françois vialatte laboratoire plasticité du cerveau
TRANSCRIPT
![Page 1: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/1.jpg)
CAN WE BRIDGE THE GAP BETWEEN SUBJECTIVE AND
OBJECTIVE EVALUATIONS IN NEUROFEEDBACK EXPERIMENTS
François VIALATTE
Laboratoire Plasticité du Cerveau
![Page 2: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/2.jpg)
SUBJECTIVITY IN
NEUROFEEDBACKWhy and which consequences?
![Page 3: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/3.jpg)
Subective variablesS U B J E C T I V I T Y I N N E U R O F E E D B A C K
WHY DOES SUBJECTIVE EVALUATION MATTER
4 horsemen of the apocalypse:
1. Personality traits
2. Cognitive load
3. Task engagement
4. Metacognition
Is there a behavioural or skill improvement?
Was the brain rythm effectively conditionned (within session, and after sessions)?
CLASSICAL OBJECTIVE EVALUATIONS
To understand better how NFB works
Subjective variations = uncontrolled experimental variations (experimental noise)
![Page 4: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/4.jpg)
Personality traitsS U B J E C T I V I T Y I N N E U R O F E E D B A C K
PERSONALITY AND FEEDBACK INTEGRATION
Human dimensions of NFB:
• High anxiety = strong bias towards negative feedbacks
• Locus of control towards technology
• Is there a « NFB responder » personality construct?
Personnality may differentiateresponders from non responders
![Page 5: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/5.jpg)
Cognitive load theoryS U B J E C T I V I T Y I N N E U R O F E E D B A C K
Ref.: Sweller et al., 2010
Direct relationship betweengermane and intrinsic
cognitive loads
Learning is more efficient when abstraction is made easyfor the learner (germane load)
Extraneous cognitive load= the learner has to performabstraction no learning
Know thyself: NFB to promote germane load
![Page 6: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/6.jpg)
Task engagement: motivation?S U B J E C T I V I T Y I N N E U R O F E E D B A C K
Ref.: Csikszentmihalyi, 1988
![Page 7: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/7.jpg)
Metacognition: NFB vs. Cognitive remediation
S U B J E C T I V I T Y I N N E U R O F E E D B A C K
Ref.: Wykes et al., 2011
• Cognitive remediation: maintaining altered cognitive functions, or compensating weak function using reliable ones.
• Reeducation using neuropsychological games, helped by a psychologist who encourages metacognition.
![Page 8: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/8.jpg)
Approriateresult feedback
Redundantresult feedback
Instrinsicfeedback
Result feedbacksS U B J E C T I V I T Y I N N E U R O F E E D B A C K
• LIMITED
• only if not already available as intrinsic feedback
Schmidt 1975
Result in remediation = am I succesfull with my neuropsychological task?
![Page 9: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/9.jpg)
Performance feedbackS U B J E C T I V I T Y I N N E U R O F E E D B A C K
Schmidt 1975
Performance feedback : serve the ball as if using a whip
• Useful to help the player who has to modify his movements
• Only if he can
Performance in remediation = was I focused enough?
relate the feedback to his own intrinsic signals
No externalperformance feedback in remediation
![Page 10: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/10.jpg)
Why is performance critical?S U B J E C T I V I T Y I N N E U R O F E E D B A C K
Functional deficit (performance)Example: schizophrenia Pathology (result)
Pathology = neural correlates of the deficit(or a specific cognitive function)
Performance = mastery over the correlate(modulation of neural correlate)
Attentional
deficit
Positive
symptoms
Functional
consequences
![Page 11: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/11.jpg)
Organism (action)
Cognitive functions
Process
Environment (reaction)
Correction
ResultObjective
Internal result feedback
Internal performance feedback
External result feedback
External performance feedback
Skill learningS U B J E C T I V I T Y I N N E U R O F E E D B A C K
Ref.: Gaume, et al., 2016
![Page 12: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/12.jpg)
Performance feedbackS U B J E C T I V I T Y I N N E U R O F E E D B A C K
Schmidt 1975
Performance feedback : serve the ball as if using a whip
• Useful to help the player who has to modify his movements
• Only if he can
External performance in NFB = are the neural correlates of the symptoms evolving?
relate the feedback to his own intrinsic signals
Externalperformance
feedback possible in NFB
![Page 13: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/13.jpg)
MEASURING
SUBJECTIVITYPsychological scales
![Page 14: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/14.jpg)
Trait scalesM E A S U R I N G S U B J E C T I V I T Y
Scale Description
KUT Locus of control related to technology
Learning Style Inventory Learning style
16-Personality Factor-5 Personnality
STAI-Y Anxiety
KUT not available in French validated translation
![Page 15: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/15.jpg)
M E A S U R I N G S U B J E C T I V I T Y
Scale Passation
Mindful state Before + after
Attention/Working memory Before + after
Commitment/motivation Before + after
Mood Before + after
Cognitive load Before + after
Expectations Before
Too many scales to evaluate regularly the state creation of an NFB state scale
State scales
![Page 16: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/16.jpg)
Before After
Test on 9 subjects, up to 5 sessions < 30 sec to answer in average Can be used for regular
evaluations
![Page 17: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/17.jpg)
Metacognition
M E A S U R I N G S U B J E C T I V I T Y
NFB EXPLICITATION INTERVIEW IN THE LITTERATURE
Strategies may speed up learning, and allow transfer tasks(participants their strategies in daily life situations without NFB).
