Brain research has made tremendous progress over the last few decades in nearly all areas of investigation and yet, the comprehension of cognition still eludes us because most high-level functions, such as decision making, results from the complex and dynamic interaction between several structures. On the one hand we have a fragmented collection of computational models whose unification is out of reach, on the other hand we have holistic approaches whose complexity obliterates any hope of comprehension. Informed by neuroscience and guided by philosophy, I propose to build the foundations of a research program in computational neuroscience in order to explain how cognition develop in most vertebrates, while guaranteeing its intelligibility.

1. 30
The Art of
Braincrafting
W h a t I c a n n o t c r e a t e , I d o n ' t u n d e r s t a n d
R i c h a r d F e y n m a n , 1 9 8 8
N I C O L A S R O U G I E R
E N C O D S 2 0 2 1
V I R T U A L C O N F E R E N C E
N E U R O D E G E N E R AT I V E D I S E A S E I N S T I T U T E — B O R D E A U X
1

2. 30 *Jorge Luis Borges. “Del rigor en la ciencia.” In: Los Anales de Buenos Aires 1.3 (1946)
In that Empire, the Art of Cartography attained such Perfection that the map of a single
Province occupied the entirety of a City, and the map of the Empire, the entirety of a
Province. In time, those Unconscionable Maps no longer satis
fi
ed, and the Cartographers
Guilds struck a Map of the Empire whose size was that of the Empire, and which coincided
point for point with it. The following Generations, who were not so fond of the Study of
Cartography as their Forebears had been, saw that that vast map was Useless, and not
without some Pitilessness was it, that they delivered it up to the Inclemencies of Sun and
Winters. In the Deserts of the West, still today, there are Tattered Ruins of that Map,
inhabited by Animals and Beggars; in all the Land there is no other Relic of the Disciplines
of Geography.
Suárez Miranda, Travels of Prudent Men, Book Four, Ch. XLV, Lérida, 1658*
‟
2

3. 30
How to model a brain
tame the complexity?
&
3

4. 30
Hebb
Hubel & Wiesel
O’Keefe
Pitts
McCulloch
Pavlov
Eccles
Pen
fi
eld
Hodgking
Huxley Skinner
Golgi & Ramón y Cajal (1906), Neuron doctrine
Pavlov (1927), Classical conditioning
McCulloch & Pitts (1943), Formal Neuron
Tolman (1948), Cognitive maps
Hebb (1949), The organization of Behavior
Pen
fi
eld & Rasmussen (1950), Sensory and motor homonculus
Skinner (1951), Operant conditioning
Eccles, Hodgkin & Huxley (1952), Conductance based model
Hubel & Wiesel (1959), Tuning curves
O’Keefe (1971), Place cells
Marr (1982), Three levels analysis
Varela, Thompson & Rosch (1991), The embodied mind
Rizzolati (1992), Mirror neurons
Churchland & Sejnowski (1992), The computational brain
Schulz (1998), Predictive reward signal
and many more…
A century in Neuroscience
4

5. 30
Are the basal ganglia actually controlling movement or quite the opposite?1 If we consider the
functional connectivity in the motor loop of human basal ganglia2 and the basal ganglia role
in learning rewarded actions and executing previously learned choices3 (and more speci
fi
cally,
the activation of cerebellum and basal ganglia during the observation and execution of
manipulative actions4), we might conclude — considering the neurophysiology of the
pedunculopontine tegmental nucleus5 — that the organization of the basal ganglia functional
connectivity network is non-linear in Parkinson's disease6, hence shedding light on
dyskinesias7.
1 Mov Disord. 2016 31(9) DOI: 10.1002/mds.26680 2 Brain Imaging Behav. 2017 11(2) DOI: 10.1007/s11682-016-9512-y
3 PLoS One. 2020 15(2) DOI: 10.1371/journal.pone.0228081 4 Sci Rep. 2020 10(1) DOI: 10.1038/s41598-020-68928-w
5 Neurobiol Dis. 2019 128 DOI: 10.1016/j.nbd.2018.03.004 6 Neuroimage Clin. 2019 22 DOI: 10.1016/j.nicl.2019.101708
7 Eur J Neurosci. 2021 53(7) DOI: 10.1111/ejn.14777
How do we assemble knowledge
to build a systemic model
without (too much) cherry-picking?
5

6. 30
Trying to understand vision by studying only
neurons is like trying to understand bird
fl
ight
by studying only feathers: it just cannot be done.
David Marr, 1982
‟
6

