
ing ex ensi e co ical lesions by exposing animals o mo e challenging and
dynamic en i onmen s. De ailed analysis o he momen by momen be-
ha iou o lesioned animals e ealed a numbe o in iguing obse a ions,
he implica ions o which we discuss in Chap e 4.
32
Re e ences
Aalo, T. N. & G aziano, M. S. A. (2006). Possible o igins o he com-
plex opog aphic o ganiza ion o mo o co ex: educ ion o a mul-
idimensional space on o a wo-dimensional a ay.
The Jou nal o
neu oscience : he ocial jou nal o he Socie y o Neu oscience
,
26
(23), 628897. doi:10.1523/JNEUROSCI.0768-06.2006
Ala e dash ili, M. & Whishaw, I. Q. (2008). Mo o co ex s oke im-
pai s indi idual digi mo emen in skilled eaching by he a .
The
Eu opean jou nal o neu oscience
,
28
(2), 31122. doi:10.1111/j.1460-
9568.2008.06315.x
A ms ong, D. M. & D ew, T. (1984). Discha ges o py amidal ac and
o he mo o co ical neu ones du ing locomo ion in he ca .
The
Jou nal o Physiology
,
346
, 471495.
Ashby, W. R. (1960).
Design o a b ain: The o igin o adap i e beha io
.
New Yo k: Wiley.
Bake , M. R., Ja id, M. & Edgley, S. A. (2001). Ac i a ion o ce ebella
climbing b es o a ce ebella pos e io lobe om mo o co ical
ou pu pa hways.
Jou nal o Physiology
,
536
(3), 825839. doi:10.
1111/j.1469-7793.2001.00825.x
Ba low, H. (1985). Ce eb al co ex as model builde .
Models o he isual
co ex
, (1985), 3746.
Be nha d, C. G. & Bohm, E. (1954). Co ical ep esen a ion and unc ional
signicance o he co icomo oneu onal sys em.
A chi es o Neu o-
logy And Psychia y
,
72
(4), 473502. doi:10.1001/a chneu psyc.1954.
02330040075006
Be ns ein, J. (1868). Uebe den zei lichen Ve lau de nega i en Schwankung
des Ne ens oms.
Püge s A chi
,
1
, 173207. doi:10 . 1007 /
BF01640316
33
Be z, W. (1874). Ana omische Nachweis zweie Gehi ncen a.
Cen albla
ü die medizinischen Wissenscha en
,
12
, 57880, 59599.
Bju s en, L. M., No sell, K. & No sell, U. (1976). Beha iou al epe o y
o ca s wi hou ce eb al co ex om in ancy.
Expe imen al b ain e-
sea ch
,
130
(2857), 115130.
B odmann, K. (1909).
Localisa ion in he Ce eb al Co ex
. Sp inge . doi:10.
1097/00005053-191012000-00013
B ösamle, C. & Schwab, M. E. (2000). Ipsila e al, en al co icospinal
ac o he adul a : Ul as uc u e, myelina ion and synap ic con-
nec ions.
Jou nal o Neu ocy ology
,
29
, 499507. doi:10.1023/A:
1007297712821
Campbell, A. W. (1905).
His ological s udies on he localisa ion o ce eb al
unc ion
.
Chu chland, M. M. & Shenoy, K. V. (2007). Tempo al complexi y and
he e ogenei y o single-neu on ac i i y in p emo o and mo o co ex.
J Neu ophysiol
,
97
(6), 42354257. doi:10.1152/jn.00095.2007
Chu chland, M. M., Cunningham, J. P., Kau man, M. T., Fos e , J. D.,
Nuyujukian, P., Ryu, S. I. & Shenoy, K. V. (2012). Neu al popula ion
dynamics du ing eaching.
Na u e
,
487
(7405), 516. doi:10.1038/
na u e11129
Cla k, S. L. & Wa d, J. W. (1937). Elec ical s imula ion o he co ex
ce eb i o ca s.
A ch Neu Psych
,
38
(5), 927943.
Cla ke, E. & O'Malley, C. (1996).
The Human B ain and Spinal Co d: A
His o ical S udy Illus a ed by W i ings om An iqui y o he Twen-
ie h Cen u y
. No man Publishing.
Cole, J. (1952). Th ee es s o he s udy o mo o and senso y abili ies
in monkeys.
Jou nal o Compa a i e and Physiological Psychology
,
45
(3), 226230.
Da ling, W. G., Pizzimen i, M. A. & Mo ec a , R. J. (2011). Func ional e-
co e y ollowing mo o co ex lesions in non-human p ima es: expe i-
34
men al implica ions o human s oke pa ien s.
Jou nal o In eg a i e
Neu oscience
,
10
(3), 353384. doi:10.1142/S0219635211002737
de Ba enne, J. G. D. (1933). "Co icaliza ion" o unc ion and unc-
ional localiza ion in he ce eb al co ex.
A chi es o Neu ology
And Psychia y
,
30
(4), 884901. doi:10.1001/a chneu psyc.1933.
02240160196012
Donoghue, J. & Wise, S. P. (1982). The mo o co ex o he a : cy oa chi-
ec u e and mic os imula ion mapping.
The Jou nal o compa a i e
neu ology
,
212
(1), 7688. doi:10.1002/cne.902120106
Doya, K. (1999). Wha a e he compu a ions o he ce ebellum, he basal
ganglia and he ce eb al co ex?
Neu al ne wo ks : he ocial jou nal
o he In e na ional Neu al Ne wo k Socie y
,
12
(7-8), 961974.
D ew, T., Jiang, W., Kably, B. & La oie, S. (1996). Role o he mo o co -
ex in he con ol o isually igge ed gai modica ions.
Canadian
jou nal o physiology and pha macology
,
74
(4), 42642.
D ew, T., Jiang, W. & Widajewicz, W. (2002). Con ibu ions o he mo o
co ex o he con ol o he hindlimbs du ing locomo ion in he ca .
B ain esea ch. B ain esea ch e iews
,
40
(1-3), 17891.
du Bois-Reymond, E. (1843). Vo läuge Ab iss eine Un e suchung übe
den sogenann en F oschs om und übe die elek omo o ischen
Fische.
Annalen de Physik und Chemie
,
58
, 130.
E lich, J. C., Bialek, M. & B ody, C. D. (2011). A co ical subs a e o
memo y-guided o ien ing in he a .
Neu on
,
72
(2), 33043. doi:10.
1016/j.neu on.2011.07.010
E a s, E. V. (1968). Rela ion o py amidal ac ac i i y o o ce exe ed
du ing olun a y mo emen .
Jou nal o neu ophysiology
,
31
(1), 14
27.
Fe ie , D. (1873). Expe imen al Resea ches in Ce eb al Physiology and
Pa hology.
Jou nal o ana omy and physiology
,
8
(P 1), 1525.
doi:10.1136/bmj.1.643.457
35

Fe ie , D. & Yeo, G. (1884). A Reco d o Expe imen s on he Eec s
o Lesion o Die en Regions o he Ce eb al Hemisphe es.
Philo-
sophical T ansac ions o he Royal Socie y o London (1776-1886)
,
175
(Janua y), 479564.
Fo ssbe g, H., G illne , S. & Rossignol, S. (1975). Phase dependen eex
e e sal du ing walking in ch onic spinal ca s.
B ain Resea ch
,
85
(1),
103107. doi:10.1016/0006-8993(75)91013-6
Fos e , M. & She ing on, C. S. (1897).
A ex book o physiology
(7 h).
London: Macmillan.
F i sch, G. & Hi zig, E. (1870). Übe die elek ische e egba kei des g oss-
hi ns.
A chi u Ana omie, Physiologie und Wissenscha liche
,
37
,
300332.
Gal ani, L. (1791). De i ibus elec ici a is in mo u muscula is com-
men a ius.
De Bononiensi Scien ia um e A ium Ins i u o a que
Academia Commen a ii
,
7
, 363418. doi:10 . 1097 / 00000441 -
195402000-00023
Geo gopoulos, A., Schwa z, A. & Ke ne , R. (1986). Neu onal popula ion
coding o mo emen di ec ion.
Science
,
233
(4771), 14161419. doi:10.
1126/science.3749885
Glees, P. & Cole, J. (1950). Reco e y o skilled mo o unc ions a e small
epea ed lesions in mo o co ex in macaque.
J Neu ophysiol. 13
(2),
137148.
Gol z, F. (1888). Übe die e ich ungen des g osshi ns.
Püge s A chi
gesam e Physiologie des Menschen und de Tie e
,
42
, 419467.
Gol z, F. & Ewald, J. R. (1896). De Hund mi e kü z em Rückenma k.
Püge 's A chi ü die gesam e Physiologie des Menschen und de
Tie e
,
63
(7), 362400.
G aham B own, T. & She ing on, C. S. (1912). On he Ins abili y o a
Co ical Poin .
P oceedings o he Royal Socie y B: Biological Sci-
ences
,
85
(July), 250277. doi:10.1098/ spb.1912.0050
36
G aham B own, T. (1911). The In insic Fac o s in he Ac o P og es-
sion in he Mammal.
P oceedings o he Royal Socie y B: Biological
Sciences
,
84
(572), 308319. doi:10.1098/ spb.1911.0077. a Xi :
NIHMS150003
G aham B own, T. & She ing on, C. S. (1913). No e on he unc ions o
he co ex ce eb i.
Jou nal o Physiology
,
46
(suppl), xxii.
G aziano, M. S. A., Taylo , C. S. R. & Moo e, T. (2002). Complex Mo e-
men s E oked by Mic os imula ion o P ecen al Co ex.
Neu on
,
34
,
841851.
G illne , S. & Shik, M. L. (1973). On he descending con ol o he lum-
bosac al spinal co d om he "mesencephalic locomo o egion".
Ac a Physiol Scand. 87
(0001-6772), 320333.
G illne , S. & Zangge , P. (1975). How de ailed is he cen al pa e n gen-
e a ion o locomo ion?
B ain Resea ch
,
88
(2), 367371. doi:10.1016/
0006-8993(75)90401-1
G illne , S. (1981). Con ol o locomo ion in bipeds, e apods, and sh.
Handbook o Physiology, The Ne ous Sys em II
, 11791236. doi:10.
1002/cphy.cp010226
G oss, C. G. (2007). The disco e y o mo o co ex and i s backg ound.
Jou nal o he his o y o he neu osciences
,
16
(3), 32031. doi:10.
1080/09647040600630160
G ünbaum, A. S. F. & She ing on, C. S. (1903). Obse a ions on he
Physiology o he Ce eb al Co ex o he An h opoid Apes.
P oceed-
ings o he Royal Socie y o London
,
72
(477-486), 152155. doi:10.
1098/ spl.1903.0033
Hebb, D. O. (1949).
The O ganiza ion o Beha io : A Neu opsychological
Theo y
. New Yo k: Wiley.
Hene , R. S. & Mas e on, R. B. (1983). The ole o he co icospinal
ac in he e olu ion o human digi al dex e i y.
B ain, beha io
and e olu ion
,
23
, 165183.
37
Hill, D. N., Cu is, J. C., Moo e, J. D. & Klein eld, D. (2011). P ima y
mo o co ex epo s ee en con ol o ib issa mo ion on mul iple
imescales.
Neu on
,
72
(2), 344356. doi:10.1016/j.neu on.2011.09.
020. a Xi : NIHMS150003
Hyland, B. (1998). Neu al ac i i y ela ed o eaching and g asping in
os al and caudal egions o a mo o co ex.
Beha iou al B ain
Resea ch
,
94
(2), 255269. doi:10.1016/S0166-4328(97)00157-5
Jackson, J. H. (1870). A s udy o con ulsions.
T ansac ions o he Sain
And ews Medical G adua es Associa ion
,
3
, 162204. doi:10.1001/
a chneu .1970.00480200090012
James, W. (1885). Localiza ion o unc ions in he b ain.
Science
,
6
(143),
379380. doi:10.1126/science.ns-6.143.379
Ja a , H. & Hyland, B. (1999). Neu onal ac i i y in a ed nucleus du ing
o elimb each- o-g asp mo emen s.
Neu oscience
,
88
(2), 629642.
doi:10.1016/S0306-4522(98)00227-9
Kawai, R., Ma kman, T., Podda , R., Ko, R., Fan ana, A. L., Dhawale,
A. K., Kamp, A. R. & Öl eczky, B. P. (2015). Mo o Co ex Is
Requi ed o Lea ning bu No o Execu ing a Mo o Skill.
Neu on
,
113. doi:10.1016/j.neu on.2015.03.024
Kuma , S. & Hedges, S. (1998). A molecula imescale o e eb a e e ol-
u ion.
Na u e
,
392
, 917920.
Kushchaye , S. V., Moskalenko, V. F., Wiene , P. C., Tsymbaliuk, V. I.,
Che kaso , V. G., Dzya ulska, I. V., Ko alchuk, O. I., Sonn ag,
V. K. H., Spe zle , R. F. & P eul, M. C. (2012). The disco e y o
he py amidal neu ons: Vladimi Be z and a new e a o neu oscience.
B ain
,
135
(1), 285300. doi:10.1093/b ain/aw 276
Kuype s, H. G. J. M. (1981). Ana omy o he descending pa hways.
Com-
p ehensi e Physiology
, 597666.
Kwakkel, G., Kollen, B. J., an de G ond, J. & P e o, A. J. (2003). P ob-
abili y o Regaining Dex e i y in he Flaccid Uppe Limb: Impac o
38
Se e i y o Pa esis and Time Since Onse in Acu e S oke.
S oke
,
34
(9), 21812186. doi:10.1161/01.STR.0000087172.16305.CD
Lac oix, S., Ha on, L. a., McKay, H., Yang, H., B an , A., Robe s, J.
& Tuszynski, M. H. (2004). Bila e al co icospinal p ojec ions a ise
om each mo o co ex in he macaque monkey: a quan i a i e s udy.
The Jou nal o compa a i e neu ology
,
473
(2), 147161. doi:10.1002/
cne.20051
Langley, J. N. & She ing on, C. S. (1884). Seconda y Degene a ion o
Ne e T ac s ollowing emo al o he Co ex o he Ce eb um in he
Dog.
Jou nal o Physiology
,
5
(2), 4965. doi:10.1113/jphysiol.1884.
sp000151
Laplane, D., Talai ach, J., Meininge , V., Bancaud, J. & Boucha eine, A.
