TheEconomistFebruary26th 2022 Science&technology 75hasgrown,evenasthenumberoffactory
robotsincreased(seechart).
Thereisasimilarfearinhealthcarethat
robotswilldestroyjobs.Butthisisamyth,
Michelle Johnsontoldthemeeting.Dr
JohnsonisthedirectoroftheRehabilita-
tionRoboticsLabattheUniversityofPenn-
sylvania,andcurrentlyworksinBotswana
onwaystouserobotstohelppeoplerecov-
erfromillnessandinjury.EveninAmeri-
ca,letaloneAfrica,“therearejustnot
enoughclinicianstodothejob,”sheadds.
DrJohnsonhasaparticularinterestin
helpingpeoplerecoverfromstrokes.This
sometimesrequiresintensivetherapyfor
longperiods.Butpublichealth-caresys-
temsareoftentoostretchedtoofferany-
thingbutlimitedtreatment.Robotscan
helphere,andinsomecircumstancesmay
bebetter,even,thanhumanphysiothera-
pists,sincetheyarebothtirelessandreli-
able.Theycanexerciseaperson’slimbs
withconsistentmovementsandtakeob-
jectivemeasurementsofrecovery.This,
saysDrJohnson,allowsasingleoccupa-
tionaltherapist,assistedbyatechnician,
tolookafter,atthesametime,half-a-doz-
enorsopatientswhowouldotherwise
needone-to-oneattention.
Robotsthatworkwithpeopleinsuch
waysdo,though,requirespecialtraining.
Andthereisalongwaytogotoimprove
thatsaysJulieShah,wholeadstheInterac-
tiveRoboticsGroupattheMassachusetts
InstituteofTechnology.Mostrobotsper-
formnarrowlydefinedtasks,withmobile
onesusingtheirsensorstoavoidbumping
intopeople.“Robotsneedtoseeusasmore
than just an obstacle to manoeuvre
around,”addsDrShah.“Theyneedtowork
withusandanticipatewhatweneed.”
Studyingwhat happens in factories
showsthatthemostsuccessfulapplica-
tionsemployrobotsprogrammedbyan
engineerwhoisworkingside-by-sidewith
someone(aso-called“domainexpert”)ful-
lyversedinthetasksathand.Tomakethat
easier,sheandhercolleaguesaredevelop-
ingaisystemswhichcanschoolarobot
usingnatural-languagecommands.
Althoughallthreeexpertsbelievero-
botswillenhancehumancapability,one
problemisthatregulationlagstechnology.
Withcovid-19,saysDrJohnson,somecli-
niciansworriedthateventhespreadofte-
lemedicinemightaffecttheirindemnity
insurance,letalonerobots.Andalthougha
longroadremainsaheadforthedevelop-
mentofautonomousdeliveryvansand
lorries,DrChristensenfindsit“ludicrous”
thatatestvehicledrivingacrossastate
borderinAmericamaythereafterhaveto
complywithacompletelydifferentsetof
regulationsfromthosewhichpertainedin
theplacewhenceitcame.Itseemsanaw-
fullotofmeetingslieaheadforroboticists
and regulators todetermine how ma-
chinesandpeoplewillworktogether.Bots at work
United StatesSources:UniversityofCalifornia,SanDiego;International
FederationofRobotics;BureauofLabourStatistics
4
3
19152010Robot sales,
’
3.
.
.
.
. 19152010Manufacturing
employment, mNeuroscienceFrom 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 different. Nor, often,
do they make good models for neurologi-
cal conditions that affect human beings.
Recently, however, two halfway-house
approaches have been developed. One is to
grow so-called brain organoids from hu-
man tissue. The other is to create animals
with human-derived 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—artificial
equivalents of embryonic cells. The pro-
cess is now sufficiently 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 specific
instances of disease to be investigated.
Dr Arlotta herself employs organoids to
study autistic-spectrum disorders (asds),
using versions which incorporate muta-
tions of three genes seemingly linked to
those conditions. Though these genes
work in different ways, she found that mu-
tating any one of them induces the same
effects. 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, different 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 different 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 defini-
tive “no”. Those currently emerging from
the culture tanks are under 5mm across, so
have less than a ten-thousandth 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 co-inventor of “Brainbow
mice”—creatures that have had individual
neurons in their brains “painted” using
proteins that fluoresce in different 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 scientific
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 scientific establish-Organoids and neuron transplants give
new ways to study the brain