7. 30
Could a neuroscientist
understand a microprocessor?
Jonas & Kording, PLoS Comp. Bio., 2017
Neuroscience is held back by the fact that it is hard to evaluate if a
conclusion is correct; the complexity of the systems under study
and their experimental inaccessability make the assessment of
algorithmic and data analytic techniques challenging at best. We
thus argue for testing approaches using known artifacts, where the
correct interpretation is known. Here we present a microprocessor
platform as one such test case. We
fi
nd that many approaches in
neuroscience, when used naïvely, fall short of producing a
meaningful understanding.
‟
7

8. 30
Neuroscience needs behavior:
correcting a reductionist bias
Krakauer et al., Neuron, 2017
A
C
D
B E
The Multiple Potential Mappings between Neural
Activity Patterns and Natural Behaviors
(A) Of all the possible activity patterns of a brain
in a dish (big pale blue circle), only a subset of
these (medium dark blue circle) will be relevant
in behaving animals in their natural environment
(big magenta circle).
(B) Designing behavioral tasks that are
ecologically valid (small magenta circle) ensures
discovery of neural circuits relevant to the
naturalistic behavior (small blue circle). Tasks
that elicit species-typical behaviors with species-
typical signals are examples (see Box 1).
(C) In order to study animal behavior in the lab,
the task studied (small white circle) might be so
non-ecological it elicits neural responses (small
blue circle) that are never used in natural
behaviors.
(D) Multiple Realizability: different patterns of
activity or circuit con
fi
gurations (small blue
circles) can lead to the same behavior (small
magenta circle).
(E) The same neural activity pattern (small blue
circle) can be used in two different behaviors
(two magenta circles). The circle with dashed
perimeter in (B)–(E) is the subset of all possible
neural activity patterns that map onto natural
behaviors (from A).
Natural behaviors
Neural activity patterns
Subset of possible
activity patterns
Smaller subset of
activity patterns
All natural behaviors
Subset of
natural behaviors
Unnatural
activity patterns
Behavior outside
natural repertoire
Single natural
behavior
Multiple natural
behaviors
Single
pattern activity
Multiple possible
patterns of activity
8

9. 30
A TREATISE OF HUMAN NATURE (1739)
David Hume (1711-1776)
THE LOGIC OF SCIENTIFIC DISCOVERY (1930)
Karl Popper (1902-1994)
THE STRUCTURE OF SCIENTIFIC REVOLUTIONS (1962)
Thomas Kuhn (1922-1996)
THE METHODOLOGY OF SCIENTIFIC RESEARCH PROGRAMS (1970)
Imre Lakatos (1922-1974)
AGAINST SCIENCE (1975)
Paul Feyerabend (1924-1994)
CAN THEORIES BE REFUTED? (1976)
Essays on the Duhem-Quine Thesis (edited by Sandra G. Harding)
9

10. 30
1. Explain (very distinct from predict)
2. Guide data collection
3. Illuminate core dynamics
4. Suggest dynamical analogies
5. Discover new questions
6. Promote a scienti
fi
c habit of mind
7. Bound (bracket) outcomes to plausible ranges
8. Illuminate core uncertainties.
9. Offer crisis options in near-real time
10. Demonstrate tradeoffs / suggest ef
fi
ciencies
11. Challenge the robustness of prevailing theory through perturbations
12. Expose prevailing wisdom as incompatible with available data
13. Train practitioners
14. Discipline the policy dialogue
15. Educate the general public
16. Reveal the apparently simple (complex) to be complex (simple)
Why Model?
Sixteen reasons other than prediction
Epstein, Journal of Arti
fi
cial Societies and Social Simulation 2008
10

11. 30
1. Explain (very distinct from predict)
2. Guide data collection
3. Illuminate core dynamics
4. Suggest dynamical analogies
5. Discover new questions
6. Promote a scienti
fi
c habit of mind
7. Bound (bracket) outcomes to plausible ranges
8. Illuminate core uncertainties.
9. Offer crisis options in near-real time
10. Demonstrate tradeoffs / suggest ef
fi
ciencies
11. Challenge the robustness of prevailing theory through perturbations
12. Expose prevailing wisdom as incompatible with available data
13. Train practitioners
14. Discipline the policy dialogue
15. Educate the general public
16. Reveal the apparently simple (complex) to be complex (simple)
Why Model?
Sixteen reasons other than prediction
Epstein, Journal of Arti
fi
cial Societies and Social Simulation 2008
A model is
fi
rst and foremost a tool
that is used to answer a question. If
there is no question, there's no need
for a model. The effectiveness of the
model is measured relatively to the
extent it allows to answer the initial
question.
11