(1977). Mo o consequences o mo o a ea abla ions in man.
Jou nal
o he Neu ological Sciences
,
31
(1), 2949. doi:10 . 1016 / 0022 -
510X(77)90004-1
Lashley, K. S. (1921). S udies o ce eb al unc ion in lea ning. III. The
mo o a eas.
B ain
,
44
, 255285. doi:10.1037/h0070668
Lashley, K. S. (1924). S udies o ce eb al unc ion in lea ning: V. The
e en ion o mo o habi s a e des uc ion o he so-called mo o
a eas in p ima es.
A chi es o Neu ology and Psychia y
,
12
(3), 249
276. doi:10.1001/a chneu psyc.1924.02200030002001
Lashley, K. S. (1950). In sea ch o he eng am.
Symposia o he Socie y o
Expe imen al Biology
.
Law ence, D. G. & Kuype s, H. G. J. M. (1968a). The unc ional o ganiz-
a ion o he mo o sys em in he monkey. I. The eec s o bila e al
py amidal lesions.
B ain
,
91
(1), 114.
Law ence, D. G. & Kuype s, H. G. J. M. (1968b). The unc ional o gan-
iza ion o he mo o sys em in he monkey. II. The eec s o lesions
o he descending b ain-s em pa hways.
B ain
,
91
(1), 1536.
39

Chap e 2
Rapid P o o yping Tools o
he S udy o Beha iou
I is no ue ha  he labo a o y can ne e be like li e. The
labo a o y
mus
be like li e!
James J. Gibson
,
The Ecological App oach o Visual
Pe cep ion
(1979)
47
2.1 Chap e Summa y
The s udy o animal beha iou has p o ided emendous insigh in o he
unc ions o he b ain. Howe e , in he labo a o y, beha iou is o en s ud-
ied in impo e ished and epe i i e egimes in o de o con ol i s complex-
i y. In his chap e , we in oduce a se o ha dwa e and so wa e ools ha
make i easie o apidly su ey a la ge ange o en i onmen s wi hou
losing con ol o e obse able beha iou al da a.
In he  s pa o he chap e , we in oduce an a chi ec u e o a
mul i-pu pose modula beha iou box. This a chi ec u e makes i possible
o use simple ab ica ion and apid p o o yping ools o quickly econgu e
a physical en i onmen o die en assays equi ing complex combina ions
o senso s and ac ua o s.
In he second pa , we p esen Bonsai, a high-pe o mance isual p o-
g amming language o con olling and moni o ing eal- ime da a s eams
on a digi al compu e . We desc ibe Bonsai's co e p inciples and a chi ec-
u e and demons a e how i allows o he apid and exible p o o yping
o in eg a ed expe imen al designs in neu oscience. We specically high-
ligh some applica ions ha equi e he combina ion o many die en
ha dwa e and so wa e componen s, including ideo acking o beha io ,
elec ophysiology and closed-loop con ol o s imula ion.
All he esul s conce ning he Bonsai amewo k ha e been published
as: Lopes, G., Bonacchi, N., F azão, J., Ne o, J., A allah, B., Soa es,
S., Mo ei a, L., Ma ias, S., I sko , P., Co eia, P., Medina, R., Calca-
e a, L., D eos i, E., Pa on, J. & Kamp, A. (2015). Bonsai: An e en -
based amewo k o p ocessing and con olling da a s eams.
F on ie s in
Neu oin o ma ics
,
9
(7). doi:10.3389/ nin .2015.00007
48
2.2 In oduc ion
The o mal s udy o animal beha iou has a long his o y spanning hund eds
o yea s ac oss he elds o e hology, expe imen al psychology and neu -
oscience. While he e hologis s mainly endea ou ed o s udy beha iou
in i s na u al en i onmen , he psychologis s and neu ophysiologis s ha e
classically eso ed, o necessi y, o mo e con olled labo a o y se ings.
The eason is mainly one o complexi y. Beha iou is a highly mul i-
dimensional, mul i-scale phenomenon ha o en allows no clea sepa a ion
be ween ele an and i ele an a iables (Gomez-Ma in, Pa on, Kamp,
Cos a & Mainen, 2014). I is in gene al impossible o p edic wha an an-
imal is going o do simply because some o he c ucial in o ma ion is no
e en accessible o measu emen . In o de o mi iga e his p oblem, neu os-
cien is s eso o making impo e ished p epa a ions whe e he numbe o
a iables ha a e changing a any gi en momen is low and e y ca e ully
con olled. The hope is ha in his way he in e p e a ion o b ain signals
eco ded simul aneously wi h animal beha iou will be acili a ed.
Depending on he kind o ques ion a neu oscien is is a e , an ap-
p op ia e beha iou pa adigm is se up. Anaes he ized and head-xed
p epa a ions, as well as classical o ope an condi ioning boxes a e egu-
la ly employed o d i e he beha iou o he animal o oscilla e be ween a
se o epea edly ep oducible s a es mo e amenable o s a is ical analysis.
Building such beha iou assays o en equi es e y specialized enginee ing
skills and long de elopmen cycles o ial and e o in o de o ensu e all
he ele an a iables a e con olled acco dingly. Because o his, he end-
ency o he eld has been o concen a e on a small se o s anda dized
assays which ha e been shown o wo k o one a ea o esea ch o o he .
Small a ia ions o he s anda d asks a e g adually in oduced in o de
o p obe die en aspec s o he sys em. The complexi y o beha iou
49
s udies in neu oscience has hus adi ionally p og essed by a i ion and
pains aking accumula ion o small pe u ba ions o o e all design pa e ns.
In e es ingly, howe e , many o he mos signican concep ual ad-
ances in ou unde s anding o b ain unc ion ha e in ac de eloped
pa i
passu
wi h o ays in o en i ely new beha iou spaces. Mo ing om anaes-
he ized o awake physiology comple ely changed he way we unde s and
he neu al p ocessing o senso y s imuli (Selle s, Benne , Hu , Williams
& F öhlich, 2015). Simila ly, mo ing om head-xed o eely mo ing be-
ha iou led o he disco e y o place elds in hippocampus (O'Kee e &
Dos o sky, 1971). Single ial analysis o simul aneously eco ded e-
sponses ha e e ealed pa e ns o neu al ac i i y such as hippocampal
ipples ha a e simply impossible o eco e om s a is ical a e ages o
epe i i e beha iou episodes (Fos e & Wilson, 2006; Da idson, Kloos e -
man & Wilson, 2009). Each o hese de elopmen s has equi ed signican
ad ances in ools used o eco d and con ol beha iou al da a a a ne
scale. Un o una ely, he echnical cos and scien ic isk o ying some-
hing no el means ha such ad ances a e s ill much ewe and a be ween
han would be desi able.
F om he beginning o his wo k i was unde s ood ha e ealing he
eleology o co ical con ol o e beha iou would equi e jus his kind o
o ay in o di e se and po en ially unknown beha iou spaces. We ag eed
ha i migh be wo h o y and de elop a oolki o he beha iou al
neu oscien is ha would accele a e he explo a ion o his as space.
One o he  s ob ious a ge s o imp o emen was he beha iou box.
T adi ionally, when a gi en beha iou assay is ound o p oduce in e es ing
esul s, i s design is p og essi ely weaked so as o exace ba e he ea u es
o he o iginal eec . In his wo k, we s a ed by b eaking apa his
concep o he polished beha iou box, and wonde ed wha would happen
i ins ead o a s anda d box, we could ha e a box o s anda ds.
50
2.3 Me hods
2.3.1 The Modula Beha iou Box
A he ou se i was decided ha he scale o he modula a chi ec u e
would p obably ha e o ma ch a gi en animal model, gi en he as ly di -
e en size scales be ween oden s, ca s and p ima es. Ou animal model
o choice is he oden
a us no egicus
, and all o ou p oposed design
choices a ge i s size scale. Small adjus men s could, howe e , be eas-
onably made up o a poin o o he mammals o simila s a u e, such as
mice.
The main componen and in e ace o he modula box is he indi idual
1×1
module (Figu e 2.1A). This module denes a s anda dized oo p in
(
12 cm ×12 cm
), agains which all o he modules a e measu ed. E e y
newly ab ica ed module is buil o specica ion o ma ch a mul iple o
his s anda dized oo p in (e.g. i is possible o ha e
2×1
,
2×2
,
4×
1
o any o he mul iple combina ion o he s anda d size). Inside he
module oo p in he module designe places a single logical componen
o a beha iou box and ensu es ha i can ope a e in isola ion. Figu e
2.1 shows some examples o eusable modules de eloped h oughou he
p ojec .
One o he p incipal equi emen s o assembling a box is as ening all
i s componen s oge he . By ha ing a s anda d oo p in , i is possible
o design a se o egula ly spaced moun ing poin s ha allows he ex-
pe imen alis o quickly gene a e an en i ely new congu a ion by simply
swapping modula componen s inside he box (Figu e 2.2, 2.3). Fo his
wo k, we ook ad an age o an exis ing aluminium s uc u al aming sys-
em (Bosch Rex o h, DE) o build he common moun ing poin s (Figu e
2.1A). Modules a e as ened agains pos -inse ion nu s which a e able o
slide ac oss he whole leng h o he aluminium ail. Each o he modules
is as ened by ou sc ews, one in each co ne . In o de o ensu e modules
51

A B
C D
Figu e 2.1.
Some examples o s anda dized beha iou modules. (
A
) De-
ail o a
1×1
module moun ed in suppo ame. Fixa ion is achie ed
by d i ing a sc ew h ough pos -inse ion nu s placed in he s uc u al
aming (see ex ). (
B
) Example ewa d po module which can be oo -
o wall-moun ed. All ele an elec onics and wa e dis ibu ion ci cui s
a e assembled on he back o he module (no shown). (
C
) Wall-moun ed
econgu able obs acle cou se s eppe module. S eppe mo o s moun ed
on he back o he module allow o dynamic econgu a ion o he o ien -
a ion o each s ep. (
D
) Floo -moun ed obs acle cou se s ep pai . Mul iple
o hese modules can be iled oge he o assemble obs acle cou ses o
a bi a y leng h.
52
Figu e 2.2.
Example o a linea shu ling box assembled om a
1 m ×1 m
modula s uc u e using ewa d po and obs acle cou se s ep modules.
Figu e 2.3.
Side iew o he linea shu ling box.
53
AB
Figu e 2.4.
Example o e ical assembly. (
A
) De ail o a
1×1
wall-
moun ed pla o m module. (
B
) Example o a e ical maze congu a ion.
can be igh ly and secu ely xed one nex o he o he , we used a sys em
o egula ly spaced double ails (Figu e 2.2). This gi es he ame he
exibili y o easily eposi ion and ea ange indi idual modules iling he
en i e oo p in o any a bi a ily la ge box.
I he suppo ame is laid ou e ically, i is possible o c ea e mod-
ula walls o a bi a y dimensions. Some o he modules can be moun ed
equally well on a e ical o ho izon al congu a ion, such as ewa d po s
(Figu e 2.1B). The h ee-dimensionali y o he design has e en been ex-
ploi ed o c ea e e ical mazes (Figu e 2.4) o g ea success.
Th oughou he p ojec we made he base o e e y module om
5 mm
ac ylic pieces. While no an absolu e equi emen o he design, his
choice o plas ic ma e ial has he ad an age ha a lase cu e can be used
o e y quickly p oduce a la ge collec ion o cus om-buil modules. In
addi ion, pa e ns can be eng a ed o cu on he base o p o ide addi ional
moun ing poin s o ha dwa e embedded in he module. The use o such
apid p o o yping ab ica ion ools alongside wi h o he shel a ailable
elec onic senso s and ac ua o s mean we we e able o comple ely edesign
he en i e beha iou box, some imes in a ma e o days.
54
2.3.2 The Bonsai F amewo k
Mode n scien ic expe imen s c ucially depend on he con ol and mon-
i o ing o many pa allel s eams o da a. Mul iple measu emen de ices,
om ideo came as, mic ophones, and p essu e senso s o neu al elec-
odes, mus simul aneously send hei da a in eal- ime o a eco ding
sys em. Gene al pu pose digi al compu e s ha e g adually eplaced many
o he specialized analog and digi al echnologies used o his kind o da a
acquisi ion and expe imen con ol, la gely due o he exibili y o p o-
g amming and he exponen ial g ow h in compu ing powe . Howe e , he
se ial na u e o p og amming ins uc ions and sha ed memo y makes i a
challenge, e en o expe ienced p og amme s, o de elop so wa e ha can
elegan ly deal wi h he asynch onous, pa allel na u e o scien ic da a.
Ano he challenge a ises om he need o so wa e in eg a ion. Each
ha dwa e endo p o ides hei own se o d i e s and p og amming in e -
aces o congu ing and acqui ing da a om hei de ices. In addi ion,
he g ow h o he open-sou ce mo emen has g ea ly inc eased he num-
be o eely a ailable echnologies o die en da a p ocessing domains.
In eg a ion o hese di e se so wa e and ha dwa e componen s emains a
majo challenge o esea che s.
These dicul ies lead o inc eased de elopmen imes when se ing up
an expe imen . Mo eo e , i equi es expe imen e s o pu sue specialized
aining ou side hei domain o esea ch. This limi s he abili y o apidly
p o o ype and y ou new designs and can quickly become he ac o
limi ing he kinds o ques ions ha a e amenable o scien ic in es iga ion.
He e we desc ibe Bonsai, an open-sou ce isual p og amming ame-
wo k o p ocessing da a s eams. The main goal o Bonsai is o simpli y
and accele a e he de elopmen o so wa e o acqui ing and p ocessing
he many he e ogeneous da a sou ces commonly used in (neu o) scien ic
esea ch. We aim o acili a e he as implemen a ion o s a e-o - he-a
expe imen al designs and o encou age he explo a ion o new pa adigms.
55
show up as p ope ies o he g oup node i sel . This allows o he pa a-
me e iza ion o nes ed da aows and inc eases hei euse possibili ies. In
addi ion, encapsula ed da aows a e used o speci y mo e complica ed, ye
powe ul, ope a o s such as i e a ion cons uc s ha allow o he compac
desc ip ion o complex da a p ocessing scena ios ha can be cumbe some
o speci y in pu e da aow isual languages (Mosconi & Po a, 2000) (see
below).
