experimental design - university of · pdf fileexperimental design butterfly categorical...
TRANSCRIPT
ExperimentalExperimental DesignDesign
Thomas WolbersThomas WolbersSpace and Aging LaboratorySpace and Aging LaboratoryCentre for Cognitive and Neural SystemsCentre for Cognitive and Neural Systems
SPM Course Edinburgh 2010
Design Design of functional neuroimaging studiesof functional neuroimaging studiesCategoricalCategorical designsdesignsFactorialFactorial designsdesignsParametricParametric designsdesignsfMRIfMRI adaptationadaptationControl conditionsControl conditionsTrial timingTrial timing
Overview
SPM Course Edinburgh 2010Categorical Designs
F.C. Donders
Subtraction logicSubtraction logic in in psychophysicspsychophysicsAssumption of pure insertion:Assumption of pure insertion:⇒⇒ you can insert a component process into a task without you can insert a component process into a task without
disrupting the other componentsdisrupting the other components⇒⇒ you can estimate duration of a cognitive process by you can estimate duration of a cognitive process by
comparing reaction times between different conditionscomparing reaction times between different conditions
SPM Course Edinburgh 2010Categorical Designs
DetectStimulus
PressButton
DetectStimulus
PressButton
Discriminate Color
DetectStimulus
PressButton
Discriminate Color
ChooseButton
T3: Choice Reaction Time• Hit left button when light is green and right button when light is red
T1: Simple Reaction Time• Hit button when you see a light
T2: Discrimination Reaction Time• Hit button when light is green but not red
SPM Course Edinburgh 2010Categorical Designs
DetectStimulus
PressButtonT1
DetectStimulus
PressButton
Discriminate ColorT2
-
Discriminate Color
=
SPM Course Edinburgh 2010Categorical Designs
DetectStimulus
PressButton
Discriminate ColorT2
-
=
DetectStimulus
PressButton
Discriminate Color
ChooseButtonT3
ChooseButton
SPM Course Edinburgh 2010
BackgroundBackground⇒⇒ pure insertion: you can insert a component process into a task pure insertion: you can insert a component process into a task
without disrupting the other componentswithout disrupting the other components⇒⇒ addtional cognitiveaddtional cognitive processesprocesses always evoke the same additional always evoke the same additional
activation!activation!⇒⇒ activation due to baselineactivation due to baseline task task unaffected!unaffected!
moving dotsmoving dots fixationfixation
SimpleSimple subtractionsubtraction•• you can identify functionally specialised regions with regionallyou can identify functionally specialised regions with regionally y
specific activation differencesspecific activation differences
Categorical Designs
SPM Course Edinburgh 2010
SerialSerial subtractionsubtractionQuestion:Question: Is inferiorIs inferior temporaltemporal cortexcortex (IT) (IT) involvedinvolved in in phonological retrieval phonological retrieval during object recognition?during object recognition?
CognitiveCognitive processesprocesses
⇒⇒ visualvisual analysis:analysis: occipitaloccipital cortexcortex
⇒⇒ object recognition:object recognition: ??????
⇒⇒ phonologicalphonological retrieval:retrieval: ??????
⇒⇒ verbalverbal outputoutput: : BrocaBroca‘‘s areas area
Categorical Designs
SPM Course Edinburgh 2010
A A say say „„yesyes““ when you see an abstract imagewhen you see an abstract image((vis. vis. analysis,analysis, verbal outputverbal output)) ⇒⇒ yesyes
B B say say „„yesyes““ when you see a concrete when you see a concrete objectobject
((vis. vis. analysis,analysis, object recognition, verbalobject recognition, verbaloutputoutput))
⇒⇒ yesyes
C C namename concreteconcrete objectobject((vis. analysis, object recognition, vis. analysis, object recognition, phonological retrieval, verbal outputphonological retrieval, verbal output))
⇒⇒
ExperimentalExperimental designdesign
butterflybutterfly
Categorical Designs
SPM Course Edinburgh 2010
visual analysisobject recognitionphonological retrievalphonological retrievalverbal output
Problem:Problem: unjustified assumption that IT response to object recognition isunjustified assumption that IT response to object recognition iscontext independentcontext independent!!
psychophysicspsychophysics ≠≠ neurophysiologyneurophysiology
visual analysisverbal output
visual analysisobject recognitionverbal output
-------------------AA
⇒⇒ significantsignificant IT activationIT activation ⇒⇒ object recognition!object recognition!B B -- AA
⇒⇒ nono significant ITsignificant IT activationactivation ⇒⇒ no evidence for IT no evidence for IT involvement in phonological retrieval!involvement in phonological retrieval!
C C -- BB
visual analysisobject recognitionobject recognitionverbal output
visual analysisobject recognitionphonological retrievalphonological retrievalverbal output
BB CC
Categorical Designs
SPM Course Edinburgh 2010
A B C D task
interaction terminteraction term(task B x task D)(task B x task D)
C C -- DD = - =
- =
-
=interactioninteraction
BackgroundBackground•• ““the whole is more than just the sum of its partsthe whole is more than just the sum of its parts””•• cognitive processes are interdependent cognitive processes are interdependent ⇒⇒ task A interacts with task B, task A interacts with task B,
A modulates sensitivity to B ... A modulates sensitivity to B ...