12. 30
Maximal brain complexity
Maximal model complexity
Minimal brain complexity
Maximal model complexity
Human Brain Project
Adult human brain
Brain complexity
10
2
10
3
10
4
10
5
10
6
10
7
10
8
10
9
10
10
10
11
Human
Macaque
Cat
Octopus
Rat
Mouse
Frog
Spider
Ant
Leech
C. Elegans
Model
complexity
Open Worm
C.Elegans
High brain complexity
Maximal model complexity
Blue Brain Project
Cat Brain
Hodgkin & Huxley
Realistic Neuron
McCulloch & Pitts
Formal Neuron
Medium brain complexity
Medium model complexity
Spaun
2.5M LIF neurons
Systemic models
& Unexplored territories
Predictive
Explicative
12

13. 30
2 3 4
✔
✘ ✘ ✘
✔
✔
✔
✔
1
Minimum Viable Product
Release early, release often,
and listen to your customers.
13

14. 30
The Cathedral and the Bazaar
Musings on Linux and Open Source by an Accidental Revolutionary
Raymond, 1999
How it started (1991)
How it is going (2017)
~ 10k lines of code (1991)
~ 29M lines of code (2021)
14

15. 30
Computational Neuroscience
mostly produces (a lot of) tires
we may very well have a thousand models of V1, none of them can see.
V1 Basal ganglia
Hippocampus
15

16. 30
The opposite of 'open' isn't closed
The opposite of open is broken.
John Wilbanks, 2012
Don't be this guy
(even if he's cute)
Much more fun
when we share code & data
16

17. 30
Minimal brain complexity
Maximal model complexity
Human Brain Project
Adult human brain
Brain complexity
10
2
10
3
10
4
10
5
10
6
10
7
10
8
10
9
10
10
10
11
Human
Macaque
Cat
Octopus
Rat
Mouse
Frog
Spider
Ant
Leech
C. Elegans
Model
complexity
Open Worm
C.Elegans
High brain complexity
Maximal model complexity
Blue Brain Project
Cat Brain
Hodgkin & Huxley
Realistic Neuron
McCulloch & Pitts
Formal Neuron
Medium brain complexity
Medium model complexity
Spaun
2.5M LIF neurons
Maximal brain complexity
Maximal model complexity
Increasing brain complexity
Increasing model complexity
17

18. 30
Bootstrapping
Cognition
o n e s p e c i e s a t a t i m e
18

19. 30
Topalidou, M., Kase, D., Boraud, T., Rougier, N.P., A
Computational Model of Dual Competition between
the Basal Ganglia and the Cortex, eNeuro, 2018.
Leblois, A., Boraud, T., Meissner, W., Bergman, H.,
Hansel, D., Competition between feedback loops
underlies normal and pathological dynamics in the
basal ganglia. The Journal of Neuroscience, 2006.
Guthrie, M., Leblois, A., Garenne, A., Boraud, T.,
Interaction between cognitive and motor cortico-
basal ganglia loops during decision making: a
computational study. Journal of Neurophysiology,
2013.
Topalidou, M., Leblois, A., Boraud, T., Rougier, N.P., A
long journey into reproducible computational
neuroscience, Frontiers in Computational
Neuroscience, 2015.
Piron, C., Kase, D., Topalidou, M., Goillandeau, M., Rougier,
N. P., Boraud, T., The globus pallidus pars interna in
goal-oriented and routine behaviors: Resolving a long-
standing paradox, Movement Disorders, 2016.
Boraud, T., Leblois, A., Rougier, N.P., A Natural
History of Skills, Progress in Neurobiology, 2018.
Saline or muscimol injection
into the internal part of
the Globus Pallidus (GPi)
15 minutes before session
Cue
presentation
(1.0
- 1.5
second)
Trial Start
(0.5
- 1.5
second)
Decision
(1.0
- 1.5
second)
Go
Signal
Reward
Up
Down
Left
Right
Reward (water) delivered
according to the reward
probability associated
with the chosen stimulus
C
o
n
t
r
o
l
2006
2013
2015
2016
2018
2018
STN1 STN2
GPi1 GPi2
S
Unit 1
STR2
STR1
THL1 THL2
SNc
Unit 2
Reward
Basal
Ganglia
Thalamus
action 1 action 2
STN1 STN2
GPi1 GPi2
S
Unit 1
STR2
STR1
C1 C2
IN2
IN1
SNc / VTA
Unit 2
Reward
Basal
Ganglia
Frontal Cortex
Thalamus
THL1 THL2
action 1 action 2
Inhibitory
Excitatory
Modulatory
Hebbian learning
Reinforcement learning
A B
Plastic
w ∝η, N1, N2
N2
N1
Dopamine (D)
w ∝η, D, N1, N2
N2
N1
Striatum
Cognitive 4 units
GPi
Cognitive 4 units
Thalamus
Cognitive 4 units
STN
Cognitive 4 units
Cognitive loop
Cortex
Cognitive 4 units
Striatum
Motor 4 units
GPi
Motor 4 units
Thalamus
Motor 4 units
STN
Motor 4 units
Motor loop
Cortex
Motor 4 units
Associative loop
Task
Environment
Cue
Positions
Cue
Identities
Substantia
nigra pars
compacta
COMPETITION
COMPETITION
dopamine
reward
Lesion
sites
RL
HL
Cortex
Associative
4x4 units
Striatum
Associative
4x4 units
EXT EXT EXT EXT
COMPETITION
A long journey
in computational neuroscience
This led to the creation of
the ReScience C Journal
(and Rescience X in 2021)
19