Bonsai was designed o be a modula amewo k, which means i is pos-
sible o ex end i s unc ionali y by ins alling addi ional packages con aining
sou ces and combina o s de eloped o specic pu poses. New packages can
be w i en using C# o any o he.NET p og amming languages. Py hon
sc ip s [ ia I onPy hon (I onPy hon Communi y, 2014)] can be embedded
in he da aow as ans o ms and sinks, allowing o apid in eg a ion o
cus om code. All unc ionali y included in Bonsai was designed using hese
modula p inciples, and we hope o encou age o he esea che s o con ib-
u e hei own packages and he eby ex end he amewo k o o he applic-
a ion domains. A p esen , he a ailable packages include compu e ision
and signal p ocessing modules based on he OpenCV lib a y (I seez, 2014).
D i e s o se e al came as and in e aces o o he imaging and signal ac-
quisi ion ha dwa e we e in eg a ed as Bonsai sou ces and sinks, including
suppo o A duino mic ocon olle s (Banzi, Cua ielles, Igoe, Ma ino &
Mellis, 2014), se ial po de ices and basic ne wo king using he OSC p o-
ocol (W igh , F eed & Momeni, 2003). Gi en he specic applica ions in
he domain o neu oscience, we also in eg a ed a numbe o neu oscience
echnology packages. The Ephys package, o example, builds on he Open
Ephys ini ia i e o he sha ing o elec ophysiology acquisi ion ha dwa e
(Voig s, Siegle, Keme e, Moo e & Wilson, 2013) by p o iding suppo o
he Rhy hm open-sou ce USB/FPGA in e ace (In an Technologies, US).
The e o e, he nex gene a ion ools o elec ophysiology can al eady be
used inside Bonsai, he acqui ed physiology da a implici ly in eg a ed wi h
62

o he a ailable da a s eams and hus easily assembled in o a powe ul and
exible expe imen al neu oscience pla o m.
Ad anced Ope a o s
The mos common applica ion o Bonsai is he acquisi ion and p ocessing
o simple, independen da a s eams. Howe e , o many mode n expe -
imen s, basic acquisi ion and s o age o da a is o en no sucien . Fo
example, i can be con enien o only eco d he da a aligned on e en s o
in e es , such as he onse o specic s imuli. Fu he mo e, neu oscience
expe imen s o en p og ess h ough se e al s ages, especially o beha i-
o al assays, whe e con olled condi ions a y sys ema ically ac oss die en
sessions o ials. In o de o en o ce hese condi ions, expe imen s need
o keep ack o which s age is ac i e and use ha in o ma ion o upda e
he s a e o con ol a iables and senso y p ocessing. These equi emen s
o en canno be desc ibed by a simple linea pipeline o da a, and equi e
cus om code o handle he complica ed logic and bookkeeping o expe i-
men al s a es. Below we desc ibe a se o ad anced Bonsai ope a o s ha
can be used o exibly econgu e da a p ocessing logic o co e a la ge
numbe o scena ios. These ope a o s and hei applica ions a e all buil
on he single idea o slicing a da a s eam in o sub-sequences, called win-
dows, which a e hen p ocessed independen ly and, po en ially, in pa allel
(Figu e 2.7).
Bonsai p o ides die en combina o s ha allow he c ea ion o hese
sub-sequences om any obse able da a s eam, using elemen coun in-
o ma ion, iming, o ex e nal igge s (Figu es 2.7AC). The specic se
o ope a ions o apply on each window is desc ibed by encapsula ing a
da aow inside a Selec Many g oup, as de ailed in he signal p ocessing
example o Figu e 2.7D. The inpu sou ce in his g oup ep esen s each
o he window sub-sequences, i.e., i is as i each o he windows is a new
da a sou ce, con aining only he elemen s ha a e a pa o ha window.
63
Ready Go
Time
KeyDown
F.
A.
Time
Da a
Sou ce
S W
Window
(Elemen Coun )
Sou ce
Window
(Coun 2, Skip 2)
Window
(Coun 2, Skip 3)
Window
(Coun 2, Skip 1)
B. S W
Sou ce Window
(TimeSpan)
Sou ce
Window
(TimeSpan 1s)
D.
Raw Fil e ed
S W S
Sou ce Window Selec Many
S A W
Inpu A e age Ou pu
57 3 27 6
57 3
7 3 2
327
27 6
54 4 5
Sou ce
Window
(Coun 3, Skip 1)
Selec Many
(A e age)
C.
S W
T
Sou ce Window
(T igge ed)
T igge
T igge
Sou ce
Window
(T igge ed)
E.
VideoW i e
S V W
Inpu Ou pu
C W S
T
Came a
T igge
Selec ManyWindow
T igge
* * * * * *
File 1 File 2 File 3
Came a
Selec Many
(A e age)
Window
(T igge ed)
Ready
R G R
Go Repea
WK T
KeyDown Take (1) Ou pu
S S
Inpu S imulus
Onse
Figu e 2.7.
Using slicing and window p ocessing combina o s in Bonsai.
64
These elemen s will be p ocessed as soon as hey a e a ailable by he en-
capsula ed da aow. Windows can ha e o e lapping common elemen s,
in which case hei p ocessing will happen concu en ly. The p ocessing
ou pu s om each window a e me ged oge he o p oduce he nal esul .
In he case o Figu e 2.7D, pas and u u e samples a e g ouped in win-
dows o compu e a unning a e age o he signal h ough ime, necessa ily
ime-shi ed by he numbe o u u e samples ha a e conside ed in he
a e age.
The p ocessing o he elemen s o each window happens independen ly,
as i he e was a new isola ed da aow unning o each o he sequences.
We can exploi his independence in o de o dynamically u n da aows
on and o du ing an expe imen . In he ideo spli ing example o Figu e
2.7E, we use an ex e nal igge sou ce o chop a con inuous ideo s eam
in o many small ideo sequences, aligned when he igge  ed. We hen
nes a VideoW i e sink in o he Selec Many g oup. The VideoW i e sink
is used o encode ideo ames in o a con inuous mo ie le. I s a s by c e-
a ing he ideo le upon a i al o he  s ame, and hen encoding e e y
ame in he sequence as hey a i e. When he da a s eam is comple ed,
he le is closed. By nes ing he VideoW i e inside he Selec Many g oup,
wha we ha e eec i ely done is o c ea e a new ideo le o each o he
c ea ed windows. Whene e a new igge a i es, a new clip is c ea ed
and sa ing p oceeds, implici ly pa allelized, o ha ideo le.
Mo e gene ally, we can use his idea o implemen disc e e ansi ions
be ween die en p ocessing modes, and chain hese s a es oge he o
design complex con ol s uc u es such as ni e s a e machines (FSMs).
FSMs a e widely used o model en i onmen s and beha io al assays in
sys ems and cogni i e neu oscience. One example is illus a ed in Figu e
2.7F, whe e we depic he con ol scheme o a s imulus- esponse appa a us
o a simple eac ion ime ask. In his ask, he e a e only wo s a es:
Ready and Go. In he Ready s a e, no s imulus is p esen ed and a ime
65
is a med. Whene e he ime  es, he ask ansi ions in o he Go s a e,
and a s imulus is p esen ed. The subjec is ins uc ed o p ess a key as
as as possible upon p esen a ion o he s imulus. As soon as he key is
p essed, he sys em goes back o he Ready s a e o s a ano he ial. In a
FSM, nodes ep esen s a es, e.g., s imulus a ailabili y o ewa d deli e y,
and edges ep esen ansi ions be ween s a es ha a e caused by e en s
in he assay, e.g., a key p ess. In each s a e, a numbe o ou pu a iables
and con ol pa ame e s a e se (e.g., u ning on a ligh ) which ep esen
he beha iou o he machine in ha s a e.
In he Bonsai da aow model, da aows encapsula ed in a Selec Many
g oup can be used o ep esen s a es in a FSM (Figu e 2.7F, bo om).
Specically, a s a e is ac i a ed whene e i ecei es an inpu e en , i.e.,
he da aow nes ed inside he s a e will be u ned on. The dynamics o
he nes ed da aow de e mine he dynamics o he s a e. In he Go s a e
p esen ed in Figu e 2.7F, he ac i a ion e en is used o igge s imulus
onse . In pa allel, we s a lis ening o he key p ess which will e mina e
he s a e. Con e sely, o he Ready s a e we would igge s imulus ose
and a m he ime o p esen ing he nex s imulus. An impo an di -
e ence be ween Bonsai da aows and pu e s a e machine models is ha
a da aow is specied as a di ec ed acyclic g aph, i.e., he da a s eam
canno loop back on i sel . Howe e , by aking ad an age o he Repea
combina o , we can es a a da aow once i is comple ed, allowing us o
ese he s a e machine o he nex ial.
Many o he con ol asks in expe imen s ha e his sequen ial ial-
based s uc u e, which has allowed us o apidly p o o ype complex beha-
iou assays, such as closed-loop oden decision making asks, simply by
le e aging he exibili y o he da a s eam slicing ope a o s.
66
Al e na i es o Bonsai
Al hough g aphical use in e aces ha e played a c ucial ole in he wide-
sp ead p oli e a ion o compu ing echnology h oughou a ious scien ic
elds, he majo i y o hese in e aces end o be applied o ela i ely
na ow domains, such as he ope a ion o a specic ins umen . Thei
goal is o en o p o ide access o all he a ious congu a ion pa ame -
e s o he ha dwa e and o p o ide basic da a acquisi ion unc ionali y.
The e is o en no oppo uni y o pa ame e ize o condi ion he beha iou
o he ins umen beyond he possibili ies p esen ed by he in e ace, and
in e connec ions wi h o he de ices a e o en limi ed o simple ha dwa e
igge s. The al e na i e, when a ailable, is o access low-le el applica ion
p og amming in e aces (APIs), and p og am he desi ed beha iou om
sc a ch.
In he mo e exible domains o da a analysis, beha iou con ol and
so wa e simula ions, he use o mo e e sa ile g aphical in e aces has be-
come inc easingly p e alen . In hese scena ios, i is no uncommon o
encoun e he de elopmen o domain-specic languages (DSLs), whe e
g aphical building blocks ela ed o he domain o applica ion can be com-
bined oge he by he use o gene a e new beha io s, such as he sequence
o s eps in a psychophysics expe imen o a s a e-machine diag am used
o con ol s imuli and ewa ds in ope an condi ioning. While p o iding
mo e exibili y o he end use , such DSLs a e usually no concei ed, a
hei co e, o be applied o wildly die en domains (e.g., an ope an con-
di ioning s a e machine is no expec ed o be able o l e con inuous
elec ophysiology signals). In ac , mos DSLs will no e en allow he use
o ex end he se o buil -in ope a ions. In hose ha do, he de elope
may nd a cus omiza ion pi (Cook, Jones, Ken & Wills, 2007), whe e
concep s and ope a ions ha a e wi hin he ange o wha he DSL can
exp ess a e easy o de elop, whe eas asks ha a e a li le bi ou side o he
67

bounda ies o he language quickly become impossible o oo cumbe some
o implemen .
As he le el o exibili y o a g aphical use in e ace inc eases, we
s a o app oach he space occupied by gene al pu pose isual p og am-
ming languages (GPVPL). These a e languages ha a e designed om he
ou se o be capable o sol ing p oblems ac oss a wide a ie y o domains
using a gene al se o ope a ions. Ideally, he co e building blocks o he
language will hemsel es be domain-independen , so ha he use can eas-
ily apply he same se o ope a ions o he wides possible class o inpu s.
In o de o be e illus a e he eel and exp essi e powe o GPVPLs, and
o cla i y whe e Bonsai i sel is posi ioned, we will gi e wo examples o
popula languages ha ha e succeeded in his niche: LabVIEW (Na ional
Ins umen s, 2014) and Simulink (Ma hWo ks, 2014).
LabVIEW is one o he bes examples o a GPVPL applied o he design
and con ol o expe imen s (Ellio , Vijayakuma , Zink & Hansen, 2007).
In LabVIEW, use s c ea e i ual ins umen s (VIs) which a e composed o
a g aphical on -panel con aining an asso men o bu ons, dials, cha s
and o he objec s; as well as a back-panel whe e a owcha -like block dia-
g am can be used o speci y he beha iou o he VI. In his back-panel,
nodes and e minal elemen s can ep esen ha dwa e componen s, nume -
ical ope a ions o on -panel objec s, which a e connec ed oge he using
i ual wi es ha speci y he ow o da a be ween hem. The popula i y o
LabVIEW g ew ini ially om i s suppo o s a e-o - he-a da a acquisi-
ion ca ds and ha dwa e as well as i s da a isualiza ion capabili ies. The
modula i y o i s a chi ec u e also allowed use s o quickly de elop and
implemen new nodes wi hin he language i sel by using VIs hemsel es
as nodes.
Al hough he LabVIEW back-panel is a da aow isual p og amming
language, i s execu ion model ends o ollow a polling, a he han e en -
d i en, s a egy o dealing wi h mul iple da a s eams. In o de o p op-
68
e ly scale his model o he inc easing numbe o a ailable p ocesso co es,
LabVIEW has implemen ed sophis ica ed code analysis ools ha a emp
o iden i y pa allelizable po ions o block diag ams au oma ically (Ellio
e al., 2007). Once hese sec ions a e iden ied, LabVIEW will au oma ic-
ally gene a e pa allel p ocesses depending on he numbe o a ailable co es
and will manage he bo lenecks in he code acco dingly. Al hough his
mi iga es he limi a ions o he sequen ial polling p og amming model, i
is impo an o ealize ha he goal o such au oma ic pa alleliza ion is
s ill o p o ide he use wi h a logically synch onized p og amming model.
Simulink is a popula da aow isual p og amming language o mod-
eling, simula ing and analyzing mul i-domain dynamic sys ems. I has
become ex emely popula o modeling esponse cha ac e is ics o con ol
sys ems, allowing no only o he apid p o o yping o algo i hms, bu also
he au oma ic gene a ion o mic ocon olle code o embedded sys ems.
Again, he success o he language s emmed p ima ily om he exibili y
and ease o use o he block diag ams, as well as he numbe o p ebuil
ope a ions and da a isualiza ion ools which quickly ook ca e o many
c ucial bu edious aspec s o con ol sys ems modeling.
