Evaluative Konditionierung: Zwei Wege zum gleichen Effekt?

Download Report

Transcript Evaluative Konditionierung: Zwei Wege zum gleichen Effekt?

The Influence of a Valence
Focus on Evaluative
Conditioning
Anne Gast & Klaus Rothermund
University of Jena
overview
• Evaluative Conditioning (EC)
• Is a focus on valence necessary for EC?
– Experiment 1
• Mechanism of the valence focus
– Experiment 2
• How specific is a valence focus?
– Experiment 3
• General conclusions and discussion
Evaluative Conditioning (EC)
(Martin & Levey, 1978)
• Pairing evaluatively neutral stimulus (CS) with
positively or negatively evaluated stimulus (US)
CS changes valence towards US
(repeated) pairwise presentation
post-conditioning rating
Research question 1:
Do Evaluative Conditioning effects only
occur, if we focus on evaluation during
conditioning?
– Earlier findings
– Experiment 1
Disruptive influence of secondary task
14
12
10
8
6
4
2
0
supraliminal
subliminal
no distraction
supraliminal
subliminal
distraction
• Distraction prevents EC-effects
• not due to cognitive capacity
• Importance of a valence focus?
Field & Moore (2005)
Research question 1
• Valence focus during conditioning has
an influence on EC-effect
 Is evaluative response during
presentation of the pair decisive?
 Is response toward stimulus associated
with CS?
Positive valence
CS
US
Possibly learned associations:
Positive
valence
CS-USassociation
evaluation
Positive valence
CS-evaluationassociation
Former results on CS-USassociations
• US-revaluation
• Sensory preconditioning
US-revaluation
PRO: Baeyens et al., 1992
1. CS is paired with valent US
 CS takes over US‘ valence
2. US is revaluated with opposite information
 CS does not change its valence
30
20
10
positive
negative
0
-10
-20
congruent revaluation incongruent revaluation
CONTRA: No US revaluation effect: Baeyens, et al., 1998
experiment 1
•
•
Influence of task during conditioning (valence
judgment vs. age judgment)
Manipulation of specificity of CS-US-pairings
CS-evaluation-association
valence judgment
 EC-effect
age judgment
 no EC-effect
CS-US-association
specific pairings
non specific pairings
 strong EC-effect
 weaker EC-effect
experiment 1: procedure
1. Conditioning
positive
2. Post-Rating
negative
Paired with…
…8 different US
from one
category
Choice of
stimuli
(pilot)
8 pictures
as CS
32
adjectives
as US
strong
healthy flexible
etc.
8x
…1 US
healthy
Valence task
Age task
Positive or negative
impression?
Old or young
impression?
How
positive/negative?
difference Cspos – Csneg
experiment 1: results
3
4
*
3
positive
*
2
1
ns
2
ns
1
negative
0
-1
0
-2
-1
-3
specific non specific specific non specific
-2
valence judgment
specific pairing
age judgment
valence judgement
age judgement
non specific pairing
EC-effect
F(1,60) = 9,75, p < .01, η = .14
EC * judgement
F(1,60) = 4,89, p < .05, η = .075
EC * specificity
F(1,60) < 1, ns.
EC * judgement * specificity
F(1,60) < 1, ns.
conclusion experiment 1
• Focus on valence is important
• Specificity of pairing is not important
 Is the response given during conditioning
associated with the CS?
research question 2:
What is the mechanism?
• EC is due to association between CS and
evaluation.
Effect only if the response is evaluation
• EC is due to association between CS and
US, but due to the non-evaluative task USvalence is temporarily inhibited
 Reactivation of US-valence returns effect
experiment 2: procedure
2. US-Reactivation
1. Conditiong
8x
healthy
Valence vs. age judgement
Evaluative reaction
on single US
3. Post-rating
Healthy
Positive or
negative?
How
positive/negative?
experiment 2: results
**
**
Main effect valence
F(1,62) = 20.524, p < .001, η = .249
Valence * judgement
F(1,62) < 1
Conclusion from Exp 1 & 2
• non-evaluative task hinders EC- effect
(Exp. 1)
• This is due to a surpression of US‘ valence
if evaluation is not task relevant
• reactivation of US returns effect
(Exp. 2)
CS-US-association + US-valence  EC-effect!
Research question 3:
How stimulus specific is the valence
focus effect?
• Is it the specific stimulus that is judged on
valence that is „switched on“?
