The Economist - USA (2022-02-26)

TheEconomistFebruary26th 2022 Science&technology 75

hasgrown,evenasthenumberoffactory
robotsincreased(seechart).
Thereisa similarfearinhealthcarethat
robotswilldestroyjobs.Butthisisa myth,
Michelle Johnson told the meeting. Dr
JohnsonisthedirectoroftheRehabilita­
tionRoboticsLabattheUniversityofPenn­
sylvania,andcurrentlyworksinBotswana
onwaystouserobotstohelppeoplerecov­
erfromillnessandinjury.EveninAmeri­
ca, let aloneAfrica, “there are just not
enoughclinicianstodothejob,”sheadds.
DrJohnsonhasa particularinterestin
helpingpeoplerecoverfromstrokes.This
sometimesrequiresintensivetherapyfor
longperiods.Butpublichealth­caresys­
temsareoftentoostretchedtoofferany­
thingbutlimited treatment.Robotscan
helphere,andinsomecircumstancesmay
bebetter,even,thanhumanphysiothera­
pists,sincetheyarebothtirelessandreli­
able.Theycan exerciseaperson’slimbs
withconsistentmovementsandtakeob­
jective measurements of recovery.This,
saysDrJohnson,allowsa singleoccupa­
tionaltherapist,assistedbya technician,
tolookafter,atthesametime,half­a­doz­
enor sopatients who would otherwise
needone­to­oneattention.
Robotsthatworkwithpeopleinsuch
waysdo,though,requirespecialtraining.
Andthereisa longwaytogotoimprove
thatsaysJulieShah,wholeadstheInterac­
tiveRoboticsGroupattheMassachusetts
InstituteofTechnology.Mostrobotsper­
formnarrowlydefinedtasks,withmobile
onesusingtheirsensorstoavoidbumping
intopeople.“Robotsneedtoseeusasmore
than just an obstacle to manoeuvre
around,”addsDrShah.“Theyneedtowork
withusandanticipatewhatweneed.”
Studying what happens in factories
showsthatthemostsuccessfulapplica­
tionsemployrobots programmedby an
engineerwhoisworkingside­by­sidewith
someone(aso­called“domainexpert”)ful­
lyversedinthetasksathand.Tomakethat
easier,sheandhercolleaguesaredevelop­
ingaisystemswhichcanschoola robot
usingnatural­languagecommands.

Althoughallthreeexpertsbelievero­

botswillenhancehumancapability,one

problemisthatregulationlagstechnology.

Withcovid­19,saysDrJohnson,somecli­

niciansworriedthateventhespreadofte­

lemedicine mightaffecttheirindemnity

insurance,letalonerobots.Andalthougha

longroadremainsaheadforthedevelop­

ment ofautonomous deliveryvans and

lorries,DrChristensenfindsit“ludicrous”

thatatestvehicledrivingacrossastate

borderinAmericamaythereafterhaveto

complywitha completelydifferentsetof

regulationsfromthosewhichpertainedin

theplacewhenceitcame.Itseemsanaw­

fullotofmeetingslieaheadforroboticists

and regulators to determine how ma­

chinesandpeoplewillworktogether.n

Bots at work

United States

Sources:UniversityofCalifornia,SanDiego;International

FederationofRobotics;BureauofLabourStatistics

50

40

30

20

0

0

19152010

Robot sales,

’000

3.0

2.5

2.0

.5

.0

19152010

Manufacturing

employment, m

Neuroscience

From here to

humanity

S

tudying thehuman brain is hard. Oth­

er  animals’  brains  provide  clues.  But

they cannot reveal the special essence that

makes human brains different. Nor, often,

do  they  make  good  models  for  neurologi­

cal conditions that affect human beings. 

Recently,  however,  two  halfway­house

approaches have been developed. One is to

grow  so­called  brain  organoids  from  hu­

man tissue. The other is to create animals

with  human­derived  neurons  in  their  bo­

dies.  As  the  aaasheard,  both  approaches

are yielding results. But they also raise eth­

ical questions of their own.