Like LabVIEW, he execu ion model o Simulink gene a ed code is
s ill based on polling s a egies, whe e eady o execu e da aow nodes
a e upda ed in u n as inpu s become a ailable. Again, s a egies o scale
he ou pu o Simulink o mul iple co es ha e been p oposed based on
analyzing and segmen ing he model in o pa allelizable sec ions which can
be con e ed in o equi alen pa allel execu ion code o mic ocon olle s
(Kumu a, Nakamu a, Ishiu a, Takeuchi & Imai, 2012).
Simila o LabVIEW and Simulink, Bonsai was designed as a gene al
pu pose modula language. The co e a chi ec u e o Bonsai is domain-
independen and p o ides a gene al amewo k o compose asynch onous
da a s eams. A gene al se o composi ion ope a o s, o combina o s,
p o ides suppo o i e a ion, segmen a ion and me ging o pa allel da a
69
s eams, as well as o he common manipula ions on obse able sequences.
Bo h he sou ces o da a and a ailable p ocessing ope a ions can be ex en-
ded wi hin he language i sel using nes ing o da aows. Da a isualize s
and a g owing lib a y o da a s eam acquisi ion, p ocessing and logging
modules a e p o ided o allow apid p o o yping o a la ge numbe o di -
e en applica ions.
Howe e , in con as o LabVIEW o Simulink, Bonsai adop s a e y
die en s a egy o implemen da aow execu ion. Ra he han ying
o de i e a global sequen ial execu ion o de o da aow nodes based on
he numbe o ac i e inpu s, Bonsai nodes simply eac o incoming in-
pu s immedia ely, wi hou he need o wai o all o hem o be ac i e.
When mul iple obse able sequences a e p esen , his allows o a choice
o die en concu ency composi ion s a egies. Ne e heless, as he es-
ul o he composi ion is an obse able sequence i sel , such concu ency
managemen can emain unc ionally isola ed om he combina o ha is
handling he composi ion. F om he poin o iew o downs eam ope a -
o s, hey a e simply ecei ing an obse able sequence. The e is a adeo,
o cou se, ha mo e esponsibili y o managing he ow o da a is passed
o he end use , bu i also allows o a ne g ained con ol o concu ency
ha is c i ical o he specica ion o pa allel applica ions.
One impo an ca ea o de eloping asynch onous sys ems is ha de-
bugging can be mo e dicul in si ua ions whe e he p ecise iming and
o de ing o e en s is equi ed o ep oduce an oending beha iou . In syn-
ch onized and sequen ial execu ion en i onmen s, one can easily go s ep
by s ep h ough he p ecise cascade o ans o ma ions ha esul ed in a
p oblem. In con as , when mul iple p ocesses a e execu ing concu en ly,
i can be ha de o analyze he p og am ow in a simila ly ep oducible,
de e minis ic manne . Howe e , i should be no ed ha his issue is no
unique o eac i e en i onmen s wi h eal asynch onous de ices. A sequen-
70
ial polling s a egy will be equally decien in ep oducing a pa icula
execu ion sequence when da a om pa allel inpu de ices is being accessed.
Ano he impo an ca ea is ha Bonsai cu en ly uns exclusi ely
in Windows ope a ing sys ems. Howe e , Mic oso has ecen ly open-
sou ced he execu ion engine o he.NET amewo k and will pu sue imple-
men a ions o all he majo ope a ing sys ems (Linux/Mac). This aises
he in e es ing possibili y o e en ually ex ending he Bonsai use base in o
hese impo an pla o ms.
2.4 Resul s
2.4.1 Ge ing S a ed wi h Bonsai
Communi y
The Bonsai amewo k can be downloaded a h ps://bi bucke .o g/
ho izongi /bonsai and ins alled on Windows ope a ing sys ems s a ing
wi h Windows 7 and abo e. The websi e is o ganized in o die en
sec ions: Downloads (whe e he la es ins alle is loca ed), Wiki (wi h a
Ge ing S a ed guide, u o ials and (FAQ) equen ly asked ques ions),
and Issues (whe e bugs can be epo ed). We ha e also c ea ed a use
o um (add ess is lis ed in he FAQ sec ion) whe e he communi y o
Bonsai use s ha e been sha ing hei eedback, ques ions and expe iences.
A ideo u o ial in oduc ion o Bonsai is included wi h his publica ion
(Video 2.1).
Ex ending Bonsai
Bonsai was designed om he ou se o suppo many die en laye s o
ex ensibili y:
(a) Da aows: The  s laye is h ough he c ea ion o Bonsai da aow
les hemsel es. Exis ing da aows can be di ec ly eused inside
71
elec ophysiology o o he digi al acquisi ion sys ems whe e mul iple da a
samples, om one o mo e channels, a e synch onously acqui ed, bue ed
and s eamed o he compu e . These bue s a e o en ep esen ed as da a
ma ices, whe e ows a e channels and columns ep esen indi idual da a
samples h ough ime, o ice- e sa. Suppo o simple band-pass l e s,
h esholding and igge ing allowed us o build exible spike de ec ion and
wa e o m ex ac ion sys ems (Figu e 2.8F). Using In an's Rhy hm API,
we in eg a ed in o Bonsai suppo o a a ie y o nex -gene a ion elec o-
physiology de ices using In an's digi al amplie echnology, such as he
Open Ephys acquisi ion sys em (Voig s e al., 2013) o In an's e alua ion
boa d (RHD2000, In an Technologies, US). This sys em was success ully
used o acqui e and isualize simul aneous eco dings om dense silicon
p obes whe e spikes om a loose-pa ch jux acellula pipe e we e used as
igge s o align and ex ac wa e o m da a appea ing on he mul i-channel
ex acellula p obe. Responses om e e y silicon p obe si e could hen be
supe imposed on an accu a e endi ion o he p obe geome y, in eal- ime.
The abili y o apidly in eg a e new modules allowed us o suppo he
de elopmen and c oss- alida ion o new ools o beha io al neu oscience.
A pa adigma ic example was he yPAD, a new me hod o quan i ying
eeding beha iou in D osophila melanogas e by measu ing changes in
elec ode capaci ance induced by he p oboscis ex ension o a y (I sko
e al., 2014). The in eg a ion o he yPAD in Bonsai allowed esea ch-
e s o quickly ge s a ed using his app oach o design new expe imen s.
Fu he mo e, i also allowed he alida ion o he ool by enabling simul-
aneous acquisi ion o high-speed ideo eco dings o y beha iou which
we e la e used o anno a ion and classica ion o he senso eeding aces
(Figu e 2.8G).
In a die en se o expe imen s, Bonsai was used o implemen a a i-
a ion on a popula wo-al e na i e o ced choice (2AFC) decision-making
ask o oden s (Figu e 2.8H). In his ype o ask, animals a e placed in
78

an en i onmen wi h h ee po s. They a e p esen ed wi h a s imulus in
he cen e po and a e wa ds epo hei pe cep ion o he s imulus by
going ei he o he le o igh choice po s. In he a ia ion we p esen in
his wo k, he wo choice po s we e eplaced by egions o in e es whe e
he ac i i y o he animal is analyzed using compu e ision. This example
oe ed unique challenges as i combined sophis ica ed sequen ial con ol o
a ask en i onmen wi h con inuous da a s eam p ocessing o ideo and
senso da a.
The in eg a ion o all hese di e se componen s o da a acquisi ion
and expe imen con ol does no only allow o he apid deploymen o
es ablished p o ocols. In ac , he modula na u e o hei in eg a ion (i.e.,
how hey can be combined oge he ) opens up new a enues o esea ch, by
allowing a ich, apid explo a ion o no el me hodologies. To demons a e
his, we c ea ed a dynamic i ual en i onmen o eely mo ing oden s
whe e he isual p esen a ion o a s imulus is igh ly con olled in closed-
loop o he ac ions o he animal. We used a p ojec ion se up simila o
he low-cos mul i- ouch sensing able p oposed by (Han, 2005), whe e a
isible ligh ea -p ojec ion sys em is coupled wi h in a ed illumina ion
and an in a ed imaging senso o de ec in eal- ime whe e he animal is
loca ed wi h espec o he isual display su ace (Video 2.2).
2.5 Discussion
A e abou a yea o using Bonsai in an ac i e neu oscience esea ch ins i-
u e, dozens o die en expe imen al p o ocols and da a analysis pipelines
ha e been success ully implemen ed using he p o ided building blocks
(Gou êa, Mon ei o, Soa es, A allah & Pa on, 2014; I sko e al., 2014;
Tecuape la e al., 2014). We we e su p ised by he di e si y o applica-
ions and by he pace a which new modules and de ices we e de eloped
and in eg a ed.
79
The pe o mance achie ed by Bonsai da aow p ocessing was an im-
po an conside a ion h oughou . Video p ocessing can be pa icula ly
challenging o handle gi en he bandwid h equi ed o quickly acqui e and
p ocess la ge da a ma ices. In o de o co ela e con inuous measu es o
beha iou wi h neu al ac i i y, i is use ul o hose measu emen s o ha e
bo h high spa ial and high empo al esolu ion. Using Bonsai, we we e
able o simul aneously p ocess and comp ess g ayscale image sequences
om high esolu ion (
1280 ×960
) and high ame a e (120 Hz) came as
using s anda d o- he-shel desk op compu e s (In el Co e i7, 8 GB RAM).
In ac , many o he epo ed assays use mul iple (>2) such ideo s eams
wi h success and ac ually p ocess he beha iou ideo online ei he o con-
ol s a es o he beha iou p o ocol o o p e-p ocess ideo da a o oine
analysis.
One o he a eas whe e we see he applica ion o Bonsai becoming
mos signican is in he de elopmen o dynamic beha iou assays (en-
i onmen s) using eac i e con ol s a egies. B ains e ol ed o gene a e
and con ol beha io s ha can deal wi h he complexi y o he na u al
wo ld. Howe e , when neu oscien is s y o in es iga e hese beha io s in
he lab, i is o en dicul o design equi alen en i onmen al complexi y
in a con olled manne . As an example, conside a simple o aging scen-
a io in which a land animal mus collec , in a imely manne , ood i ems
ha become a ailable a andom in e als in many si es. I he i em is
no collec ed in ime, i o s o ge s ea en by compe i o s. In he case o
a single o aging si e, a FSM desc ip ion in ui i ely ep esen s he wo k-
ings o he en i onmen (Figu e 2.9A). Howe e , le us now conside a
si ua ion whe e he en i onmen has wo o hese ood si es ope a ing in-
dependen ly, hus in oducing he possibili y o die en e en s occu ing
simul aneously a each o he si es. I ou en i onmen is modeled as a
ni e-s a e machine, hen we mus ep esen e e y possible combina ion
o s a es and ansi ions, as in Figu e 2.9B. In he classical s a e machine
80
Figu e 2.9.
Desc ibing he beha iou o dynamic en i onmen s using ei he
s a e-machines o da aows.
o malism he machine can only be in one s a e a a ime, which means
we now need o model each s a e as he combina ion o he indi idual
independen s a es a each ewa d loca ion. Fu he mo e, because ans-
i ions be ween hese s a es a e asynch onous and independen , we hus
ha e edges be ween nea ly e e y pai o nodes, as each ewa d si e can
change i s s a e a any poin in ime ela i e o he o he .
How would designing such a scena io eel like in a eac i e p og am-
ming language? Figu e 2.9C shows a possible specica ion o he 1-si e
o aging ask in eac i e e ms. In his case, we ha e wo sou ces o e en s
om he en i onmen : one ime signaling he a ailabili y o ewa d (A);
and a sampling e en (S) which is igge ed e e y ime he animal checks
81
he loca ion o ood. Bo h o hese e en s can occu independen ly o
each o he , bu when a sampling e en coincides wi h ewa d a ailabili y
(C), hen ewa d (R) is deli e ed. Because his desc ip ion is in insically
asynch onous and pa allel, i makes i ex emely easy o scale he ask
o a la ge se o loca ions: jus eplica e he da aow o each o he
o he loca ions (Figu e 2.9D). In his example, he design space was made
mo e in ui i e by in oducing he pa allel and asynch onous na u e o a
eal-wo ld si ua ion in o ou modeling o malism.
Ano he dicul y o he classical s a e machine o malism is dealing
wi h con inuous a iables. The na u al en i onmen p o ides cons an
eal- ime eedback ha igh ly co ela es wi h he ac ions o an animal.
Rep oducing such closed-loop in e ac ion and manipula ing i s dynamics
is a necessa y ool o ully in es iga ing b ain unc ion. Such models a e
i ually impossible o ep esen in a machine o ni e s a es, gi en he po-
en ial inni ude o eedback esponses. Howe e , he da aow o malism
o asynch onous e en sou ces can easily accommoda e such models. In
ac , his is hei na u al ba leg ound; nodes ep esen eac i e ope a o s
ha p omp ly espond o inpu alues b oadcas ed by e en sou ces. These
models o asynch onous compu a ion a e hus ideal o ec ea ing he com-
plex disc e e and con inuous aspec s o na u al en i onmen s ha b ains
e ol ed o mas e . We hus p opose Bonsai as a new ool o neu oscien -
is s ying o unde s and how he b ain deals wi h eal wo ld complexi y.
2.6 Acknowledgemen s
We hank João Bá olo Gomes o sugges ing he name Bonsai; Danbee
Kim o ea ly discussions on c ea ing i ual en i onmen s o oden s;
Joana Noguei a, Geo ge Dimi iadis and all he membe s o he In elligen
Sys ems Labo a o y o help ul discussions and commen s on he manu-
sc ip . We also hank all he membe s o he Champalimaud Neu oscience
82
P og amme who used Bonsai o se up hei da a analysis and acquisi ion
expe imen s and in so doing p o ided aluable eedback o imp o e he
amewo k. The esea ch leading o hese esul s has ecei ed unding
om he Eu opean Union's Se en h F amewo k P og amme (FP7/2007-
2013) unde g an ag eemen no. 600925 and he Bial Founda ion (G an
190/12). GL is suppo ed by he PhD S uden ship SFRH/BD/51714/2011
om he Founda ion o Science and Technology. The Champalimaud
Neu oscience P og amme is suppo ed by he Champalimaud Founda ion.
