Upload
nguyenminh
View
226
Download
6
Embed Size (px)
Citation preview
Bottom-up and top-downeffects of unconsciousprocessingprocessing
Eva Van den Bussche & Bert Reynvoet
SCAN 16-06-2011
6/23/2011 1Herhaling titel van presentatie
Overview
• Introduction
• Bottom-up: The depth of unconscious processing
Pag.
processing
• Top-down: Is consciousness necessary for cognitive control?
• A new paradigm
• General conclusions
16-06-2011 2SCAN Utrecht
Introduction
• Can unconsciously / subliminally presented information influence our behaviour?
17ms
33ms
9
##
6
Pag.
�Masked priming paradigm (Marcel, 1983)
16-06-2011 3SCAN Utrecht
Larger than 5Smaller than 5
33ms
480ms
6
##
Introduction
�Masked priming paradigm
Congruent
condition
Incongruent
condition
Pag.16-06-2011 4SCAN Utrecht
condition condition
Prime 6 1
Target 9 9
Both > 5
Faster RTs
One <, one > 5
Slower RTs
Difference (RT_I - RT_C) = priming effect
Bottom-up
• Can unconsciously / subliminally presented information influence our behaviour? YES!
Pag.
behaviour? YES!
• BUT: how far does this bottom-up strength reach?
• How deep can unconscious stimuli be processed?
16-06-2011 5SCAN Utrecht
Bottom-up
• How deep can unconscious stimuli be processed? � 2 classes of theories:
– Semantically (e.g. Dehaene et al.,
Pag.
– Semantically (e.g. Dehaene et al., 1998)
– Non-semantically, but due to low-level Stimulus-Response (S-R) mappings (e.g. Damian, 2001; Kunde et al., 2003)
16-06-2011 6SCAN Utrecht
Bottom-up
Larger than 5!
Response = Congruent
Prime6
1. Semantic account:
Smaller than 5!
Response = Incongruent
Prime1
Pag.16-06-2011 7SCAN Utrecht
Response = press right
Congruent � faster RT!
Target9 Larger than 5!
Response = press right
Response = press left
Incongruent � Slower RT!
Bottom-up
Larger than 5!
Response =
Prime6
2. Non-semantic account:
Response =
Prime1
Pag.16-06-2011 8SCAN Utrecht
Response = press right
Congruent � faster RT!
Target9
Response = press right
Response = press LEFT
Incongruent � Slower RT!
Bottom-up
• BUT: formation of S-R links limited to:
– Expected stimuli (e.g. repeated primes, same format)
Pag.
– Small number of stimuli (e.g. small sets, small categories)
� NO subliminal priming expected for:
unexpected stimuli, large stimulus sets, large stimulus categories
16-06-2011 9SCAN Utrecht
Bottom-up
• However:
- Subliminal priming for large stimulus sets, large categories (for a meta-analysis see Van den Bussche et al., 2009a, Psych Bull)
Pag.
Van den Bussche et al., 2009a, Psych Bull)
- Subliminal priming for unexpected formats (Dell’Acqua & Grainger, 1999; Van den Bussche et al., 2009b; Van den Bussche et al., under review)
16-06-2011 10SCAN Utrecht
Bottom-up
• Conclusion:
Despite the limited bottom-up strength, unconscious stimuli can influence
Pag.
unconscious stimuli can influence behavior and be processed up to a high semantic level!
~ consensus
16-06-2011 11SCAN Utrecht
Top-down
• But, can an unconscious stimulus modify unconscious processing by exerting top-down control?
Pag.
= heavily debated!
• Subliminal processing = ‘‘a condition of information inaccessibility where bottom–up activation is insufficient … for global ignition’’ (Dehaene et al., 2006, p.3).
16-06-2011 12SCAN Utrecht
Top-down
• Double prediction (GNW):
1. Unconscious processes are sensitive to conscious strategic and contextual influences
Pag.
influences
2. However, an unconscious stimulus itself cannot modify unconscious processing by exerting top-down control
16-06-2011 13SCAN Utrecht
Top-down
1. Unconscious processes are sensitive to conscious strategic and contextual influences
� Yes!