Organoids  are  usually  grown  from  in­

duced  pluripotent  stem  cells—artificial

equivalents  of  embryonic  cells.  The  pro­

cess  is  now  sufficiently  well  understood

for them to be mass produced and Paola Ar­

lotta of Harvard University described ways

they  are  being  put  to  use.  These  include

studying  brain  development,  examining

the pathology and genetics of disease, and

screening  potential  drugs.  A  particularly

exciting  idea  is  to  grow  organoids  using

cells from people with known, genetically

related  problems.  That  will  allow  specific

instances of disease to be investigated.

Dr Arlotta herself employs organoids to

study  autistic­spectrum  disorders  (asds),

using  versions  which  incorporate  muta­

tions  of  three  genes  seemingly  linked  to

those  conditions.  Though  these  genes

work in different ways, she found that mu­

tating  any  one  of  them  induces  the  same

effects. These include accelerated develop­

ment of cells called gabaergic neurons and

a consequent slowing of the rate, and dimi­

nution of the amplitude, of electrical spik­

ing. This is of interest because other stud­

ies  suggest  that  disrupted  gabaergic  sig­

nalling is indeed associated with asds.

Dr Arlotta also described how it is now

possible to make organoids that resemble,

in their mix of cells, different parts of the

central  nervous  system  (for  example,  the

cerebral cortex and the spinal cord) and to

link these together, and also to muscle or­

ganoids,  to  create  what  are  known  as  as­

sembloids.  That  permits  preliminary  in­

vestigation of how different regions of the

brain  connect  up,  and  how  the  brain  con­

nects with the rest of the body.

A  thoughtful  individual  might,  at  this

point, be tempted to stop and ask whether

brain  organoids  themselves  can  do  any­

thing  remotely  like  thinking.  At  the  mo­

ment, the answer to that is a pretty defini­

tive  “no”.  Those  currently  emerging  from

the culture tanks are under 5mm across, so

have less than a ten­thousandth of the vol­

ume of an adult human brain. More impor­

tantly,  microscopic  examination  shows

that they have little of the complex organi­

sation found in real brains. And they have

no sensory connections through which to

learn  about  the  world.  But  technology

moves  on.  As  Bernard  Lo,  a  medical  ethi­

cist  at  the  University  of  California,  San

Francisco,  told  the  meeting,  “this  science

is  developing  rapidly,  and  we  don’t  know

what will be possible in a decade.”

The  second  approach,  putting  human

neurons into living animals, was outlined

by  Joshua  Sanes,  who  also  works  at  Har­

vard.  Dr  Sanes  came  to  public  attention  a

few years ago as co­inventor of “Brainbow

mice”—creatures that have had individual

neurons  in  their  brains  “painted”  using

proteins  that  fluoresce  in  different  col­

ours. Recently, though, he has found him­

self  struggling  against  the  limits  of  what

can  be  learned  from  laboratory  animals,

and  has  become  interested  in  the  idea  of

partly “humanising” them.

Transplanting  neurons  is  an  old  tech­

nique, and is even being tested therapeuti­

cally  for  the  treatment  of  Parkinson’s  dis­

ease. But they can also be transplanted be­

tween  species,  and  human  neurons  have,

indeed,  been  transplanted  into  mice.

Some,  though,  talk  of  going  further,  and

transplanting human neural stem cells in­

to embryonic mouse brains. The intention

would  be  to  create  a  “chimera”  in  which

brain  cells  from  both  sources  were  inter­

mixed and interfunctional. 

At  the  moment,  formidable  technical

obstacles  stand  in  the  way  of  doing  this.

But  Dr  Lo’s  observation  about  scientific

progress  is  equally  applicable  here.  And

chimeric animals of this sort, which might

even exhibit humanlike behaviours, are an

idea  at  least  as  disturbing  as  brain  orga­

noids’  becoming  conscious.  In  2020,

therefore,  America’s  scientific  establish­

Organoids and neuron transplants give

new ways to study the brain