2.7 Au ho con ibu ions
Concei ed and designed he modula box: GL, ARK; Concei ed and de-
eloped he Bonsai amewo k: GL; Technical ad iso y boa d: NB, JF;
Concei ed and de eloped he expe imen al applica ions: GL, NB, JF, JPN,
BVA, SS, LM, SM, PMI, PAC, REM, LC, ED, JJP, ARK; Pe o med and
analyzed expe imen s: GL, JF, JPN, BVA, SS, LM, SM, PMI, PAC, REM,
LC, ED; W o e he Bonsai manusc ip : GL, ARK.
83

Re e ences
Bainomugisha, E., Ca e on, A. L., Cu sem, T. V., Mos inckx, S. & Meu-
e , W. D. (2013). A su ey on eac i e p og amming.
ACM Com-
pu ing Su eys
,
45
(4), A icle No. 52. doi:10.1145/2501654.2501666.
a Xi : 1005.3014
Banzi, M., Cua ielles, D., Igoe, T., Ma ino, G. & Mellis, D. (2014). A -
duino.
Cook, S., Jones, G., Ken , S. & Wills, A. C. (2007).
Domain Specic De el-
opmen wi h Visual S udio DSL Tools
. Addison-Wesley P o essional.
Da idson, T. J., Kloos e man, F. & Wilson, M. A. (2009). Hippocam-
pal Replay o Ex ended Expe ience.
Neu on
,
63
(4), 497507. doi:10.
1016/j.neu on.2009.07.027
Ellio , C., Vijayakuma , V., Zink, W. & Hansen, R. (2007). Na ional In-
s umen s LabVIEW: A P og amming En i onmen o Labo a o y
Au oma ion and Measu emen .
Jou nal o Labo a o y Au oma ion
,
12
(1), 1724. doi:10.1016/j.jala.2006.07.012
Fos e , D. J. & Wilson, M. a. (2006). Re e se eplay o beha iou al se-
quences in hippocampal place cells du ing he awake s a e.
Na u e
,
440
(7084), 6803. doi:10.1038/na u e04587
Ga ido-Ju ado, S., Munõz-Salinas, R., Mad id-Cue as, F. J. & Ma ín-
Jiménez, M. J. (2014). Au oma ic gene a ion and de ec ion o highly
eliable ducial ma ke s unde occlusion.
Pa e n Recogni ion
,
47
(6),
22802292.
Gomez-Ma in, A., Pa on, J. J., Kamp, A. R., Cos a, R. M. & Mainen,
Z. F. (2014). Big beha io al da a: psychology, e hology and he
ounda ions o neu oscience.
Na u e Neu oscience
,
17
(11), 1455
1462. doi:10.1038/nn.3812
Gou êa, T. S., Mon ei o, T., Soa es, S., A allah, B. V. & Pa on, J. J.
(2014). Ongoing beha io p edic s pe cep ual epo o in e al du -
84
a ion.
F on ie s in neu o obo ics
,
8
(Ma ch), 10. doi:10.3389/ nbo .
2014.00010
Han, J. (2005). Low-cos mul i- ouch sensing h ough us a ed o al in-
e nal eec ion.
P oceedings o he 18 h annual ACM symposium on
Use in e ace so wa e and echnology (UIST '05)
, 115118.
I onPy hon Communi y. (2014). I onPy hon.
I seez. (2014). OpenCV (Open Sou ce Compu e Vision).
I sko , P. M., Mo ei a, J.-M., Vinnik, E., Lopes, G., Sa a ik, S., Dickinson,
M. H. & Ribei o, C. (2014). Au oma ed moni o ing and quan i a i e
analysis o eeding beha iou in D osophila.
Na u e Communica ions
,
5
, 4560. doi:10.1038/ncomms5560
Johns on, W. M., Hanna, J. R. P. & Milla , R. J. (2004). Ad ances in
da aow p og amming languages.
ACM Compu ing Su eys
,
36
(1),
134. doi:10.1145/1013208.1013209
Ka o, H. & Billinghu s , M. (1999). Ma ke acking and hmd calib a ion
o a ideo-based augmen ed eali y con e encing sys em. In
Aug-
men ed eali y, 1999.(iwa '99) p oceedings. 2nd ieee and acm in e -
na ional wo kshop on
(pp. 8594). IEEE.
Kumu a, T., Nakamu a, Y., Ishiu a, N., Takeuchi, Y. & Imai, M. (2012).
Model Based Pa alleliza ion om he Simulink Models and Thei
Sequen ial C Code. In
P oc. o he 17 h wo kshop on syn hesis and
sys em in eg a ion o mixed in o ma ion echnologies (sasimi 2012)
(pp. 186191).
Lopes, G., Bonacchi, N., F azão, J., Ne o, J., A allah, B., Soa es, S., Mo -
ei a, L., Ma ias, S., I sko , P., Co eia, P., Medina, R., Calca e a, L.,
D eos i, E., Pa on, J. & Kamp, A. (2015). Bonsai: An e en -based
amewo k o p ocessing and con olling da a s eams.
F on ie s in
Neu oin o ma ics
,
9
(7). doi:10.3389/ nin .2015.00007
Ma hWo ks. (2014). Simulink.
85
Mic oso Open Technologies. (2014). Reac i e Ex ensions o he .NET
amewo k (Rx).
Mosconi, M. & Po a, M. (2000). I e a ion cons uc s in da a-ow isual
p og amming languages.
Compu e Languages
,
26
(2000), 67104.
Na ional Ins umen s. (2014). LabVIEW.
O'Kee e, J. & Dos o sky, J. (1971). The hippocampus as a spa ial map:
P elimina y e idence om uni ac i i y in he eely mo ing a .
B ain esea ch
,
34
(1), 171175.
Selle s, K. K., Benne , D. V., Hu , A., Williams, J. H. & F öhlich,
F. (2015). Awake s. anes he ized: laye -specic senso y p ocessing
in isual co ex and unc ional connec i i y be ween co ical a eas.
Jou nal o Neu ophysiology
,
113
(10), 37983815. doi:10.1152/jn.
00923.2014
Tecuape la, F., Ma ias, S., Dugue, G. P., Mainen, Z. F. & Cos a, R. M.
(2014). Balanced ac i i y in basal ganglia p ojec ion pa hways is c i -
ical o con a e si e mo emen s.
Na u e communica ions
,
5
, 4315.
doi:10.1038/ncomms5315
Voig s, J., Siegle, J. H., Keme e, C., Moo e, C. & Wilson, M. (2013).
A low-cos , open-sou ce sys em o combining high-channel coun
elec ophysiology wi h closed-loop op ogene ic eedback. In
Socie y
o neu oscience (san diego, ca, 13 no embe )
.
W igh , M., F eed, A. & Momeni, A. (2003). Open Sound Con ol: S a e o
he A 2003. In
P oceedings o he 2003 con e ence on new in e aces
o musical exp ession (nime-03)
(pp. 153159). Mon eal, Canada.
86
Chap e 3
Mo ing wi h and wi hou
Mo o Co ex
I  s became scep ical o he supposed pa h o he condi ioned
eex when I ound ha a s, ained in a die en ial eac ion
o ligh , showed no educ ion in accu acy o pe o mance when
almos he en i e mo o co ex, along wi h he on al poles o
he b ain, was emo ed.
Ka l S. Lashley
,
In Sea ch o he Eng am
(1950)
87
measu ed ela i e o B egma, and la e used o econs uc he p ecise
placemen o all g id elec odes in he b ain. A he end o he su ge y, a
i anium sc ew was inse ed pos e io ly o he c anio omy in con ac wi h
he b ain in o de o be used as e e ence o he eco ding sys em. The
s abili y o he implan depends c i ically on he absence o mo emen in
he bony pla es o he skull du ing de elopmen , which can comp omise
he mechanical xa ion o he eco ding chambe o he head (Dimi iadis
e al., 2014). Fo his eason, i is ecommended ha a s unde going his
p ocedu e should be olde han 7 mon hs (Dimi iadis e al., 2014).
Beha iou assay:
Du ing each session he animal was placed inside a
beha iou box o
30 min
, whe e i could collec wa e ewa ds by shu ling
back and o h be ween wo nose pokes (Island Mo ion Co po a ion, USA).
To do his, animals had o c oss a
48 cm
obs acle cou se composed o eigh
2 cm
aluminium s eps spaced by
4 cm
(Figu e 3.1). The s uc u e o he
assay and each s ep in he obs acle cou se was buil ou o aluminium
s uc u al aming (Bosch Rex o h, DE,
20 mm
se ies). The walls o he
a ena we e ab ica ed wi h a lase -cu e om
5 mm
hick opaque black
ac ylic and xed o he s uc u al aming. A anspa en ac ylic window
pa i ion was posi ioned in on o he obs acle cou se in o de o p o ide
a clea iew o he animal. All expe imen s we e un in he da k by ha ing
he beha iou al appa a us enclosed in a ligh igh box.
A mo o ized b ake allowed us o lock o elease each s ep in he obs acle
cou se (Figu e 3.2). The sha o each o he obs acles was coupled o an
ac ylic piece used o con ol he o a ional s abili y o each s ep. In o de
o lock a s ep in a xed posi ion, wo se o mo o s a e ac ua ed o p ess
agains he ac ylic piece and hold i in place. Two o he ac ylic pieces
we e used as s ops o ensu e a maximum o a ion angle o app oxima ely
+/-
100◦
. Two small nu s we e a ached o he bo om o each s ep o
wo k as a coun e weigh ha gi es he obs acles a endency o e u n o
hei o iginal a congu a ion. In o de o ensu e ha noise om se o
94

s able
uns able
andom
Figu e 3.1.
Schema ic o he appa a us and summa y o he die en
condi ions in he shu ling p o ocol.
mo o ac ua ion could no be used as a cue o ell he animal abou he
s a e o each s ep, he mo o s we e always se o p ess agains an ac ylic
piece, ei he he piece ha keeps he s ep s abilized, o he ac ylic s ops.
A he beginning o each ial, he mo o s we e un h ough a andomized
sequence o posi ions in o de o mask in o ma ion abou s a e ansi ions
and also o ensu e he s eps we e ese o hei o iginal congu a ion.
Con ol o he mo o s was done using a Mo o uino boa d (A ica, PT)
along wi h a cus om wo kow w i en in he Bonsai isual p og amming
language (Lopes e al., 2015).
P io o he mic o-ECoG eco dings, each s ep in he obs acle cou se
was ou  ed wi h a mic o load cell (CZL616C, Phidge s, CA) secu ed
be ween he s ep on holde and he base (Figu e 3.2). This allowed us
95
se o mo o s
load cell (0-750g)
Figu e 3.2.
Schema ic o he s ep locking mechanism and load cell senso .
o eco d a a ying ol age signal p opo ional o he load applied by he
animal on each s ep. This load signal was acqui ed simul aneously on all
eigh s eps and digi ized synch onously wi h he ECoG da a acquisi ion
sys em.
Da a acquisi ion:
The beha iou o he animals was eco ded wi h a
high-speed and high- esolu ion ideog aphy sys em (1280x680 @
120 Hz
)
using an in a ed came a (Flea3, Poin G ey, CA), supe -b igh in a ed
LED on ligh s (SMD5050, 850 nm) and a a i- ocal lens (Fujinon, JP)
posi ioned in on o he anspa en window pa i ion. A op iew o
he assay was simul aneously eco ded wi h he same sys em a a lowe
ame- a e (
30 Hz
) o moni o ing pu poses. All ideo da a was encoded
wi h MPEG-4 comp ession o subsequen oine analysis. Beha iou da a
acquisi ion o he nose poke beam b eaks was done using an A duino
boa d (Uno, A duino, USA) and s eamed o he compu e ia USB. All
ideo and senso da a acquisi ion was eco ded in pa allel using he same
Bonsai wo kow used o con ol he beha iou assay.
Fo he mic o-ECoG eco dings, all elec ophysiological signals we e
amplied, digi ized and mul iplexed using wo 64-channel amplie boa ds
(RHD2164, In an Technologies, US) connec ed o he elec ode in e ace
boa d (EIB) on he eco ding chambe . The amplie boa ds we e hen
96
connec ed h ough a dual heads age adap e (C3440, In an Technologies,
US) o he main da a acquisi ion USB in e ace boa d (RHD2000-E al,
In an Technologies, US). In o de o acili a e he ee mo emen o he
animal in he beha iou box, he single cable connec ing he head o he an-
imal o he USB in e ace boa d was passed h ough a slip ing (MMC235,
Moon, CN) and hooked in o a nylon s ing c ossing he op o he assay.
In his way, mo emen and o a ion o he e he ed animal we e com-
pensa ed o a oid unwan ed s ain and wis ing on he cables du ing he
en i e eco ding pe iod.
In o de o synch onize he ideog aphy and ECoG eco ding sys ems,
we connec ed he s obe ou pu o he came a o a digi al inpu in he In an
USB in e ace boa d using a GPIO cable (ACC-01-3000, Poin G ey, CA).
The came a s obe ou pu is elec onically coupled o indi idual ame
exposu es (i.e. shu e opening and closing e en s), and can be used o
sub-millisecond eadou o indi idual ame acquisi ion imes. The s obe
signal was acqui ed and digi ized synch onously wi h ECoG da a acquis-
i ion, and used o
pos -hoc
econs uc ion o p ecise ame iming. Da a
acquisi ion om he USB in e ace boa d was eco ded using a Bonsai
wo kow and ca e was aken ha i was always s a ed  s and e min-
a ed las in o de o ensu e ha no ex e nal synch oniza ion e en s we e
los .
Beha iou p o ocol:
The animals we e kep in a s a e o wa e
dep i a ion o
20 h
p io o each daily session. Fo e e y ial, a s we e
deli e ed a
20
µ
L
d op o wa e . A he end o each day, hey we e gi en
ee access o wa e o
10 min
be o e ini ia ing he nex dep i a ion pe iod.
Sessions las ed o six days o he week om Monday o Sa u day, wi h a
day o ee access o wa e on Sunday. Be o e he s a o he wa e dep i a-
ion p o ocol, animals we e un on a single habi ua ion session whe e hey
we e placed in he box o a pe iod o
15 min
.