2 strategies:- self-regulation (breathing, relaxation, etc…)- Modulation (imagination or conditioning)
Ref.: Kober et al., 2013; Nan et al., 2012; Hampson et al., 2011; Caria et al., 2007; Leary et al., 2006
Weak as compared to cognitive remediation creation of a metacognitive interview
strategy to modulateactivity in nucleus accumbens (fMRIneurofeedback)
+ some false assumptions regarding the « absence of strategies »
(confusion with introspection)
![Page 18: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/18.jpg)
• Utiliser des stratégiesd’imagination (imaginer de lalumière, des sensationscorporelles, qui font varier lefeedback)
• Utiliser des conditionnementsexistants (penser à des souvenirs, àdes mots, qui changent le signal defeedback)
• Soutien du feedback par desstratégies psycho-corporelles(changer mon rythme respiratoire,pianoter avec mes doigts, pourinfluer le feedback)
• Régulation du stress (respiration, relaxation)• Soutien à la motivation (imaginer des gens
qui m’applaudissent, me répéter « bravo »intérieurement, me visualiser grand et fort)
• Efforts de concentration (rester immobile etfocalisé, ne penser qu’à la tâche)
• Comparer le feedback et mesperceptions internes (mon ressentiest-il en accord avec le feedback ?)
• Essayer de prédire le feedback avantqu’il ne s’affiche.
• Être attentif (conscient de) à sacharge cognitive
• Utiliser une technique demindfulness (être dans l’instantprésent, être plutôt que faire)
Modulation
Introspection
Auto-régulation
• Répéter l’instruction enboucle à voix haute
• Réencoder l’instructionavec mes propres mots
• Procéder étapes parétapes
• Faire des pauses• Compenser ses faiblesses
Etayages
![Page 19: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/19.jpg)
Test on 9 subjects, up to 5 sessions100% of subjects found thisinterview very useful at the first session AND at subsequent sessions >60% of subjects reported « no strategy » at first (and afterwardsrecognized 2~3 strategies duringinterview)
![Page 20: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/20.jpg)
BRIDGING THE GAPModeling subjective variables
![Page 21: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/21.jpg)
Modeling personality dimensionsB R I D G I N G T H E G A P
Complex dataComputer learns
« by itself »
The fashionable so-called « bayesian » approach isone of many versions of machine learning
![Page 22: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/22.jpg)
• Multi-layer perceptron
• Support vector machines
Machine learningB R I D G I N G T H E G A P
𝑥
𝑦
𝑥
𝑦
Linear regression Non-linera regression
𝑥
𝑦
Univeral approximators
𝑦∗ = 𝑎𝑛𝑥𝑛 +⋯+ 𝑎1𝑥
1 + 𝑎0𝑦∗ = 𝑎𝑥 + 𝑏
![Page 23: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/23.jpg)
Example : age and fMRIB R I D G I N G T H E G A P
Ref.: La Corte, et al., 2016
![Page 24: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/24.jpg)
ERN in feedback tasks : fERNB R I D G I N G T H E G A P
Using fERN to objectivize feedback integration
Source : DLPFC / ACC
Ref.: Walsh & Anderson, 2012
-9,3µV -2,0µV
![Page 25: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/25.jpg)
• Ability to « feel oneself » (state perception)
• Integration of an « interoceptive consciousness »in the insular cortex
• Possible neural correlate: Heart Evoked Potential
Germane load = interoception?B R I D G I N G T H E G A P
Ref. : Damasio et al., 2003, 2015; Barrett & Simmons, 2015; Luft & Bhattacharya , 2015.
![Page 26: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/26.jpg)
APPENDIX
![Page 27: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/27.jpg)
Problème 1 : termes croisésD É M A R C H E D E P S Y C H O I N G É N I E R I E
Le « ou exclusif »x2
x1
x1
x2Exemple type :
Indice de massecorporelle
![Page 28: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/28.jpg)
0 5 10 15 20 25 30 35
NSS
Neurological soft signs
0 5 10 15 20 25 30 35
0
10
20
30
40
50
60
70
80
NSS
SM
NSS + saccades mémorisées
30
20
0
01
10
20
23
4 0
40
60
80
SM
NSS + SM + prosaccades/antisaccades
(a)(b)
(c)
Contrôles
Patients souffrant de schizophrénie
Termes croisés : illustrationD É M A R C H E D E P S Y C H O I N G É N I E R I E
![Page 29: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/29.jpg)
Problème 2 : bruitI NGÉNIER IE DES ICM
LES DONNÉES CONTIENNENT DU BRUIT
- Limites de précision des instruments
- Erreurs de mesures (erreurs humaines, erreurs liées au contexte)
- Bruits technologiques (pollution électromagnétique)
Donnéescollectées
Distributionsous-jacente
Les données contiennent une
part d’information, et une part de bruit
![Page 30: François VIALATTE Laboratoire Plasticité du Cerveau](https://reader036.vdocuments.us/reader036/viewer/2022062420/62ae5248d15c887e834e1538/html5/thumbnails/30.jpg)
Sur-ajustementD É M A R C H E D E P S Y C H O I N G É N I E R I E
Sous-ajustement : le modèle n’a pas appris toute l’information
Sur-ajustement : le modèle a appris « par cœur » les données d’apprentissage (dont le bruit)