• Are all stimuli present in the context are
„switched on“?
experiment 3
How specific is the valence focus?
Judgmental Task is manipulated within participants. Different
CS-US-pairs are combined with the two different tasks:
Pair 1: US1 – CS1: valence task
Pair 2: US2 – CS2: age task
Pair 3: US1 – CS3: age task (but US is in valence task in pair 1)
Hypotheses:
Pair 1: EC-effect
(valence of US is activated in these trials)
Pair 3: EC-effect
(valence of US is activated, in other trials)
Pair 2: EC-effect
(valence of US is not directly activated, however
evaluation takes place in the context)
experiment 3: procedure
1. Conditiong
US evaluated here
Pair 1:
evaluate!
healthy
Evaluation in context
Pair 2:
judge age!
flexible
Pair 3:
judge age!
healthy
Effect here
 context based
US evaluated in
other pair
2. Post-rating
CS-evaluation after conditioning
Experiment 3 – results
1,5
ns
+
1
ns
0,5
positive
0
negative
-0,5
-1
-1,5
evaluation
task
evaluated in
other pair
age task
df
F
p
partial η2
EC-effect
(1,57)
4.724
< .05
.077
EC * Focus
(2,114)
<1
conclusion experiment 3
• Evaluative focus is not stimulus specific. In
a context were some stimuli are evaluated
the valence of all other similar stimuli is
active aswell.
General conclusions
• Evaluative Effects are only found if an evaluative
focus is active during the learning trials
• This is due to temporal supression of stimulus
valence if only non-evaluative dimensions are
considered.
• If the valence of only some stimuli is task
relevant, this is enough for all stimuli‘s valence
to be activated.
• Evaluative learning takes places in evaluative
contexts and less when attention is on other
dimensions
Thank you for your attention!
Experiment 1 : results
Mediation analysis of valence judgment (itemwise)
judgment
β = .794***
US
valence
β = .719***
CS
valence
β = .211* (without judgment)
β = -.359** (with judgment)
Judgment: Times judged positive – times judged negative during conditioning
experiment 1 (unspecific pairings):
results
3
2,5
2
1,5
CS pos
1
CS neg
0,5
0
-0,5
valence task
age task
-1
Main effect US-type: F(30, 1) = 2.41, p = .131
US-type*task: F(30,1) = 3.875, p = .058, ηpartial2 = .114
US-type under valence task: t(15) = 2.481, p < .05, d = .62
US-type under age task: t(15) = -.295, p = .772
Experiment 1: „valence focus“
Manipulated: task focus indirectly via a
secondary task during conditioning
1. Categorize in respect to valence
2. Categorize in respect to style
3. Control: no task
Experiment 2: procedure
3. conditioning
1. Baselineevaluation
CS
USneg
USpos CS
N-L
N-D
4 pairs
choice
4. Postconditioningevaluation
4 pairs
Pairwise presentation (5x)
CS + US
valence task
style task
Do you like this
garment?
Casual or evening?
Experiment 2: results
Evaluative conditioning effects (difference CSpos – CSneg)
under different task foci
2,5
2
1,5
high cog. Load
low cog. Load
1
0,5
0
valence task
style task
no task
Main effect valence: F(1,97) = 23.369, p< .001, ηpartial2 = .194
US-valence x task: F(2,97) = 2.61, p = .079, ηpartial2 = .194
Contrast style task – valence task and control: t(99) = 1.892, p = .061
experiment 2: results
1060
1040
7.5
US-judgement in ms
number of correct valence
judgements
8
7
6.5
6
5.5
5
4.5
1020
1000
980
960
940
920
900
880
860
840
4
valence judgement
age judgement
valence judgement
age judgement
Sensory Pre-conditioning
Hammerl & Grabitz, 1996; Walther, 2002
1.
CS1 (neutral) is paired with CS2 (neutral)
2.
Only CS1 is paired with US (valent)
CS1 changes into the direction of the US
also CS2 changes into direction of US
30
20
20
10
10
positive
control
0
0
-20
-20
-30
sensory
preconditioning
Walther (2002), experiment 1
control
-10
-10
direct pairing
negative
direct pairing
sensory
preconditioning
Walther (2002), experiment 2
results of studies on US-revaluation
and sensory pre-conditioning
• After pairing CS-valence depends on USvalence.