97
The ollowing sequence o condi ions we e p esen ed o he animals
o e he cou se o a mon h (see also Figu e 3.1): day 0, habi ua ion o
he box; day 1-4, all he s eps we e xed in a s able congu a ion; day
5, 20 ials o he s able congu a ion, a e which he wo cen e s eps
we e made uns able (i.e. ee o o a e); day 6-10, he cen e wo s eps
emained uns able; day 11, 20 ials o he uns able congu a ion, a e
which he wo cen e s eps we e again xed in a s able s a e; day 12, all
he s eps we e xed in a s able congu a ion; day 13-16, he s a e o he
cen e wo s eps was andomized on a ial-by- ial basis o be ei he s able
o uns able. Following he end o he andom p o ocol, animals con inued
o be es ed in he assay o a a iable numbe o days (up o one week)
in die en condi ions. A he end o he es ing pe iod, all animals we e
exposed o a nal session whe e all s eps we e made ee o o a e in o de
o assay locomo ion pe o mance unde challenging condi ions.
Fo he mic o-ECoG eco dings, he basic beha iou p o ocol was ad-
jus ed o allow o ex a eco ding ime du ing condi ions o in e es . Fi s ,
all session imes we e doubled o he eco dings (e.g.
30 min
o he ha-
bi ua ion session, and
60 min
o all o he sessions). Second, he numbe
o days on each condi ion was also ex ended o allow ex ac ing mo e i-
als om each animal o analysis. Finally, he condi ion whe e he cen e
wo s eps we e eliably uns able was eplaced wi h a condi ion o a e in-
s abili y. In his condi ion, a e he animal is exposed o an uns able
congu a ion, he s eps a e e e ed back o being s able o ano he 20
ials, a e which hey become again uns able o one ial, and so on.
Da a analysis:
All sc ip s and cus om code used o da a analysis
a e a ailable online
1
. The aw ideo da a was  s p e-p ocessed using
a cus om Bonsai wo kow in o de o ex ac ea u es o in e es (Figu e
3.3). T acking o he nose was achie ed by backg ound sub ac ion and
connec ed componen labelling o segmen ed image elemen s. Fi s we
1
h ps://bi bucke .o g/kamp-lab/shu ling-analysis
98
Figu e 3.3.
Example ideo ame om he beha iou acking sys em.
Colou ed o e lays ep esen egions o in e es and ea u e aces ex ac ed
au oma ically om he ideo.
compu e he ellipse bes - o he la ges objec in he image. We hen
ma k he ip o he nose as he u he mos poin , in he segmen ed shape
o he animal, along he majo axis o he ellipse. In o de o analyse
s epping pe o mance, egions o in e es we e dened a ound he su ace
o each s ep and in he gaps be ween he s eps. Backg ound sub ac ed
ac i i y o e hese egions was eco ded o e e y ame o subsequen
de ec ion and classica ion o s eps and slips.
Analysis ou ines we e un using he NumPy scien ic compu ing pack-
age ( an de Wal , Colbe & Va oquaux, 2011) and he Pandas da a
analysis lib a y (McKinney, 2010) o he Py hon p og amming language.
C ossings we e au oma ically ex ac ed om he nose ajec o y da a by
 s de ec ing consecu i e ime poin s whe e he nose was posi i ely iden-
ied in he ideo. In o de o hese pe iods o be success ully ma ked as
c ossings, he s a ing posi ion o he nose mus be loca ed on he opposi e
side o he ending posi ion. Inside each c ossing, he momen o s epping
wi h he o elimb on he cen e s eps was ex ac ed by looking a he  s
peak abo e a h eshold in he  s de i a i e o he ac i a ion signal in
99

he co esponding egion o in e es . False posi i e classica ions due o
hindlimb o ail ac i a ions we e elimina ed by en o cing he cons ain
ha he posi ion o he head mus be loca ed be o e he nex s ep. Visual
con ma ion o he classied imepoin s showed ha spu ious ac i a ions
we e all bu elimina ed by his p ocedu e as s epping wi h he hindlimb
o ail equi es he head o be u he ahead in space unless he animal
u ned a ound (in which case he ajec o y would no be ma ked as a
c ossing anyway). The posi ion o he nose a he momen o each s ep
was ex ac ed and ound o be no mally dis ibu ed, so s a is ical analysis
o he s ep pos u e in he andom condi ion used an unpai ed - es o
check o independence o die en measu emen g oups.
In o de o e alua e he dynamics o c ossing in he andom condi ion,
we  s measu ed o e e y ial he speed a which he animals we e
mo ing on each spa ial segmen o he assay. To minimize o e all ial-by-
ial a ia ion in indi idual animal pe o mance, we used he a e age speed
a which he animal app oached he manipula ed s ep as a baseline and
sub ac ed i om he speed a each indi idual segmen . To summa ize
die ences in pe o mance be ween s able and uns able ials, we hen
compu ed he a e age speed p ole o each condi ion, and hen sub ac ed
he a e age speed p ole o uns able ials om he a e age speed p ole
o s able ials. Finally, we compu ed he sum o all hese speed die ences
a e e y segmen in o de o ob ain he speedup index o each animal, i.e.
an index o whe he he animal ends o accele a e o decele a e ac oss he
assay on s able e sus uns able ials.
Fo he mic o-ECoG expe imen s, e oked po en ials we e analysed by
spli ing he aw physiological ol age aces in o
750 ms
windows, whe e
ime ze o was aligned o he momen o s epping wi h he o elimb on
one o he obs acles in he cou se (see below). Each indi idual ime se ies
was low-pass l e ed a
50 Hz
(4 h o de Bu e wo h l e , wo-pass) and
baselined by sub ac ing he a e age o he  s
250 ms
be o e e en onse
100
in o de o compensa e o cons an ol age shi s be ween he wo g ids.
Some o he channels in each g id we e en i ely excluded om he ana-
lysis due o po en ially damaged su ace con ac s, as e idenced by wide
ampli ude, andom oscilla o y beha iou , which was o en ma ched by he
p esence o high impedance measu emen s ex ac ed om he elec ode si e
in i o. In one o he sessions, he cable connec ing he heads age o he
in e ace boa d was acciden ally emo ed by he animal, and all he ials
alling du ing his pe iod had o be excluded om analysis. Co espond-
ence be ween indi idual ECoG samples and ideo ames was compu ed by
ma ching he indi idual ha dwa e ame coun e wi h he sequence o all-
ing edges de ec ed in he shu e s obe signal acqui ed om he in a ed
came a.
Video classica ion:
Classica ion o paw placemen aul s (i.e.
slips) was pe o med in semi-au oma ed ashion. Fi s , possible slip
imepoin s we e de ec ed au oma ically using he peak de ec ion me hod
ou lined abo e. All cons ain s on head posi ion we e elaxed o his
analysis in o de o exclude he possibili y o alse nega i es. A human
classie hen p oceeded o manually go h ough each o he slip candida es
and inspec he ideo a ound ha imepoin in o de o assess whe he
he ac i a ion peak was a genuine paw placemen aul . Examples o alse
posi i es include ail and head ac i a ions as well as paw ac i a ions ha
occu while he animal is ac i ely engaged in explo a ion, ea ing, o o he
ac i i ies ha a e un ela ed o c ossing he obs acles.
A simila echnique was used o de ec and classi y he e en onse s o
he analysis o e oked po en ials in he mic o-ECoG expe imen s. In his
case, a p elimina y classica ion o each ideo ame in o le and igh
o elimb was achie ed by  s compu ing he b igh ness his og am o each
ame, which was used o encode he image as a lowe -dimensional ec o .
The ec o s o all s ep ames we e subsequen ly clus e ed using K-means
and hen manually inspec ed o label co ec ion.
101
Classica ion o beha iou esponses ollowing  s exposu e o he un-
s able condi ion was done on a ame-by- ame analysis o he high-speed
ideo aligned on  s con ac wi h he manipula ed s ep. The ame o
 s con ac was dened as he  s ame in which he e is no iceable
mo emen o he s ep caused by animal con ac . Th ee main ca ego -
ies o beha iou we e obse ed o ollow he  s con ac : compensa ion,
in es iga ion and hal ing. Beha iou sequences we e  s classied as be-
longing o one o hese ca ego ies and hei onse s and ose s de e mined
by he ollowing c i e ia. Compensa ion beha iou is dened by a apid
and adap i e pos u al co ec ion o he locomo ion pa e n in esponse o
he pe u ba ion. Onse o his beha iou is dened by he  s ame in
which he e is isible apid con ac ion o he body muscula u e ollowing
 s con ac . In es iga ion beha iou consis s o pe iods o a ge ed in e -
ac ion wi h he s eps, o en in ol ing manipula ion o he eely mo ing
obs acle wi h he o epaws. The onse o his beha iou is dened by he
animal o ien ing i s head down o one o he manipula ed s eps, ollowed
by subsequen in e ac ion. Hal ing beha iou is cha ac e ized by a pe iod
in which he animal s ops i s ongoing mo o p og am, and main ains he
same body pos u e o se e al seconds, wi hou swi ching o a new beha-
iou o o ien ing specically o he manipula ed s eps. This beha iou is
dis inc om a eezing esponse, as occasional mo emen s o he head a e
seen. Onse o his beha iou is dened by he momen whe e locomo ion
and o he mo o ac i i ies besides mo emen o he head come o a s op.
A human classie blind o he lesion condi ion was gi en desc ip ions o
each o hese h ee main ca ego ies o beha iou and asked o no e onse s
and ose s o each beha iou h oughou he ideos. These classica ions
p o ide a isual summa y o he  s esponse ideos; he comple e da ase
used o his classica ion is included as supplemen a y mo ies.
102
3.4 Resul s
To in es iga e whe he he in ac mo o co ex is equi ed o he obus
con ol o mo emen in esponse o unexpec ed pe u ba ions, we designed
a econgu able dynamic obs acle cou se whe e indi idual s eps can be
made s able o uns able on a ial-by- ial basis (Figu e 3.1, also see Me h-
ods). In his assay, a s shu le back and o h ac oss he obs acles, in he
da k, in o de o collec wa e ewa ds. We specically designed he assay
such ha modica ions o he physics o he obs acles could be made co -
e ly. In his way, he animal has no explici in o ma ion abou he s a e
o he s eps un il i ac ually con ac s hem. Wa e dep i ed animals we e
ained daily o 4 weeks, h oughou which hey encoun e ed inc easingly
challenging s a es o he obs acle cou se. Ou goal was o cha ac e ize p e-
cisely he condi ions unde which mo o co ex becomes necessa y o he
con ol o mo emen , and his mo i a ed us o in oduce an en i onmen
wi h g aded le els o unce ain y.
We compa ed he pe o mance o 22 animals: 11 wi h bila e al ibo enic
acid lesions o he p ima y and seconda y o elimb mo o co ex, and 11
age and gende ma ched con ols (5 sham su ge y, 6 wild- ypes). Animals
we e gi en ample ime o eco e , 4 weeks pos -su ge y, in o de o specic-
ally isola e beha iou s ha a e ch onically impai ed in animals lacking he
unc ions enabled by mo o co ical s uc u es. His ological examina ion o
se ial co onal sec ions e ealed signican a iabili y in he ex en o dam-
aged a eas (Figu e 3.4), which was likely caused by mechanical blockage
o he injec ion pipe e du ing lesion induc ion a some si es. Ne e he-
less, olume econs uc ion o he se ial sec ions allowed us o accu a ely
quan i y he size o each lesion, iden i y each animal ( om Lesion A o
Lesion K; la ges o smalles ), and use hese alues o compa e obse ed
beha iou al eec s as a unc ion o lesion size.
103
p og ession
heigh
A
s able uns able s able andom
0
1
2
3
4
5
6
7
8
p og ession (cm)
s ep pos u e ac oss ials (Con ol A)
B
Figu e 3.8.
Measu ing pos u al app oach o he obs acles. (
A
) Schema ic
o pos u al analysis image p ocessing. The posi ion o he animal's nose is
ex ac ed whene e he paw ac i a es he ROI o he  s manipula ed s ep
(see me hods). (
B
) The ho izon al posi ion, i.e. p og ession, o he nose
in single ials o one o he con ol animals s epping ac oss he die en
condi ions o he shu ling p o ocol.
pos u e o he animals when s epping on he obs acles changed signican ly
o e ime (Figu e 3.8B, 3.9A). Specically, he cen e o g a i y o he body
was shi ed u he o wa d and highe du ing la e sessions, in a manne
p opo ional o pe o mance. Howe e , a e he obs acles changed o he
uns able s a e, we obse ed an immedia e and pe sis en adjus men o
his c ossing pos u e, wi h animals assuming a lowe cen e o g a i y and
educing hei speed as hey app oached he uns able obs acles (Figu e
3.9A,B). In e es ingly, we also no iced ha a g oup o animals adop ed a
die en s a egy. Ins ead o lowe ing hei cen e o g a i y, hey ei he
kep i unchanged o shi ed i e en mo e o wa d and pe o med a jump
o e he uns able obs acles (Figu e 3.10A,B). These wo s a egies we e
ema kably consis en ac oss he wo g oups, bu he e was no co ela ion
be ween he s a egy used and he deg ee o mo o co ical lesion (Figu e
110

s able uns able s able andom
1.5
1.0
0.5
0.0
0.5
1.0
1.5
2.0
2.5
p og ession (cm)
* * *
a e age s ep pos u e
con ol
lesion
A
4 2 0 2 4
p og ession (cm)
40
20
0
20
40
60
80
speed (cm / s)
0
200
n = 3550 ials
0 200 400
s able
uns able
B
4 2 0 2 4
p og ession (cm)
4
2
0
2
4
6
heigh (cm)
0
100
200 n = 1995 ials
0 200
s able
uns able
con ols
C
4 2 0 2 4
p og ession (cm)
4
2
0
2
4
6
heigh (cm)
0
50
100
n = 1555 ials
0 100 200
s able
uns able
lesionsD
Figu e 3.9.
Ra s adap hei pos u al app oach o he obs acles a e a
change in physics. (
A
) A e age ho izon al posi ion o he nose ac oss he
die en p o ocol s ages o bo h lesion and con ol animals. As e isks
indica e he a e age nose posi ion on he 20 ials immedia ely p eceding
a change in p o ocol condi ions (see ex ). (
B
) Dis ibu ion o ho izon al
posi ion agains speed o he las wo days o he s able (blue) and un-
s able (o ange) p o ocol s ages. (
C-D
) Dis ibu ion o nose posi ions o
con ol and lesion animals o e he same sessions.