20. 30
A Computational Model of Dual Competition
between the Basal Ganglia and the Cortex
Topalidou et al., eNeuro, 2018
Bilateral inactivation of the globus pallidus interna, by
injection of muscimol, prevents animals from learning
new contingencies while performance remains intact,
although slower for the familiar stimuli.
Saline or muscimol injection
into the internal part of
the Globus Pallidus (GPi)
15 minutes before session
Cue
presentation
(1.0
- 1.5
second)
Trial Start
(0.5
- 1.5
second)
Decision
(1.0
- 1.5
second)
Go
Signal
Reward
U
p
D
o
w
n
L
e
f
t
R
i
g
h
t
Reward (juice) delivered
according to the reward
probability associated
with the chosen stimulus
Control
P=0.75
P=0.25 Mean of last 25 trials
1.0
0.8
0.6
0.4
0.2
0.0
0 20 40 60 80 100 120
Number of trials
Mean
success
rate
HC NC
saline
muscimol
Saline Muscimol
1.0
0.8
0.6
0.4
0.2
0.0
Mean
success
rate
*
*
*
*
HC NC HC NC
20

21. 30
CN
+
+
+
+
+
–
–
P
Brain stem structures
(e.g., superior colliculus, PPN)
pr
pc
VTA
–
–
+
Prefrontal
cortex
Premotor cortex
Motor
cortex
Parieto-
temporo-
occipital
cortex
Hippocampus
Thalamus
Globus
pallidus
Striatum
Cerebellum
Amygdala
Subthalamic
nucleus
Ventral
pallidum
Nucleus
accumbens
Substantia
nigra
C
l
a
u
s
t
r
u
m
External current External current
2
1
Thalamus
cognitive
Thalamus
motor
STN
cognitive
GPi
motor
GPi
cognitive
Striatum
cognitive
Striatum
associative
Cortex
motor
Cortex
cognitive
- - -
- -
-
+ +
GPe
cognitive
-
-
+
External current
INDIRECT
PATHWAY
-
-
HYPERDIRECT
PATHWAY
-
-
Striatum
motor
DIRECT
PATHWAY
GPe
motor
STN
motor
Cortex
associative
1
A Computational Model of Dual Competition
between the Basal Ganglia and the Cortex
Topalidou et al., eNeuro, 2018
21

22. 30
The blueprint of the
decision-making network in vertebrates
Saline or muscimol injection
into the internal part of
the Globus Pallidus (GPi)
15 minutes before session
Cue
presentation
(1.0
- 1.5
second)
Trial Start
(0.5
- 1.5
second)
Decision
(1.0
- 1.5
second)
Go
Signal
Reward
U
p
D
o
w
n
L
e
f
t
R
i
g
h
t
Reward (juice) delivered
according to the reward
probability associated
with the chosen stimulus
Control
P=0.75
P=0.25
22

