G. DehaeneG. DehaeneG. DehaeneG. Dehaene----LambertzLambertzLambertzLambertzINSERM/CEA/CNRSINSERM/CEA/CNRSINSERM/CEA/CNRSINSERM/CEA/CNRS

Cognitive Cognitive Cognitive Cognitive NeuroImagingNeuroImagingNeuroImagingNeuroImaging UnitUnitUnitUnit

FranceFranceFranceFrance

A fast structural development

A slow functional development

of the human brain

Questions on connectivity and dynamics

from a developmental perspective

Humans

Chimpanzees

cm3 Brain growth

Humandevelopmentis slow and not comparable to that of other animals

1 5 10 16 years

Synaptic densityBirth Puberty

Acoustic radiations

Cortico-corticalAssociative fibers

from Yakovlev & Lecours, 1967

1 year

Myelogenesis

Visual Cortex

Optic radiations

Gogtay & al, 2004

Cortical thicknessdecreases with age

Humans

Chimpanzees

cm3

Brain growth

Human development is fast

1 5 10 16 years

22 wGA Term 3 months 6 months 1 year

3rd trimester of gestation 2 years31 wGA 2 years

Specific human capacities developduring this period

(verbal communication and social cognition, Metacognition ??)

27w 29w

31w

36w

Preterm newborns (27 – 36w)

Newborns at term and infants

30w 30w 30w 31w

32w 32w 33w

34w 34w 35w

40w 44w 48w

28w 30w

Sulcation development during the last trimester of pregnancy

The third trimester of gestation: Neuronal migration and Sulcation development

J. Dubois (Neurospin) and P. Hüppi (Hôpitaux de Geneve)

Dubois , et al, Cerebral Cortex 2008

The first neurons make the preplate.Then, the migrating neurons will

split the preplate in two: - the marginal zone (layer I)- the subplate (layer VII)

Inside-out placement of the migrating neurons

between the MZ and the SPguided by the Cajal-Retzius cells

(Reelin)in the marginal zone (Layer I)

Marin-Padilla, TINS, 1998

Organization of the Cortical Plate

11 weeks of gestation

Marginal zone

Cortical Plate

Subplate

All pyramidal neurons of the neocortex are anchored to layer I by their original connections and grow by elongating their apical dendrites

creating an horizontal and vertical network

Organization of the Cortical Plate

30 wGA

40 wGA Marin-Padilla, 1982, 1998

1) Progressive organization in 6 layers from the depth to the surface2) Functional connections are first established in the subplate

MZ Marginal zoneCP cortical plateSP SubplateIZ intermediate zone (zone de migration)VZ Ventricular zone

The 6 layers organization and the differentiation between cortical areas is visible after 31 weeks of

gestation

A particular circuitry

� 2 thalamo-cortical networks during the last weeks of gestation: - transient circuits within the subplate- permanent circuit in the cortical plate

� connections with layer 1 ??

� Interneurons are migrating tangentielly and arrive in their location after the pyramidal neurons are in place

CP

35-50mm 50-65mm 65-80mm 80-95mm 95-110mm 110-130mm20-35mm

150-175mm 175-200 mm

Adults (Guevara & al, Neuroimg 2012)

Connectomist, Duclapet Poupon, 2012

Atlas of cortical fibersin full-term newborns (J. Dubois & al, in preparation)

Resting stateNetworks

Doria et al, PNAS, 2010lateral visual

medialvisual

Auditory

Somato-sensory

motor

cerebellum

BrainstemThalami

Default Mode

Left DorsalVisualStream

Executive Control

Right DorsalVisualStream

29-32 SA 33-37 SA 39-43 SA

(n=17) (n=21) (n=24) (n=8)

Full-term

Cortical plate

Subplate

30 wGA

Dark staining of the thalamo-cortical fibers

28 wGA

40 wGA

35 wGA

400 to < 20μv

Resting stateNetworks

Doria et al, PNAS, 2010lateral visual

medialvisual

Auditory

Somato-sensory

motor

cerebellum

BrainstemThalami

Default Mode

Left DorsalVisualStream

Executive Control

Right DorsalVisualStream

29-32 SA 33-37 SA 39-43 SA

(n=17) (n=21) (n=24) (n=8)

Full-term

Cortical plate

Subplate

30 wGA

Dark staining of the thalamo-cortical fibers

•Activity in the subplate in relation with thalamicrelays•Layer I activity•Astrocytes, •Vascular factors related to metabolic demand due to brain construction•etc…