111
Con ol A
s able
uns able
Lesion B
Con ol H Lesion C
A
10 15 20 25 30 35 40
p og ession (cm)
0
1
2
3
4
5
6
heigh (cm)
c ossing ajec o ies (n = 21)
con ol
lesion
B
200 250 300 350 400 450 500 550 600 650
weigh (g)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
p (skip middle s eps)
co ela ion o jumping wi h weigh
con ol
lesion
C
Figu e 3.10.
Animals use die en s a egies o dealing wi h he un-
s able obs acles. (
A
) Example a e age p ojec ion o all pos u e images
o s able (g een) and uns able ( ed) sessions o wo non-jumpe ( op)
and wo jumpe (bo om) animals. (
B
) A e age nose ajec o ies o indi-
idual animals c ossing he uns able condi ion. The shaded a ea a ound
each line ep esen s he 95% condence in e al. (
C
) Co ela ion o he
p obabili y o skipping he cen e wo s eps wi h he weigh o he animal.
3.9C,D, 3.10C). In ac , we ound ha he use o a jumping s a egy was
bes p edic ed by he body weigh o he animal (Figu e 3.10C).
Du ing he wo days whe e he s able s a e o he en i onmen was
eins a ed, he pos u e o he animals was g adually es o ed o p e-
manipula ion le els (Figu e 3.8B, 3.9A), al hough in many cases his
adjus men happened a a slowe a e han he ansi ion om s able o
uns able. Again, his pos u al adap a ion was independen o he p esence
o absence o o epaw mo o co ex.
We nex looked in de ail a he days whe e he s a e o he obs acle
cou se was andomized on a ial-by- ial basis. This s age o he p o ocol
is pa icula ly in e es ing as i eec s a si ua ion whe e he en i onmen
has a pe sis en deg ee o unce ain y. Fo his analysis, we we e o ced o
exclude he animals ha employed a jumping s a egy, as hei expe ience
wi h he manipula ed obs acles was he same i espec i e o he s a e o
he wo ld. Fi s , we epea ed he same pos u e analysis compa ing all
112
he s able and uns able ials in he andom p o ocol in o de o con ol
o whe he he e was any sub le cue in ou mo o ized se up ha he an-
imals migh be using o gain in o ma ion abou he cu en s a e o he
wo ld. The e was no signican die ence be ween andomly p esen ed
s able and uns able ials on he app oach pos u e o he animal (Figu e
3.11A). Howe e , classi ying he ials on he basis o pas ial his o y e-
ealed a signican eec on pos u e (Figu e 3.11B). This sugges ed ha
he animals we e adjus ing hei body pos u e when s epping on he a -
ec ed obs acles on he basis o hei cu en expec a ion abou he s a e
o he wo ld, which is upda ed by he p e iously expe ienced s a e. Su -
p isingly, his eec again did no depend on he p esence o absence o
on al mo o co ical s uc u es (Figu e 3.11C,D).
Finally, we decided o es whe he gene al mo o pe o mance was a -
ec ed by he andomized s a e o he obs acles. I he animals do no know
wha s a e he wo ld will be in, hen he e will be an inc eased challenge
o hei s abili y when hey c oss o e he uns able obs acles, possibly de-
manding a quick change in s a egy when hey lea n whe he he wo ld
is s able o uns able. In o de o e alua e he dynamics o c ossing, we
compa ed he speed p ole o each animal ac oss hese die en condi ions
(Figu e 3.12, see Me hods). In e es ingly, wo o he animals wi h he
la ges lesions appea ed o be signican ly slowed down on uns able ials,
while con ols and he animals wi h he smalles lesions ins ead ended o
accele a e a e encoun e ing an uns able obs acle. Howe e , he o e all
eec o lesions e sus con ols was no s a is ically signican (Figu e
3.12C).
Ne e heless, we we e in igued by his obse a ion and decided o
in es iga e, in de ail, he  s momen in he assay when a pe u ba ion
is encoun e ed. In he andom p o ocol, e en hough he s a e o he
wo ld is unp edic able, he animals know ha he obs acles migh become
uns able. Howe e , he e y  s ime he en i onmen becomes uns able,
113
4 2 0 2 4
p og ession (cm)
4
2
0
2
4
heigh (cm)
0
200
400 p = 0.296
n = 3562 ials
0 500
s able
uns able
cu en s a e
A
4 2 0 2 4
p og ession (cm)
4
2
0
2
4
heigh (cm)
0
100
p = 4.3e-13
n = 1717 ials
0 200 400
p e ious
s able
p e ious
uns able
p e ious s a e
B
4 2 0 2 4
p og ession (cm)
4
2
0
2
4
heigh (cm)
0
100 p = 4.87e-06
n = 1018 ials
0 100 200
p e ious
s able
p e ious
uns able
con ols
C
4 2 0 2 4
p og ession (cm)
4
2
0
2
4
heigh (cm)
0
50 p = 1.12e-09
n = 699 ials
0 100
p e ious
s able
p e ious
uns able
lesions
D
Figu e 3.11.
Animals adjus hei pos u e on a ial-by- ial basis o he
expec ed s a e o he wo ld. (
A
) Dis ibu ion o nose posi ions on he
andomized p o ocol when s epping on he  s manipula ed obs acle, o
ials in which he cu en s a e was s able (blue) o uns able (o ange).
(
B
) Dis ibu ion o nose posi ions o ials in which he p e ious wo
ials we e s able (blue) o uns able (o ange). (
C-D
) Same da a as in (
B
)
spli by he con ol and lesion g oups.
p
alues om S uden 's unpai ed
- es a e indica ed.
114
15 10 5 0 5 10 15 20 25
p og ession (cm)
20
10
0
10
20
30
speed (cm/s)
a e age speed (Con ol A)
s able
uns able
15 10 5 0 5 10 15 20 25
p og ession (cm)
20
10
0
10
20
30
speed (cm/s)
a e age speed (Lesion B)
s able
uns able
con ol lesion
100
50
0
50
100
150
200
speedup (cm/s)
p = 0.18
speed p o ile di e ence
A B C
Figu e 3.12.
Encoun e ing die en s a es o he andomized obs acles
causes he animals o quickly adjus hei mo emen ajec o y. (
A
) Ex-
ample a e age speed p ole ac oss he obs acles o s able (blue) and un-
s able (o ange) ials in he andomized sessions o a con ol animal (see
ex ). The shaded a ea a ound each line ep esen s he 95% condence
in e al. (
B
) Respec i ely o one o he la ges lesions. (
C
) Summa y o
he a e age die ence be ween he speed p oles o s able and uns able
ials ac oss he wo g oups o animals. E o ba s show s anda d e o o
he mean.
p
alue om S uden 's unpai ed - es is indica ed.
he collapse o he obs acles is comple ely unexpec ed and demands an
en i ely no el mo o esponse.
A de ailed analysis o he esponses o he  s collapse o he s eps
e ealed a s iking die ence in he s a egies deployed by he lesion and
con ol animals. Upon he  s encoun e wi h he manipula ed s eps,
we obse ed h ee ypes o beha iou al esponses om he animals (Video
3.6): in es iga ion, in which he animals immedia ely s op hei p og ession
and o ien owa ds, whisk, and physically manipula e he al e ed obs acle;
compensa ion, in which he animals apidly adjus hei beha iou o nego-
ia e he unexpec ed ins abili y; and hal ing, in which he ongoing mo o
p og am ceases and he animals' beha iou simply comes o a s op o
se e al seconds. Rema kably, hese esponses depended on he p esence
o absence o mo o co ex (Figu e 3.13). Animals wi h he la ges mo o
co ical lesions, upon hei  s encoun e wi h he no el en i onmen al
115

0.0 0.5 1.0 1.5 2.0 2.5 3.0
im el om l i s lcon ac l(s)
Lk
Lj
Li
Lh
Lg
L
Le
Ld
Lc
Lb
La
Ck
Cj
Ci
Ch
Cg
C
Ce
Cd
Cc
Cb
Ca
e hog am lo l i s lcon ac lwi hlm anipula edl ail
Con ols
Lesions
Fi s lCon ac
in es iga ion
compensa ion
hal ing
no mallmo ion
Δ
AB
Figu e 3.13.
Responses o an unexpec ed change in he en i onmen . (
A
)
Response ypes obse ed ac oss indi iduals upon  s encoun e ing an un-
p edic ed ins abili y in he s a e o he cen e obs acles. (
B
) E hog am o
beha iou al esponses classied acco ding o he h ee c i e ia desc ibed in
(
A
) and aligned (0.0) on  s con ac wi h he newly manipula ed obs acle.
Black dashes indica e when he animal exhibi s a p onounced ea ick.
Whi e indica es ha he animal has c ossed he obs acle cou se.
obs acle, hal ed o se e al seconds, whe eas animals wi h an in ac mo o
co ex, and hose wi h he smalles lesions, we e able o apidly eac wi h
ei he an in es iga o y o compensa o y esponse (Video 3.7,3.8).
The esponse o animals wi h ex ended lesions was e en mo e s iking.
In wo o hese animals, he e was a ailu e o ecognize ha a change
had occu ed a all (Video 3.9). Ins ead, hey kep walking ac oss he
now uns able s eps o se e al ials, ne e s opping o assess he new si u-
116
a ion. One o hem g adually no iced he manipula ion and s opped his
p og ession, while he o he one only ully ealized he change a e inad-
e en ly hi ing he s eps wi h i s snou (Video 3.9: Ex ended Lesion A).
This was he  s ime we e e obse ed his beha iou , as all animals wi h
o wi hou co ical lesions always displayed a clea swi ch in beha iou al
s a e ollowing he  s encoun e wi h he manipula ion. In he emaining
animals wi h ex ended lesions, wo o hem clea ly hal ed hei p og ession
ollowing he collapse o he obs acles, in a way simila o he la ge mo o
co ex ibo enic lesions (Video 3.10). The hi d animal (Ex ended Lesion
B) ac ually collapsed upon con ac wi h he manipula ed s ep, alling o e
i s paw and digi s awkwa dly and hi ing he obs acles wi h i s snou .
Sho ly a e his he e was a swi ch o an explo a o y beha iou s a e, in
a way simila o Ex ended Lesion A.
Towa ds a neu obiological basis o obus esponses
In o de o in es iga e he neu ophysiological co ela es o hese obus e-
sponses in he mo o co ex, in h ee animals we implan ed exible su ace
elec ode g ids abo e he du a in one hemisphe e o he in ac b ain (Figu e
3.15A, also see Me hods). Each s ep o he obs acle cou se was ou  ed
wi h a load cell senso o measu e he p ecise iming o con ac and he
amoun o weigh placed on each limb du ing locomo ion. The en i e elec-
oco icog aphy (ECoG) sys em was synch onized on a ame-by- ame
basis wi h he high-speed ideo acquisi ion so we could econs uc he
de ailed beha iou o he animal a any poin o he physiological ace as
well as ela e he con inuous load p ole on indi idual s eps wi h die en
phases in he locomo ion cycle (Figu e 3.14).
We  s asked whe he he e we e esponses in he ECoG signal o e
o elimb mo o co ex ha we e modula ed by s epping beha iou . Align-
ing he ECoG aces o he e en o s epping on a pe manen ly s able s ep
wi h he con ala e al paw e ealed he dis inc p esence o an e oked po-
117
0.0 0.1 0.2 0.3 0.4 0.5 0.6
ime (s)
0
1
2
3
4
5
6
load (A.U.)
Figu e 3.14.
Example load cell ac i a ion p ole du ing s epping on a
ung.
en ial on he an e io g id channels ha was absen when s epping wi h
he ipsila e al paw (Figu e 3.15B, op ace). On close inspec ion, i could
be seen ha he beginning o he nega i e deec ion sligh ly p ecedes he
ime o con ac wi h he s ep, sugges ing a non-senso y con ibu ion o
he e oked esponse. Synap ic ac i i y in he long and hick apical dend-
i es o py amidal cells a e hough o be one o he main con ibu o s o
co ically eco ded ex acellula eld po en ials (Buzsáki, Anas assiou &
Koch, 2012). In he ca , a sizeable p opo ion o py amidal ac neu -
ons in he mo o co ex ha e been ound o discha ge hy hmically du ing
unimpeded locomo ion (A ms ong & D ew, 1984; D ew, Jiang, Kably &
La oie, 1996), a phenomenon ha is e y likely o be coupled wi h ob-
se able synap ic ac i i y in he po en ial aces and could accoun o he
s ep-aligned e oked esponses ha we obse ed du ing locomo ion o a s
in he s able obs acle cou se.
118
Nex , we asked whe he he e was any modula ion o he e oked e-
sponse when na iga ing he uns able obs acle cou se. In o de o y and
maximize he numbe o ials in which he encoun e wi h he uns able
s ep is unexpec ed, we adjus ed he beha iou al p o ocol a he ans-
i ion be ween he s able and uns able es pe iods. This ime, ins ead o
pe manen ly swi ching he cen e s eps o he uns able congu a ion, we
decided o immedia ely e e he s eps back o he s able s a e a e he
 s exposu e o he ins abili y. A e 20 subsequen ials in he s able
s a e, he s eps we e again made uns able, and his pa e n was epea ed
o se e al days.
Su p isingly, when we aligned he ECoG aces o con ala e al paw
s eps on he manipula ed obs acle in uns able ials, we obse ed a second
e oked nega i i y, delayed in ime ela i e o he p e iously obse ed s able
s ep e oked esponse, and wi h a much la ge ampli ude ac oss he channels
in he an e io g id (Figu e 3.15B, middle le ace). Rema kably, e en
in he p esence o such a small numbe o ials, he consis ency o he
esponse in e e y ial p o ided a good enough signal- o-noise a io o he
a e age esponse o be clea ly isible. In e es ingly, his nega i i y was
ound o be apidly ollowed by an equally la ge posi i e deec ion in he
po en ial which decayed o baseline wi h a much la ge ime cons an , a
esponse ha was en i ely absen om he e oked po en ial o s epping on
a s able s ep. In con as , he esponse o uns able s eps wi h he ipsila e al
paw did no e eal such la ge deec ions om he baseline, al hough a
consis en nega i i y could s ill be seen ac oss he g id a ound he same
ime poin (Figu e 3.15B, middle igh ace). The ampli ude and iming
o e oked esponses when s epping wi h he con ala e al paw on he same
manipula ed s ep in s able ials was la gely iden ical o he condi ion
whe e he s ep was pe manen ly s able, and again was ound o be absen
when s epping wi h he ipsila e al paw (Figu e 3.15B, bo om ace).