Pag.
� Yes!
Unconscious priming can be influenced by temporal attention (Naccache et al., 2002), spatialattention (Sumner et al., 2006) and strategies, such as target set (Greenwald et al., 2003) and target notation (Kunde et al., 2003)
16-06-2011 14SCAN Utrecht
Top-down
• Example: Kunde et al. (2003)Targets Primes Priming?
1, 4, … 1, 4, … Yes
1, 4, … one, four, … No
Pag.
� Unconscious priming effects were dependent
on the target notation (i.e. a conscious manipulation)
16-06-2011 15SCAN Utrecht
1, 4, … one, four, … No
one, four, … 1, 4, … No
one, four, … one, four, … Yes
Top-down
2. However, an unconscious stimulus itself cannot modify unconscious processing by exerting top-down control
� ???
Pag.
� ???
16-06-2011 16SCAN Utrecht
Top-down
• Our study (VDB et al., 2008, Con & Cog)
• Investigate the 2 predictions by manipulating the proportion (50-50, 25-75 or 75-25) of Arabic numbers (e.g. 1) versus
Pag.
or 75-25) of Arabic numbers (e.g. 1) versus number words (e.g. “one”):
– Consciously (target level): experiment 1
– Unconsciously (prime level): experiment 2
• All other task conditions were controlled
16-06-2011 17SCAN Utrecht
Top-down
Congruent trial
nine
#$#$#$33ms
Pag.16-06-2011 18SCAN Utrecht
7
$#$#$#
33ms
300ms
Larger than 5Smaller than 5
Top-down
0
5
10
15
20
Pri
min
g e
ffe
ct
(ms
)
N = 15 N = 16 N = 15 N = 16 N = 16
F(2,43) = 3.70, p = .03 F(2,44) = 0.22, p = .81
Pag.
� The conscious manipulation exerted top-down control on the priming effects, but the unconsciousmanipulation did not
16-06-2011 19SCAN Utrecht
-5
25% Arabic targets
75% Arabic targets
baseline 25% Arabic primes
75% Arabic primes
Pri
min
g e
ffe
ct
Arabic primes
Number word primes
N = 15 N = 16 N = 15 N = 16 N = 16
Top-down
2. However, an unconscious stimulus itself cannot modify unconscious processing by exerting top-down control
� Yes!
Pag.
� Yes!
� But…
16-06-2011 20SCAN Utrecht
Top-down
• Is consciousness necessary for cognitive control?
Yes: Ansorge et al., in press; Kunde, 2003; Merikle & Joordens, 1997; Nieuwenhuis et
Pag.
Merikle & Joordens, 1997; Nieuwenhuis et al., 2001; Van den Bussche et al., 2008
No: Bodner & Masson, 2001; Cohen et al., 2009; Hughes et al., 2009; Lau & Passingham, 2007; van Gaal et al., 2008; 2010
16-06-2011 21SCAN Utrecht
Top-down
• Unconscious manipulations can NOT influence priming (Merikle & Joordens, 1997; Van den Bussche et al., 2008)
• An unconscious conflict can NOT induce
Pag.
• An unconscious conflict can NOT induce conflict adaptation effects (Ansorge et al., in press;
Kunde, 2003)
• Unconscious errors do NOT lead to post-error slowing (i.e. remedial action to prevent future errors) (Nieuwenhuis et al., 2001)
16-06-2011 22SCAN Utrecht
Top-down
• Unconscious manipulations can influence priming (Bodner & Masson, 2001)
• An unconscious conflict can induce conflict adaptation effects (van Gaal et al., 2010)
Pag.
• Unconscious errors can lead to post-error slowing (Cohen et al., 2009)
• Unconscious information can act as a “stop signal” and urge subjects to interrupt their ongoing response to a task (Hughes et al., 2009; van Gaal et al., 2008)
• Unconscious information can influence task selection (Lau & Passingham, 2007)
16-06-2011 23SCAN Utrecht
Top-down
• Aforementioned studies suffer from critical problems:
1. Objective post-hoc awareness tests which are
Pag.
