37
Experimental design Alexa Morcom Edinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides

Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

  • Upload
    ngophuc

  • View
    222

  • Download
    8

Embed Size (px)

Citation preview

Page 1: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Experimental design

Alexa Morcom

Edinburgh SPM course 2013

Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides

Page 2: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

PreprocessingPreprocessing

Overview of SPM

SPMsSPMs

Image time-

series

Image time-

series

TemplateTemplate

Design matrixDesign matrixContrastsContrasts

ThresholdingThresholding

Generallinearmodel

Generallinearmodel

KernelKernel Variance

components

Variance

components

Page 3: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Overview

• Categorical designs

• Factorial designs

• Parametric designs

• fMRI adaptation

• Control condition

• Paradigm timing

Page 4: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Isolating a process

Subtraction logic and assumption of pure insertion

• Compare task conditions differing in a single process

• Measure the time the process takes

• Assume that addition of the component process

does not alter other task components

Donders (1898-9)

Page 5: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Pure insertion

Page 6: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Cognitive subtraction

Page 7: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Cognitive subtraction

Page 8: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Categorical designs

Principle

• Subtract two conditions to isolate a process

• Assume that addition of the component process

does not alter other task components

• So adding the process into different tasks should

produce the same change in activity

• (and: meaningful cognitive theory)

Page 9: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Categorical designs

Simple subtraction

• Detect regions specialised for a function by

testing for activation difference

Page 10: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Categorical designs

Serial subtraction

• Several cognitive processes in picture naming

See Friston et al. chapter in Human Brain Function (I)

Page 11: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Categorical designs

Experimental design

Page 12: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Categorical designs

Problem:

Page 13: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Factorial designs

Interactions

• The whole (task) is more than the sum of its

(interdependent) processes

• A modulates B

• Vary A and B independently

Page 14: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Factorial designs

Page 15: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Conjunction design

Two task pairs

• B – viewing concrete objects and saying “yes”

• C – naming concrete objects

Difference = phonological retrieval PLUS interaction

with object recognition

• B2 – viewing coloured shapes saying “yes”

• C2 – naming colour of coloured shapes

Diff = phonological retrieval PLUS interaction with

visual analysis

Page 16: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Conjunction design

Overlap isolates the process of interest

• Phonological retrieval

• NOT its interactions with visual processing

Overlap of 4 subtractions

Price & Friston (1997)

Page 17: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Conjunction designs

Detecting overlapping processing

• Encoding faces, different objects

• Reactivation of same regions when face,

object memories retrieved

MVPA recall study

Polyn et al. (2005)

Page 18: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Parametric designs

A continuously varying parameter

• Systematic variation in activity with process

engaged to varying degrees

• Specific: e.g. Linear? Quadratic?

• Avoids pure insertion but does assume no

qualitative change in processing

• Often less sensitive

Page 19: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Parametric designs

PET study

• Auditory words,

varying rate

• Linear relationship

of rate with

activity in primary

auditory cortex

Rest + 5 rates of auditory word presentation

Price et al. (1992)

A B

Page 20: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Parametric designs

Model based fMRI

• Computational model provides neurometric

function e.g. Rescorla-Wagner prediction

error

• Model comparison

Glascher & O’Doherty (2010)

Page 21: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Parametric factorial designs

• V1 activity =

parametric

(physiological)

predictor

• Attention to

motion =

categorical

(psychological)

predictor

z = -9

Friston et al. (1997)

Psychophysiological interaction in V5

Page 22: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

fMRI adaptation

• Repetition suppression

• = reduced BOLD response to repeated stimuli

• Accompanies priming (behavioural)

Page 23: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

fMRI adaptation

Repetition suppression as a tool

• fMRI – typical voxel = 10,000s of neurons

• FFA – a mix, tuned to diff. face orientations?

• Or: all viewpoint-invariant?

Page 24: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

fMRI adaptation

Release from adaptation => sensitivity to the changed feature

Page 25: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

fMRI adaptation

Orientation tuning in human LOC

Ident. Trans. Rotate

Page 26: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Control condition

Problem

• fMRI is a contrastive method – for many

designs, you need a control

• ‘Rest’ isn’t no processing in many areas

Page 27: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Control condition

Different stimuli and task Different stimuli similar task

Same stimuli different task Similar stimuli same task

+

‘Marilyn’ Wonder if I left

the gas on…?

‘Female’ ‘Seen before’

‘Female’ ‘Female’

‘Female’ ‘Female’

� Choice of a baseline depends on your question!

Page 28: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Control condition

Stark & Squire (2001) PNAS

Page 29: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Paradigm timing

Analysis of whole block

Large effects (=efficient)

Optimal length = 16 sec

(sluggish BOLD vs. low

frequency confounds)

Analysis of single items

Smaller effects

SOA from min ~= 2 sec

Page 30: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Paradigm timing

Advantages of event-related design

• Intermixing of conditions avoids unwanted

psychological effects e.g. habituation, expectancy,

loss of concentration

Page 31: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Randomised

O1 N1 O3O2 N2

Data

Model

O = Old Words

N = New WordsBlocked

O

1

O

2

O

3N

1

N

2

N

3

Page 32: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Paradigm timing

Advantages of event-related design

• Intermixing of conditions avoids unwanted

psychological effects e.g. habituation, expectancy,

loss of concentration

• Post-hoc classification of trials, e.g. Subsequent

memory effect

Page 33: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Paradigm timing

Page 34: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Paradigm timing

Advantages of event-related design

• Intermixing of conditions avoids unwanted

psychological effects e.g. habituation, expectancy,

loss of concentration

• Post-hoc classification of trials, e.g. Subsequent

memory effect

• Some events can only be indicated by subject at

particular time e.g. Spontaneous perceptual changes

Page 35: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

0 5 10 15 200

5

10

15

20

Inter−Reversal Time (s)

Num

ber

of P

erce

ptua

l Rev

ersa

ls

0 5 10 15 200

5

10

15

20

25

Inter−Reversal Time (s)

Page 36: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Paradigm timing

Advantages of event-related design

• Intermixing of conditions avoids unwanted

psychological effects e.g. habituation, expectancy,

loss of concentration

• Post-hoc classification of trials, e.g. Subsequent

memory effect

• Some events can only be indicated by subject at

particular time e.g. Spontaneous perceptual changes

• Some events cannot be blocked, e.g. oddball

Page 37: Thanks to Rik Henson, Thomas Wolbers, Jody Culham, · PDF fileEdinburgh SPM course 2013 Thanks to Rik Henson, Thomas Wolbers, Jody Culham, and the SPM authors for slides . PreprocessingPreprocessing

Summary

A few principles, one main take-home message

• Different designs for different questions

Want to know more?

• Temporal design efficiency

• Design optimisation

� Advanced course (Wednesday)