• After pairing CS-valence depends not on USvalence.
 majority of evidence speaks for CS-USassociations
experiment 1
material
CS: 8 pre-chosen portrait fotos (pre-study N = 38): neutral on the
dimensions age and valence
US: prechosen adjectives (pre-studies N = 17/22/15):
8 positive/young, 8 positive/old, 8 negative/young, 8 negative/old
design
1.
valence of US (within)
2.
Age of US (within)
3.
judgment task during conditioning (age/valence; between)
4.
Specificity of CS-US-pairing between)
Conditioning procedure
Picture-CS is paired with positive or negative adjectives
Task: judgment of picture and word as a whole (age or valence)
Conditioning trials:
500 ms CS only, 2200 ms CS & US, 1000 ms CS, US & response
CS-rating after conditioning
experiment 1: results
3
difference Cspos – Csneg
2
4
1
3
paired positive
0
2
paired negative
1
-1
0
-2
-1
-3
-2
specific
non specific
valence judgement
specific
non specific
valence judgment
age judgement
specific pairing
age judgment
non specific pairing
EC-effect
F(1,60) = 9,75, p < .01, η = .14
EC * judgement
F(1,60) = 4,89, p < .05, η = .075
EC * specificity
F(1,60) < 1, ns.
EC * judgement * specificity
F(1,60) < 1, ns.
valence-judgement, specific pairing
t(15) = 2.419, p < .05, d = .61
valence-judgement, non specific pairing
t(15) = 2.481, p < .05, d = .62
age-judgement, specific pairing
t(15) = 1.509, p = .152, ns.
experiment 2b
increased power - specific pairing, age judgement
CS-ratings after conditioning
2
1,5
1
0,5
0
-0,5
-1
positive pairing
negative pairing
EC-effect
t(31) = 1.059, p = .298
power-analysis
(α = .05; d = .62; N = 32): β =.924
experiment 2: results
Main effect valence
F(1,62) = 20.524, p < .001, η = .249
Valence * judgement
F(1,62) < 1
Valence in valence judgement
F(1,31) = 8.732, p < .01, η = .22
Valence in age judgement
F(1,31) =, 11.805, p < ,01, η = .276
Valence * Reinstatement
F(1,62) = 3.775, p = .057, η = .057
2
1.5
1
0.5
positive
0
negative
-0.5
-1
-1.5
-2
Correctly
remembered
pairs
EC-Effect (difference pos - neg)
EC-Effect (difference pos - neg)
Results experiment 3
2
1.5
*
+
*
1
evaluation
task
0.5
evaluated in
other pair
0
age task
-0.5
-1
-1.5
Correctly
remembered
pairs
Incorrectly
remembered
pairs
incorrectly
remembered
pairs
F
p
partial η2
Valence
6.86
<.05
.132
Awareness
1.823
.184
.039
Valence*Awareness
6.72
<.05
.13
df
experiment 3: procedure
2. USReinstat
ement
1. Conditiong
positive
negative
Paired with…
multihealthy
cultura
l
…8 different US
from one
category
flexible
etc.
8x
Evaluative
reaction on
single US
Healthy
…1 US
e.g.
Healthy
Valence task
Age task
Positive or negative
impression?
Old or young
impression?
2. Postconditioning-rating
Positive
or
negative
?
How
positive/negative?
Results experiment 3
2
**
Evaluation after conditioning
1,5
+
**
*
1
0,5
positive
negative
0
-0,5
-1
-1,5
valence task
age task
specific pairing
valence task
age task
unspecific pairing
df
F
p
partial η2
Valenz
(1,124)
26.273
< .001
.175
Valenz * task
(1,124)
<1
Valenz * Spec
(1,124)
1.220
Valenz * task * Spec
(1,124)
<1
.007
.271
.010
.001
Experiment 1,2 & 3: results
1.5
2.5
2
1.5
1
0.5
0
-0.5
-1
-1.5
-2
1
0.5
positive
negative
0
valence
task
-0.5
age task
No US-reactivation
valence
task
age task
specific positive
specific negative
unspecific positive
unspecific negative
valence
task
US-reactivation
age task
No US-reactivation
valence
task
age task
US-reactivation
-1
EC-effect
EC * judgement
EC * Spec
EC * judgement *
reactivation
EC * judgement * Spec *
reactivation
df
F
p
partial η2
(1,264)
(1,264)
(1,264)
(1,264)
40.220
8.012
<1
4.65
< .001
<.01
.132
.029
<.05
.017
(1,264)
<1