HbO signal: One frame every 110 ms

Standard:the same syllable

is repeated

(ba or ga)

Voice MismatchThe gender of the voice

changed from time to

time

Block design: 20 sec of stimulation followed by 40 s of silence

in 12 preterms at 31 wGA, 3 days after birth

Phoneme MismatchThe identity of the syllable

changed from time to time

(e.g. ba to ga)

NIRS: hemodynamic responses to syllables:

Mahmoudzadeh et al, PNAS, 2013

Permutation testsPcor< .05

ST vs PmmST vs Vmm

0.5

-1

-0.5

0

1

StandardVoice MismatchPhoneme Mismatch

Left Right

Standard

Voice Mismatch

Phoneme Mismatch

0.3

0.6

-0.3

0

CH1

0 2010

0.3

0.6

-0.3

0

CH12

0 2010

0.3

0.6

-0.3

0

CH18

0 2010

0.3

0.6

-0.3

0

CH7

0 2010

Mahmoudzadeh et al, PNAS, 2013

Responseto a change of phoneme

Responseto novelty

e.g. baf baf baf

Male voice

600 ms

t-test

340 ms

t-test: Deviant -Standard

Change of phoneme

gam

Change of voice

baf

L R

L R

L R

L R

ERP to syllables at 30.4 wGA

(19 preterms)

32-64 channels

� The classical resting networks are observed in fMRI from the

first neural connections

� Linguistic and non linguistic information are processed

differently suggesting functional segregation

BUT

the micro neural circuitry is different (layers organisation and

subplate connexions, long-distance and interhemispheric

connexions)

� What are

� the relation between hemodynamique and neural

activity?

� the relation between evoked and spontaneous activity

Post-mortem

Myelin stain

Flechsig, 1920

Full-term birth One month-old 4 month-old

32 sem 40 sem 1 mois 3 mois 6 mois

15 mois 2 ans 4 ans 6 ans

Arborisation dendritique dans le cortex moteur (LeRoy Conel1939-5

Post-term: Myelination of the tracts

Synaptogenesis and pruning

Maturation affects DTI indices:

Transverse diffusivity decreases and FA increases

Neil et al, NMR in Biomed 2002Dubois et al, NeuroImage 2006

McCulloch et al, 1999

Maturation decreases the latency of the visual P1

7 weeks

ERP to face recorded from an occipital electrode

N1

P1

14 weeks

γβα +⋅+⋅= ageindexspeed DTIvP 1

Dubois et al, J Neuroscience 2008

Correlation between anatomical

and functional development

P1

6 weeks

mean

17 weeks

10 weeks

LGN CortexFA

Front

Synaptic densityBirth Puberty

Acoustic radiations

Cortico-corticalAssociative fibers

from Yakovlev & Lecours, 1967

1 year

Myelogenesis

Visual Cortex

Optic radiations

Maturation is not homogeneous across the brain

Frontal Cortex

Auditory Cortex

Birth Puberty1 year 3 years

from Huttenlocher & Dabholkar, 1997

Maturation

+

Maturation index computed on each voxel of a T2 image in a 2-month-oldLeroy & al, J of NS, 2012

Although immature, all regions, notablyfrontal regions, are participating to infant

cognition

Dehaene-Lambertz & al, Brain and Language, 2010

Stranger

Mother

-0.4

-0.2

0

0.2

0.4 Amygdala

1 2 3 4 5 6 7 8-0.8

-0.4

0

0.4

0.8 Orbitofrontal

Prefrontal

Dorsolateral Prefrontal

% signal change % signal change

forward speech

backward speech

Dehaene-Lambertz et al, Science, 2002

AsleepAwake

Three-month-olds

Voices

Photo from Anne Guedes

But the infant brainis slowwwwwwwwww

A systematic organization of phase delays

in response to sentences

Mean phase of the

BOLD response

1 9 s5 73

x=-60

L

x=-44

L

x=-52

L

Dehaene-Lambertz, Dehaene et al. (2006). Functional segregation of cortical language areas by

sentence repetition. Hum Brain Mapp, 27(5), 360-371.

fMRI responses to a sentence are fastest around primary auditory cortex.