B B -- AA =
Factorial Designs
SPM Course Edinburgh 2010
visual analysisobject recognitionphonological retrievalphonological retrievalverbal output
•• phonological retrieval modulatesphonological retrieval modulates IT response to object recognitionIT response to object recognition⇒⇒ IT IT also involvedalso involved in in phonological retrieval!phonological retrieval!
no object no object recogn.recogn.
D D Name colour of abstract imageName colour of abstract image((vis. vis. analysis,analysis, phonological retrieval,phonological retrieval, verbal verbal outputoutput))
⇒⇒ greengreen
phonolog. phonolog. retrievalretrievalnono phonolog. phonolog. retrievalretrieval
(C (C -- D) D) –– ⇒ significant IT activation(B (B -- A)A)
visual analysisverbal output
visual analysisobject recognitionverbal output
AA
BB CC
visual analysisphonological retrievalphonological retrievalverbal output
DD
object object recognit.recognit.
visual analysisobject recognitionobject recognitionverbal output
Factorial Designs
Interaction:
visual analysisobject recognitionphonological retrievalphonological retrievalverbal output
SPM Course Edinburgh 2010
Example 1Example 1: : linear activation linear activation increase inincrease in LOC LOC with increasing with increasing object visibility!object visibility!
SystematicSystematic variationvariation of regionalof regional activation withactivation with endoendo--/exogenous/exogenous parametersparameters⇒⇒ task stays the same while the amount of processing varies; thus,task stays the same while the amount of processing varies; thus, changes to changes to
the nature of the task are less of a problem the nature of the task are less of a problem ⇒⇒ you can test for bothyou can test for both linearlinear (i.e.level(i.e.level of sensorimotor/cognitive processing)of sensorimotor/cognitive processing) andand
nonnon--linear effectslinear effects (i.e. time effects)(i.e. time effects)
categorical/factorialcategorical/factorial designsdesigns
cognitivecognitive processesprocesses
parametricparametric designsdesigns
binarybinary
continuouscontinuousS
igna
l
object visibilityobject visibility
Parametric Designs
RoseRose et al. et al. (2005).(2005). Cerebral Cerebral CortexCortex..
SPM Course Edinburgh 2010
1 2 3 4 5
Sig
nal ExampleExample 2: 2: NonNon--linearlinear decrease ofdecrease of prefrontal prefrontal
activation over time during procedural learning!activation over time during procedural learning!
Combining parametricCombining parametric andand factorialfactorial designsdesignsnonnon--linear interactionlinear interactionlinear interactionlinear interaction
V1V1
STSSTS
visibilityvisibility
Sig
nal
Sig
nal1-back
2-back
Sig
nal
object visibilityobject visibility RainerRainer et al. (2001). et al. (2001). CurrentCurrentBiologyBiology..
Parametric Designs
SPM Course Edinburgh 2010fMRI Adaptation
HensonHenson et al. (2000)et al. (2000)
FusiformFusiform gyrusgyrus
repetition suppression / fMRI adaptation / primingrepetition suppression / fMRI adaptation / priming
fMRI adaptationfMRI adaptationpriming = reduced BOLD response to primed stimuli (i.e. priming = reduced BOLD response to primed stimuli (i.e. repeated presentation)repeated presentation)
SPM Course Edinburgh 2010fMRI Adaptation
120°90°60° 150°
tuning curvefor this one neuron
Viewing angle
Neu
rona
l act
ivity
monkeymonkey STSSTS
Priming as a toolPriming as a toolincrease in spatial resolution (increase in spatial resolution (hyperresolutionhyperresolution))example: orientation tuning for face stimuliexample: orientation tuning for face stimuli
⇒⇒ viewpoint selectivity in the human FFA?viewpoint selectivity in the human FFA?
SPM Course Edinburgh 2010fMRI Adaptation
Priming as a toolPriming as a toolfMRI fMRI –– voxel typically contain ten thousands of neuronsvoxel typically contain ten thousands of neuronsFFA: mixture of neurons tuned to different orientations?FFA: mixture of neurons tuned to different orientations?
120°90°
60°150°
180°
Viewing angle
Neu
rona
l act
ivity
30°0°
-30°-60°
Tuning curvesfor three different neurons
identical BOLD response to identical BOLD response to each orientation!each orientation!
Slide modified from Jody Culham
SPM Course Edinburgh 2010
‘‘samesame’’stimulistimuli
‘‘differentdifferent’’StimuliStimuli
predicted BOLD-responses
Slide modified from Jody Culham
fMRI Adaptation
orientation tuning in the humanorientation tuning in the human FFA?FFA?
SPM Course Edinburgh 2010fMRI Adaptation
GrillGrill--SpectorSpector & & MalachMalach (2001), Acta (2001), Acta PsycholPsychol..
facesfaces vs. vs. objectsobjects fMRIfMRI adaptationadaptation
orientation tuning in the humanorientation tuning in the human FFA?FFA?