119
McKinney, W. (2010). Da a S uc u es o S a is ical Compu ing in Py-
hon.
P oceedings o he 9 h Py hon in Science Con e ence
, 5156.
Me z, G. A. & Whishaw, I. Q. (2002). Co ical and subco ical lesions
impai skilled walking in he ladde ung walking es : a new ask
o e alua e o e- and hindlimb s epping, placing, and co-o dina ion.
Jou nal o neu oscience me hods
,
115
(2), 16979.
Meye , P. M. & Meye , D. R. (1971). Neu osu gical p ocedu es wi h special
e e ence o aspi a ion lesions. In R. D. Mye s (Ed.),
Me hods in
psychobiology, ol. i
(pp. 91130). London: Academic P ess.
Mille , M. W. (1987). The o igin o co icospinal p ojec ion neu ons in a .
Exp B ain Res
,
67
(2), 339351. doi:10.1007/BF00248554
O chy, T. M., Wol, S. B. E., Rhee, J. Y., Pehle an, C., Kawai, R., Kemp ,
A., Gobes, S. M. H. & Öl eczky, B. P. (2015). Acu e o- a ge eec s
o neu al ci cui manipula ions.
Na u e
. doi:10.1038/na u e16442
Phillips, C. G. (1969). The Fe ie Lec u e, 1968: Mo o Appa a us o
he Baboon's Hand.
P oceedings o he Royal Socie y B: Biological
Sciences
,
173
(1031), 141174. doi:10.1098/ spb.1969.0044
Schindelin, J., A ganda-Ca e as, I., F ise, E., Kaynig, V., Longai , M.,
Pie zsch, T., P eibisch, S., Rueden, C., Saal eld, S., Schmid, B.,
Tine ez, J.-Y., Whi e, D. J., Ha ens ein, V., Elicei i, K., Tomancak,
P. & Ca dona, A. (2012). Fiji: an open-sou ce pla o m o biological-
image analysis.
Na u e me hods
,
9
(7), 67682. doi:10.1038/nme h.
2019
Schwa cz, R., Hök el , T., Fuxe, K., Jonsson, G., Golds ein, M. & Te enius,
L. (1979). Ibo enic acid-induced neu onal degene a ion: A mo pholo-
gical and neu ochemical s udy.
Expe imen al B ain Resea ch
,
37
(2),
199216. doi:10.1007/BF00237708
an de Wal , S., Colbe , S. C. & Va oquaux, G. (2011). The NumPy
A ay: A S uc u e o Ecien Nume ical Compu a ion.
Compu ing
in Science & Enginee ing
,
13
(2), 2230. doi:10.1109/MCSE.2011.37
126

Wa son, B. D., Die ich, W. D., Bus o, R., Wach el, M. S. & Ginsbe g,
M. D. (1985). Induc ion o ep oducible b ain in a c ion by pho o-
chemically ini ia ed h ombosis.
Annals o Neu ology
,
17
(5), 497
504. doi:10.1002/ana.410170513
Whishaw, I. Q. (2000). Loss o he inna e co ical eng am o ac ion pa -
e ns used in skilled eaching and he de elopmen o beha io al com-
pensa ion ollowing mo o co ex lesions in he a .
Neu opha maco-
logy
,
39
(5), 788805. doi:10.1016/S0028-3908(99)00259-2
127
Chap e 4
Ex ended Discussion
Le us ake a couple o examples om physical cul u e and
spo s. Downhill skiing and slalom make high demands on a
skie 's dex e i y. Wha is he die ence be ween slalom and
plain c oss-coun y skiing, which does no equi e any pa icula
dex e i y? I is in a conglome a e o unexpec ed, unique
complica ions in he ex e nal si ua ion, in a quick succession o
mo o asks ha a e all unlike each o he .
Nicholai A. Be ns ein
,
Dex e i y and i s De elopmen
(1996)
129
4.1 Chap e Summa y
We p opose a new ole o mo o co ex: ex ending he obus ness o sub-
co ical mo emen sys ems, specically o unexpec ed si ua ions demand-
ing apid mo o esponses adap ed o en i onmen al con ex . The implic-
a ions o his idea o cu en and u u e esea ch a e discussed.
130
4.2 A challenge om obo ics
Is  obus con ol a p oblem wo hy o high le el co ical inpu ? Reco -
e ing om a pe u ba ion, o main ain balance o minimize he impac
o a all, is a ole no mally assigned o ou lowe le el pos u al con ol
sys ems. The co ec i e esponses embedded in ou spinal co d (She ing-
on, 1893, 1910), b ains em (A shian e al., 2014) and midb ain (G illne
& Shik, 1973) a e clea ly impo an componen s o his s abilizing ne -
wo k, bu a e hey sucien o main ain obus mo emen in he dynamic
en i onmen s ha we encoun e on a daily basis? Some insigh in o he
equi emen s o a obus con ol sys em can be gained om enginee ing
a emp s o build obo s ha na iga e in na u al en i onmen s.
In he eld o obo ics, ea s o p ecision and ne mo emen con ol
( he mos commonly p esc ibed ole o mo o co ex), a e no a majo
sou ce o dicul y. Indus ial obo s ha e long since exceeded human pe -
o mance in bo h accu acy and execu ion speed (Senoo e al., 2009). Mo e
ecen ly, using ein o cemen lea ning me hods, hey a e now able o au o-
ma ically lea n ecien mo emen s a egies, gi en a human-dened goal
and many epea ed ials o ne- uning (Coa es, Abbeel & Ng, 2008).
Wha hen a e he ha d p oblems in obo ic mo o con ol? Why a e mos
obo s s ill conned o ac o ies, i.e. con olled, p edic able en i onmen s?
The eason is ha as soon as a obo encoun e s na u al e ain, a as
numbe o p e iously unknown si ua ions a ise. The esul ing pe u b-
a ions a e deal wi h poo ly by he s a is ical machine lea ning models
ha a e cu en ly used o ain obo s in con olled se ings.
Le 's conside a amilia example: You a e up ea ly on a Sunday mo n-
ing and head ou side o collec he newspape . I is cold ou , so you pu
on a obe and some slippe s, open he on doo , and descend he s eps
leading down o he s ee in on o you house. Unbeknowns o you, a
hin laye o ice has o med o e nigh and you oo is now quickly sliding
131

ou om unde nea h you. You a e abou o all. Wha do you do? Well,
his depends. Is he e a ailing you can g ab o ca ch you sel ? We e
you ca ying a cup o coee? Did you no ice he os on he lawn and
s ep cau iously, an icipa ing a slippe y su ace? A oiding a dange ous all,
o eco e ing g ace ully, equi es a ich knowledge o he wo ld, knowledge
ha is no immedia ely a ailable o spinal o e en b ains em ci cui s. This
ich con ex ele an o obus mo emen is eadily a ailable in co ex, and
co ex alone.
Imagine now ha you a e asked wi h building a obo o collec you
mo ning newspape . This obo , in o de o a oid a ca as ophic and
cos ly ailu e, would need o ha e all o his con ex ual knowledge as well.
I would need o know abou he s uc u e o he local en i onmen (e.g.
hand ailings ha can suppo i s weigh ), ho liquids and hei iscosi ies,
and e en he co ela ion o ozen dew wi h icy su aces. To be a uly
obus mo emen machine, a obo mus
unde s and
he physical s uc u e
o he wo ld. Mo eo e , i needs o unde s and i in abou
100 ms
a e i s
 oo  s a s o slip.
Reaching o s op a all while holding a cup o coee is no exac ly he
kind o ea o which we p aise ou a hle es and spo s champions, and
his migh explain why he dicul y o such  ea s o obus ness is o en
o e looked. Howe e , i would no be he  s ime ha we nd ou sel es
humbled by he daun ing complexi y o a p oblem ha we nai ely assumed
was  i ial. Vision, o example, has emained an imp essi ely ha d ask
o a machine o sol e a human-le el pe o mance, ye i was o iginally
p oposed as an unde g adua e summe p ojec (Pape , 1966). Pe haps a
simila mises ima e has clouded ou designa ion o he ha d mo o con ol
p oblems wo hy o co ical inpu .
Inspi ed by he challenges con on ing obo icis s, as well as ou oden
beha iou al esul s, we a e now in a posi ion o posi a new ole o mo o
co ex.
132
4.3 A p imo dial ole o mo o co ex
We a e seeking a ole o mo o co ex in non-p ima e mammals, animals
ha do no equi e his s uc u e o o e mo emen p oduc ion. The
s uggles o obo icis s highligh he dicul y o building mo emen sys-
ems ha obus ly adap o unexpec ed pe u ba ions, and he esul s we
epo in his s udy sugges ha his is, indeed, he mos conspicuous
deci o a s lacking mo o co ex. So le us p opose ha , in oden s,
mo o co ex is p ima ily esponsible o ex ending he obus ness o he
subco ical mo emen sys ems. I is no equi ed o con ol in s able,
p edic able, non-pe u bing en i onmen s, bu ins ead specically exe s
i s inuence when unexpec ed challenges a ise. This, we p opose, was
he o iginal selec i e p essu e o e ol ing a mo o co ex, and hus, i s
p imo dial ole. This ole pe sis s in all mammals, media ed ia a modu-
la ion o he subco ical mo o sys em (as is emphasized in s udies o ca
locomo ion), and has e ol ed in p ima es o include di ec con ol o he
skele al muscula u e. Ou p oposal o a  obus  eleology o mo o co ex
has a numbe o in e es ing implica ions.
4.4 Implica ions o non-p ima e mammals
One o he mos imp essi e ai s o mammals is he as ange o en i on-
men al niches ha hey occupy. While mos o he animals adap o change
o e e olu iona y ime scales, mammals excel in hei exibili y, quickly
e alua ing and esponding o unexpec ed si ua ions, and aking isks e en
when aced wi h challenges ha ha e ne e been p e iously encoun e ed
(Spinka, Newbe y & Beko, 2001). This success equi es mo e han p e-
cision, i equi es esou ce ulness: he abili y o quickly come up wi h a
mo o solu ion o any si ua ion and unde any condi ion (Be ns ein, 1996).
The Russian neu ophysiologis Be ns ein e e ed o his abili y wi h an
133
uncon en ional deni ion o dex e i y, which he conside ed o be dis inc
om a simple ha mony and p ecision o mo emen s. In his wo ds, dex e -
i y is equi ed only when he e is `a conglome a e o unexpec ed, unique
complica ions in he ex e nal si ua ions, [such as] in a quick succession o
mo o asks ha a e all unlike each o he ' (Be ns ein, 1996).
I Be ns ein's  obus dex e i y is he p ima y ole o mo o co ex,
hen i becomes clea why he eec s o lesions ha e hus a been so ha d
o cha ac e ize: assays o mo o beha iou ypically e alua e si ua ions
ha a e epea ed o e many ials in a s able en i onmen . Such epea ed
asks we e use ul, as hey oe imp o ed s a is ical powe o quan ica ion
and compa ison. Howe e , we p opose ha hese condi ions specically
exclude he scena ios o which mo o co ex o iginally e ol ed. I is no
easy o epea edly p oduce condi ions ha animals ha e no p e iously
encoun e ed, and he challenges in analysing hese unique si ua ions a e
conside able.
The assay epo ed he e ep esen s ou  s a emp a such an ex-
pe imen , and i has al eady e ealed ha such condi ions may indeed be
necessa y o isola e he ole o mo o co ex in oden s. We hus p opose
ha neu oscience should pu sue simila assays, emphasizing unexpec ed
pe u ba ions and no el challenges, and we ha e de eloped new ha dwa e
and so wa e ools o make hei design and implemen a ion much easie
(Chap e 2).
4.5 Implica ions o p ima e s udies
In con as o o he mammals, p ima es equi e mo o co ex o he di ec
con ol o mo emen . Howe e , do hey also e ain i s ole in gene a ing
obus esponses? The gene al pa esis, o e en pa alysis, ha esul s om
mo o co ical lesions in hese species obscu es he in ol emen o co ex
in di ec ing apid esponses o pe u ba ions. Ye he e is e idence ha a
134
ole in obus con ol is s ill p esen in p ima es, including humans. Fo
example, s oke pa ien s wi h pa ial lesions o he dis ibu ed mo o co -
ical sys em will o en eco e he abili y o mo e he aec ed muscula u e.
Howe e , e en a e eco e ing mo emen , s oke pa ien s a e s ill p one o
se e e impai men s in obus con ol: unsuppo ed alls a e one o he lead-
ing causes o inju y and dea h in pa ien s su i ing mo o co ical s oke
(Jacobs, 2014). We hus sugges ha s oke he apy, cu en ly ocused
on egaining di ec mo emen con ol, should also conside s a egies o
imp o ing obus esponses.
E en i we acknowledge ha a p imo dial ole o mo o co ex is s ill
appa en in p ima e mo emen con ol, i emains o be explained why he
mo o co ex o hese species acqui ed di ec con ol o basic mo emen s
in he  s place. This is an open ques ion.
4.6 Some specula ion on he ole o di ec co ical
con ol
Wha happens when co ex acqui es di ec con ol o mo emen ? Fi s , i
mus lea n how o use his inuence, bypassing o modi ying lowe mo e-
men con olle s. While unc ional co icospinal ac connec ions may be
es ablished p ena ally (Ey e, Mille , Clow y, Conway & Wa s, 2000), he
enemen o co icospinal dependen mo emen s, which mus o e ide he
lowe mo o sys em, akes much longe and coincides wi h he leng hy ma -
u a ion pe iod o co icospinal e mina ion pa e ns (Law ence & Hopkins,
1976). Humans equi e yea s o p ac ice o p oduce and ene basic loco-
mo ion and g asping (Thelen, 1985; on Ho s en, 1989), mo o beha iou s
ha a e a ailable o o he mammals almos immedia ely a e bi h. This
may be he cos o gi ing co ex di ec con ol o mo emen i akes mo e
ime o gu e ou how o mo e he bodybu wha is he bene ?
135