They become increasingly slower in STS, temporal pole and temporo-parietal junction,

and slowest in inferior frontal cortex.

also observed in 3-month-old infants

L R

0 7.2 14.4 sTime after sentence onset

Heschl gyrus

Posterior STG

Middle STSAnterior STS

Temporal pole

% s

igna

l cha

nge

0.2

0.2

0

0.2

0

0.2

Heschl gyrus

Posterior STG

Middle STSAnterior STS

Temporal pole

0

0

0

0.2

00.2

0

0.2

0

0.2

0

0.2

0 14.4 s5 10

Mean phase of the BOLD response

Dehaene-Lambertz & al, PNAS, 2006

3-month-old infants

Comparing the phases in infants and adultssame gradient but time-range larger in infants

z= 41z= 17z = 4z = -8

0 14.4 s105

L R L R L R L R

L R L R L R L R

Mean phase of BOLD response

Adults

Infants

16 33 50 66 83 1000

1

2

3

4

P3 amplitude Visibility report

16 33 50 66 83 1000

0.2

0.4

0.6

0.8

1

ERPs to thresholded

visualstimuli in

adultsDel Cul, Baillet & Dehaene, Plos Biology, 2007

� Two types of regions in adults

� Regions in line with the stimulus

� Regions supporting second-order processes (attention, memory,

consciousness)

Visibilityrating

Critical stimulus(17-300 ms)

Backward mask 1 (33 ms)

Backward mask 2(1500 ms minus

stimulus duration)

Forward mask(1500 ms)

or

Face trial Control trial

Repeat 12-14 cycles

Face

ControlP400

Late Slow Wave

µv

Kouider & al, Science, 2013

In infants: Response to masked faces

30 5-month-olds

29 12-month-olds

21 15-month-olds

300 ms

50 ms

100 ms

150 ms

200 ms

250 ms

33 ms

17 ms

12 - 15 months

µv

Time (sec)

Late Slow Wave

P400

N290

Early Posterior Negativity

12-15 month-olds

5 month-olds

Non-linear effect on the diff (Face-Control)

800-1000 ms

A) A simple rule: AAAI AAAI AAAI AAAA

4

-4

μv

Response to a rule violation (900-1200 ms)vs a P300 in adults

Violation - CorrectViolation Correct

Violation

Correct-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2

0

10

20

30

40

Time (s)-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2

0

10

20

Am

plitu

de (p

A.m

)

Time (s)

MMR LSW Basirat et al, submitted

150 ms 350 ms 1000 ms1890 ms

12 3 4

1 2 3 4Rd nb of barepetition

Ga

B) (very) Late response to a predicted deviant stimulus

babies with the same macro architecture and

similar functions

� But dynamics are different due to the

heterogenous maturation of the brain

� Relatively fast maturation of the sensory systems

� Slow (very slow) second order processes, which can be

3 or 4 times slower than in adults

� How these properties can explain

sucessful learning ?

� Synchronisation between

regions with different speed

� Stability of the coupling when

long delays

� Hierarchy between regions

� Etc..

Frequency of occurrence of activity patterns in V1 under Spontaneous activity (S, y axis) versus movie (M, x axis)

Young Ferret P29 Ad Ferret P129

Berkes et alScience 2011

EEG in infants ?

-Hensch, Neuron 2013: Opening of critical windowrelated to inhibition of the endogeneous activity

- Berkes et al, Science 2011: progressive adaptation of internal models to the statistics of natural stimuli at the neural level

� Relation between spontaneous and evoked activity ?

- Infant in the matrix ?

Some questions

to conclude

� Can analyses of the spontaneous activity give indications

about the micro-structural organization of the brain (e.g.

interneurons activity, layer I involvement, thalamic relays) ?

� Can we reconcile the differences between hemodynamic and

neural recordings, notably during preterm life?

� What computational properties to understand the world and

create cultural artefacts are given by this type of organization?

What does mean the relative fast maturation of low-level regions

and the slow maturation of high-level networks?

Are the slow responses crucial for infants’ cognition or just a by-

product of maturation?

Collaborators:

Jessica Dubois

François Leroy

Lucie Hertz-Pannier

Stanislas Dehaene

Jean François Mangin

NIRS (Amiens)

Fabrice Wallois

Madhi Mahmouzadeh

Funding agencies:

� INSERM-CNRS-CEA

� McDonnell Foundation

� Agence Nationale pour

la Recherche

� Fondation Motrice

� Fondation de France