SPM Course Edinburgh 2010Control Condition
ProblemProblem•• fMRI = fMRI = contrastivecontrastive methodmethod
⇒⇒ for many designs, you need to include adequatefor many designs, you need to include adequate control conditionscontrol conditions
„„RestRest““ = = often substantial activation in many areas!often substantial activation in many areas!⇒⇒ reason:reason: mentalmental imageryimagery / rehearsal / / rehearsal / eye movementseye movements……⇒⇒ lossloss ofof sensitivity!sensitivity!
Stark & Squire (2001) Stark & Squire (2001) –– When When zero is not zero... zero is not zero... PNASPNAS, , 98(22), 1276098(22), 12760--12766.12766.
SPM Course Edinburgh 2010Control Condition
⇒⇒ decision depends on experimentaldecision depends on experimental hypotheses!hypotheses!
AlternativesAlternatives1.1. high congruency withhigh congruency with experimental conditionsexperimental conditions2.2. additional possibilities:additional possibilities:
Stark & Squire (2001) Stark & Squire (2001) –– When When zero is not zero... zero is not zero... PNASPNAS, , 98(22), 1276098(22), 12760--12766.12766.
SPM Course Edinburgh 2010
analysis of entire block, not of single analysis of entire block, not of single stimulistimulilarge effectslarge effectsOptimal block length: ~16secOptimal block length: ~16sec
⇒⇒ allows enough time for signal to allows enough time for signal to oscillate fullyoscillate fully
⇒⇒ not near artifact frequenciesnot near artifact frequencies
analysis of single stimulianalysis of single stimulismaller effectssmaller effectsminimal SOA: ~2secminimal SOA: ~2sec
Block DesignBlock Design EventEvent--Related DesignRelated Design
Trial timing
Stimulus trainStimulus train
Hypothetical BOLD responseHypothetical BOLD response
Stimulus trainStimulus train
Hypothetical BOLD responseHypothetical BOLD response
SPM Course Edinburgh 2010Trial timing
Advantages of eventAdvantages of event--related designrelated designrandomised order avoids unwanted psychological effectsrandomised order avoids unwanted psychological effects
e.g. habituation / expectancy effects, attentional e.g. habituation / expectancy effects, attentional declinedecline
SPM Course Edinburgh 2010
RandomisedRandomised
O1 N1 O3O2 N2
DataData
ModelModel
O = Old WordsO = Old WordsN = New WordsN = New Words
BlockedBlocked
O1 O2 O3 N1 N2 N3
Trial timing
Slide modified from Rik Henson
SPM Course Edinburgh 2010Trial timing
Advantages of eventAdvantages of event--related designrelated designrandomised order avoids unwanted psychological effectsrandomised order avoids unwanted psychological effects
e.g. habituation / expectancy effects, attentional e.g. habituation / expectancy effects, attentional declinedecline
postpost--hoc/subjective classification of trials hoc/subjective classification of trials e.g. subsequent memory effecte.g. subsequent memory effect
SPM Course Edinburgh 2010
R R RF F
R = Words Later RememberedR = Words Later RememberedFF = Words Later Forgotten= Words Later Forgotten
~4s
DataModel
EventEvent--relatedrelated
Trial timing
Slide modified from Rik Henson
SPM Course Edinburgh 2010Trial timing
Advantages of eventAdvantages of event--related designrelated designrandomised order avoids unwanted psychological effectsrandomised order avoids unwanted psychological effects
e.g. habituation and expectancy effects, attentional e.g. habituation and expectancy effects, attentional declinedecline
postpost--hoc/subjective classification of trials hoc/subjective classification of trials e.g. subsequent memory effecte.g. subsequent memory effect
some events can only be indicated by subject (in time)some events can only be indicated by subject (in time)e.g. spontaneous perceptual changese.g. spontaneous perceptual changes
SPM Course Edinburgh 2010
0 5 10 15 200
5
10
15
20
Inter−Reversal Time (s)
Num
ber o
f Per
cept
ual R
ever
sals
0 5 10 15 200
5
10
15
20
25
Inter−Reversal Time (s)
Trial timing
Slide modified from Rik Henson
SPM Course Edinburgh 2010Trial timing
Advantages of eventAdvantages of event--related designrelated designrandomised order avoids unwanted psychological effectsrandomised order avoids unwanted psychological effects
e.g. habituation and expectancy effects, attentional e.g. habituation and expectancy effects, attentional declinedecline
postpost--hoc/subjective classification of trials hoc/subjective classification of trials e.g. subsequent memory effecte.g. subsequent memory effect
some events can only be indicated by subject (in time)some events can only be indicated by subject (in time)e.g. spontaneous perceptual changese.g. spontaneous perceptual changes
some trials cannot be blockedsome trials cannot be blockede.g. e.g. „„oddballoddball““ paradigmsparadigms
SPM Course Edinburgh 2010Trial timing
TimeTime…
““OddballOddball””
Slide modified from Rik Henson