23. 30
The blueprint of the
decision-making network in vertebrates
Saline or muscimol injection
into the internal part of
the Globus Pallidus (GPi)
15 minutes before session
Cue
presentation
(1.0
- 1.5
second)
Trial Start
(0.5
- 1.5
second)
Decision
(1.0
- 1.5
second)
Go
Signal
Reward
U
p
D
o
w
n
L
e
f
t
R
i
g
h
t
Reward (juice) delivered
according to the reward
probability associated
with the chosen stimulus
Control
P=0.75
P=0.25
STN1 STN2
GPi1 GPi2
S
Unit 1
STR2
STR1
THL1 THL2
SNc
Unit 2
Reward
Basal
Ganglia
Thalamus
action 1 action 2
STN1 STN2
GPi1 GPi2
S
Unit 1
STR2
STR1
C1 C2
IN2
IN1
SNc / VTA
Unit 2
Reward
Basal
Ganglia
Frontal Cortex
Thalamus
THL1 THL2
action 1 action 2
Inhibitory
Excitatory
Modulatory
Hebbian learning
Reinforcement learning
A B
Plastic
w ∝η, N1, N2
N2
N1
Dopamine (D)
w ∝η, D, N1, N2
N2
N1
STN1 STN2
GPi1 GPi2
S
Unit 1
STR2
STR1
THL1 THL2
SNc
Unit 2
Reward
Basal
Ganglia
Thalamus
action 1 action 2
STN1 STN2
GPi1 GPi2
S
Unit 1
STR2
STR1
C1 C2
IN2
IN1
SNc / VTA
Unit 2
Reward
Basal
Ganglia
Frontal Cortex
Thalamus
THL1 THL2
action 1 action 2
Inhibitory
Excitatory
Modulatory
Hebbian learning
Reinforcement learning
A B
Plastic
w ∝η, N1, N2
N2
N1
Dopamine (D)
w ∝η, D, N1, N2
N2
N1
23

24. 30
Amphibians
Reptiles Birds
Gallinacean
Mammals
Sauropsids
Humans
Corvids
Iguana
Newt
Zebrafish
Lamprey
Fishes
Rodents
Dorsal Pallium / Cortex
Basal Nuclei
Behavior Automatization
Lamprey
Zebrafish
Fishes Amphibians Sauropsids
Reptiles Birds
Iguana
Gallinacean
Mammals
A Natural History of Skills
(Boraud, Leblois & Rougier 2018)
24

25. 30
Human
Bonobo
Gorilla gorilla
Gorilla beringei
graueri
Gibon
Orangutan
Chimpanzee
Indochinese
lutung
King colobus
Hanuman langur
Moustached
guenon
Green monkey
Wooly monkey
Grey-cheeked
mangabey
Rhesus
monkey
Hamadryas
baboon
Soofy mangabey
Black-and-white
ruffed lemur
Crab-eating
macaque
Mongoose
lemur
Aye aye
Ring-tailed
lemur
Black spider monkey
Tufted capucin
White faced
sapajou
Douroucouli Cotton-top
tamarin
Black-penciled
marmoset
Squirrel
monkey
Red slender
loris
Coquerel's
mouse lemur
Demidoff's
galago
Red-tailed
sportive lemur
Grey mouse
lemur
Imaging evolution
of the primate brain
Friedrich et al., Neuroimage, 2021
25

26. 30
Measuring evolution
of the primate brain
beyond optimality
In economic decision, cognitive biases are deviations
from rationality de
fi
ned as the maximization of
expected utility. In terms of attitude toward risks and
perception of probabilities, prospect theory
(Kahneman & Tversky, 1979) has captured consistent
deviations from standard predictions.
Predation pressure
Social
pressure
Outcome
Value
Gains
Losses
Risk seeking
Risk aversion
Risk aversion
Risk seeking
A person is risk averse
for gains
A person is risk seeking
for losses
26

27. 30
Minimal brain complexity
Maximal model complexity
Human Brain Project
Adult human brain
Brain complexity
10
2
10
3
10
4
10
5
10
6
10
7
10
8
10
9
10
10
10
11
Human
Macaque
Cat
Octopus
Rat
Mouse
Frog
Spider
Ant
Leech
C. Elegans
Model
complexity
Open Worm
C.Elegans
High brain complexity
Maximal model complexity
Blue Brain Project
Cat Brain
Hodgkin & Huxley
Realistic Neuron
McCulloch & Pitts
Formal Neuron
Medium brain complexity
Medium model complexity
Spaun
2.5M LIF neurons
Maximal brain complexity
Maximal model complexity
Phylogenetic axis
Ontogenetic
axis
Increasing brain complexity
Increasing model complexity
27

28. 30
B
A
The Life of behavior
Gomez-Marin & Ghazanfar, Neuron, 2019
Neuroscience needs behavior. However, it is daunting to
render the behavior of organisms intelligible without
suppressing most, if not all, references to life. When
animals are treated as passive stimulus-response,
disembodied and identical machines, the life of behavior
perishes.
‟
28

29. 30
The Art of Braincrafting
With the advent of new practices, new tools and new theories,
the time is ripe for a radical change in our approach and practice
of computational neuroscience. We can envisage a distributed
and cooperative effort of the community towards a uni
fi
ed goal,
that is, understanding how brains work by building them.
29

30. 30
The Art of
Braincrafting
W h a t I c a n n o t c r e a t e , I d o n ' t u n d e r s t a n d
R i c h a r d F e y n m a n , 1 9 8 8
‟
30