1. Objective post-hoc awareness tests which are heavily debated
2. Visual strength of the stimuli always critically differed in the conscious and unconscious conditions
3. Different experimental paradigms � difficult to
compare
16-06-2011 24SCAN Utrecht
A new paradigm
• Our current aim: investigate whether consciousness is needed for cognitive control, but using a paradigm which
Pag.
control, but using a paradigm which avoids the problems demonstrated for previous studies!
• How?
16-06-2011 25SCAN Utrecht
A new paradigm
• Design:
#$#$#$ RED $#$#$# &&&&&How certain are you that the
colour word presented before
the target was "RED" or
"BLUE"?
Pag.
Trials are divided based on prime awareness task in conscious and unconscious trials
16-06-2011 26SCAN Utrecht
Target
classification
"BLUE"?
1 2 3 4 5
A new paradigm
• Advantages:
1. Trial-based assessment method
2. Visual strength identical in conscious and
Pag.
2. Visual strength identical in conscious and unconscious conditions
3. Paradigm can be used to examine a wide array of top-down effects (priming, conflict adaptation, …)
16-06-2011 27SCAN Utrecht
A new paradigm
• Disadvantages:
1. Requires many participants
2. Requires many trials (exclusion of uncertain
Pag.
2. Requires many trials (exclusion of uncertain trials)
3. Takes some time to master
16-06-2011 28SCAN Utrecht
A new paradigm
• Does it work?
- Took many pilot studies to get it right (masks, prime duration, PAS-scale, number of trials, etc.)
Pag.
number of trials, etc.)
- First “finished” study just completed!
16-06-2011 29SCAN Utrecht
A new paradigm
480 ms 40 ms 27 ms 2000 ms
$#$#$# RED $#$#$# &&&&&
How certain are you that the colour word RED or BLUE was presented before the coloured
symbols?
1.I am certain that I saw the colour
+Press the spacebar when you're
ready to start the next trial.Place your fingers on buttons "a"
and "p" again to be able to categorize the coloured symbols
as fast as possible
Design (320 trials):
Pag.16-06-2011 30SCAN Utrecht
1.I am certain that I saw the colour word RED: press button "1“2.I think I saw the colour word RED, but I am not certain: pressbutton "2”3.I did not see the colour word: press button "3”4.I think I saw the colour word BLUE: press button "4”5.I am certain that I saw the colour word BLUE: press button "5"
as fast as possible
If the symbols are red: press "a"If the symbols are blue: press "p"
p (BLUE)a (RED)
1 2 3 4 5
SPACE
A new paradigm
• Results- N=60 � 24 excluded
- Conscious trials (on average 99) = subjects correctly identified the primes and were 100%
Pag.
correctly identified the primes and were 100% certain
- Semi-conscious trials (on average 97) = subjects correctly identified the primes but were not certain
- Unconscious trials (on average 61) = subjects reported not being able to identify the prime
16-06-2011 31SCAN Utrecht
A new paradigm
• Results:- Repeated measures with condition (conscious,
semi-conscious or unconscious) and congruency (congruent or incongruent) as within-subject
Pag.
(congruent or incongruent) as within-subject factors
- Significant interaction between condition and congruency (F(2,33)= 15.81, p< .001)
16-06-2011 32SCAN Utrecht
A new paradigm
146ms***
Pag.16-06-2011 33SCAN Utrecht
33ms*
24ms*
A new paradigm
• Does it work?
- Yes!
- Next step: examine the effect of top-
Pag.
- Next step: examine the effect of top-down manipulations on the conscious VS unconscious trials
16-06-2011 34SCAN Utrecht
General conclusions
• Unconscious information
- Can influence our behavior
- Has limited bottom-up strength, but is
Pag.
- Has limited bottom-up strength, but is still able to be processed up to a high, semantic level
- Seems to be able to exert top-downcontrol as well, although further research using an improved paradigm is needed
16-06-2011 35SCAN Utrecht
Contact
Eva Van den Bussche
Vrije Universiteit Brussel
Department of Psychology
Pleinlaan 2
Pag.
Pleinlaan 2
1050 Brussels
Belgium
http://homepages.vub.ac.be/~evdbussc//
16-06-2011 36SCAN Utrecht