H - PowerPoint PPT Presentation

Symposium “What Neuroimaging Can Tell Us? From Correlation to Causation and Cognitive Ontologies”, OHBM 2016 June 27, 2016 H       ? Sebastian Weichwald MPI for Intelligent Systems sweichwald.de/ohbm2016 � www.neural.engineering

Motivation

Voxel not used Voxel weight for person.n.01 -4 +4 Decoding model OFA OFA FFA FFA sup. sup. RH LH anterior anterior Encoding model EBA EBA Thanks to Alexander Huth, Gallant lab, UC Berkeley 2

Voxel not used Voxel weight for person.n.01 -4 +4 Decoding model OFA OFA √ √ FFA FFA relevance sup. sup. RH LH anterior anterior √ √ Encoding model EBA EBA Thanks to Alexander Huth, Gallant lab, UC Berkeley 2

Voxel not used Voxel weight for person.n.01 -4 +4 Decoding model OFA OFA √ √ FFA FFA relevance sup. sup. RH LH anterior anterior √ √ Encoding model EBA EBA Thanks to Alexander Huth, Gallant lab, UC Berkeley 2

Encoding models 3

Encoding models GLM difference in means ⋯ 3

Encoding models GLM difference in means ⋯ “Associated with experimental condition?” 3

Reichenbach’s Common Cause Principle Reichenbach, 1956 4

Reichenbach’s Common Cause Principle stimulus voxel associated Reichenbach, 1956 4

Reichenbach’s Common Cause Principle stimulus voxel Reichenbach, 1956 4

Reichenbach’s Common Cause Principle h stimulus voxel Reichenbach, 1956 4

Reichenbach’s Common Cause Principle h stimulus voxel stimulus voxel Reichenbach, 1956 4

Reichenbach’s Common Cause Principle h stimulus voxel stimulus voxel stimulus voxel Reichenbach, 1956 4

Reichenbach’s Common Cause Principle h stimulus voxel stimulus voxel stimulus voxel Reichenbach, 1956 4

Causal interpretation chart Weichwald et al., NeuroImage , 2015 5

Causal interpretation chart Voxel relevant? Encoding Causal interpretation √ × Stimulus-based Weichwald et al., NeuroImage , 2015 5

Causal interpretation chart Voxel relevant? Encoding Causal interpretation √ effect of S × Stimulus-based Weichwald et al., NeuroImage , 2015 5

Causal interpretation chart Voxel relevant? Encoding Causal interpretation √ effect of S × no effect of S Stimulus-based Weichwald et al., NeuroImage , 2015 5

Decoding models 6

Decoding models 6

Decoding models “Does removal impair decoding performance?” 6

Decoding models “Does removal impair decoding performance?” “Associated after taking all other variables into account?” 6

Causal interpretation chart Voxel relevant? Encoding Causal interpretation √ effect of S × no effect of S Stimulus-based Weichwald et al., NeuroImage , 2015 7

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ Stimulus-based × Weichwald et al., NeuroImage , 2015 7

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ Non-effects may appear relevant: effect of S × no effect of S stimulus √ voxel A Stimulus-based × voxel B Weichwald et al., NeuroImage , 2015 7

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ Stimulus-based × Weichwald et al., NeuroImage , 2015 7

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ inconclusive Stimulus-based × Weichwald et al., NeuroImage , 2015 7

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ Effects may be missed: inconclusive Stimulus-based × stimulus voxel A voxel B Weichwald et al., NeuroImage , 2015 7

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ inconclusive Stimulus-based × Weichwald et al., NeuroImage , 2015 7

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ inconclusive Stimulus-based × inconclusive Weichwald et al., NeuroImage , 2015 7

Are decoding models useful?

Combine encoding and decoding models & 8

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ inconclusive Stimulus-based × inconclusive Weichwald et al., NeuroImage , 2015 9

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ inconclusive Stimulus-based × inconclusive √ √ √ × √ × × × Weichwald et al., NeuroImage , 2015 9

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ Encoding plus decoding allows to identify indirect effects: inconclusive Stimulus-based × inconclusive √ √ stimulus voxel A voxel B √ × √ × × × Weichwald et al., NeuroImage , 2015 9

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ inconclusive Stimulus-based × inconclusive √ √ √ × √ × × × Weichwald et al., NeuroImage , 2015 9

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ inconclusive Stimulus-based × inconclusive √ √ √ × indirect effect of S √ × × × Weichwald et al., NeuroImage , 2015 9

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ Encoding clarifies role of voxels relevant in decoding: effect of S × no effect of S stimulus √ voxel A inconclusive Stimulus-based × inconclusive √ √ √ × indirect effect of S √ voxel B × × × Weichwald et al., NeuroImage , 2015 9

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ inconclusive Stimulus-based × inconclusive √ √ √ × indirect effect of S √ × × × Weichwald et al., NeuroImage , 2015 9

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ inconclusive Stimulus-based × inconclusive √ √ effect of S √ × indirect effect of S √ × provides context × × Weichwald et al., NeuroImage , 2015 9

Causal interpretation chart Voxel relevant? Encoding Decoding Causal interpretation √ effect of S × no effect of S √ inconclusive Stimulus-based × inconclusive √ √ effect of S √ × indirect effect of S √ × provides context × × no effect of S Weichwald et al., NeuroImage , 2015 9

Wrap-up

Wrap-up ▸ Simple causal interpretation chart 10

Wrap-up ▸ Simple causal interpretation chart (also response-based paradigms) 10

Wrap-up ▸ Simple causal interpretation chart (also response-based paradigms) Relevant in encoding/correlated? × / √ (e. g. GLM) 10

Wrap-up ▸ Simple causal interpretation chart (also response-based paradigms) Relevant in encoding/correlated? × / √ (e. g. GLM) Relevant in decoding/partially correlated? × / √ (e. g. classifier) 10

Wrap-up ▸ Simple causal interpretation chart (also response-based paradigms) Relevant in encoding/correlated? × / √ (e. g. GLM) Relevant in decoding/partially correlated? × / √ (e. g. classifier) ↝ Read off causal interpretation 10

Wrap-up ▸ Simple causal interpretation chart (also response-based paradigms) Relevant in encoding/correlated? × / √ (e. g. GLM) Relevant in decoding/partially correlated? × / √ (e. g. classifier) ↝ Read off causal interpretation (Weichwald et al., NeuroImage , 2015) 10

Wrap-up ▸ Simple causal interpretation chart (also response-based paradigms) Relevant in encoding/correlated? × / √ (e. g. GLM) Relevant in decoding/partially correlated? × / √ (e. g. classifier) ↝ Read off causal interpretation (Weichwald et al., NeuroImage , 2015) Extensions 10

Wrap-up ▸ Simple causal interpretation chart (also response-based paradigms) Relevant in encoding/correlated? × / √ (e. g. GLM) Relevant in decoding/partially correlated? × / √ (e. g. classifier) ↝ Read off causal interpretation (Weichwald et al., NeuroImage , 2015) Extensions ▸ Stimulus-based causal inference 10

Wrap-up ▸ Simple causal interpretation chart (also response-based paradigms) Relevant in encoding/correlated? × / √ (e. g. GLM) Relevant in decoding/partially correlated? × / √ (e. g. classifier) ↝ Read off causal interpretation (Weichwald et al., NeuroImage , 2015) Extensions ▸ Stimulus-based causal inference ↝ Cause-effect relationships between neural processes 10