This article was downloaded by: [Van Damme, Stefaan] On: 30 November 2010 Access details: Access Details: [subscription number 930338992] Publisher Psychology Press Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 3741 Mortimer Street, London W1T 3JH, UK
The Quarterly Journal of Experimental Psychology
Publication details, including instructions for authors and subscription information: http://www.informaworld.com/smpp/title~content=t716100704
Aversive conditioning under conditions of restricted awareness: Effects on spatial cueing An K. Raesa; Ernst H. W. Kostera; Stefaan Van Dammea; Wim Fiasa; Rudi De Raedta a Ghent University, Ghent, Belgium First published on: 04 August 2010
To cite this Article Raes, An K. , Koster, Ernst H. W. , Van Damme, Stefaan , Fias, Wim and De Raedt, Rudi(2010)
'Aversive conditioning under conditions of restricted awareness: Effects on spatial cueing', The Quarterly Journal of Experimental Psychology, 63: 12, 2336 — 2358, First published on: 04 August 2010 (iFirst) To link to this Article: DOI: 10.1080/17470218.2010.492995 URL: http://dx.doi.org/10.1080/17470218.2010.492995
PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY 2010, 63 (12), 2336– 2358
Aversive conditioning under conditions of restricted awareness: Effects on spatial cueing An K. Raes, Ernst H.W. Koster, Stefaan Van Damme, Wim Fias, and Rudi De Raedt
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
Ghent University, Ghent, Belgium
In the literature on aversive conditioning there is still debate on the role of awareness. According to some authors, affective learning can occur with or without contingency learning (dual-process model), whereas others argue that a single process produces both affective responses and contingency knowledge. Although many studies have investigated these models, the results to date are inconclusive. Based on a review of the literature, a new series of experiments was designed to examine aversive conditioning in the absence of contingency awareness. In the present study we examined the effects of subliminal aversive conditioning on a spatial cueing task. Awareness was stringently tested after conditioning. Three kinds of awareness were distinguished: contingency awareness (awareness of the CS2US contingencies, where CS is the conditioned stimulus, and US is the unconditioned stimulus), perceptual awareness (awareness of the perceptual differences between the CSs), and US expectancy (awareness of a threat feeling when confronted with the CS+, but not when confronted with the CS2). The results of three experiments demonstrated that responses on the spatial cueing task were modulated by subliminal aversive conditioning. Importantly, none of the participants was contingency aware or able to perceptually discriminate between the conditioned stimuli. However, in Experiment 3, only those participants showing some level of postconditioning expectancy awareness exhibited conditioning effects. These experiments suggest that subliminal aversive conditioning produces small but significant effects, which may be modulated by expectancy awareness. Keywords: Conditioning; Awareness; Contingency; Fear; Attention.
Classical conditioning refers to the observation that after repeated pairing of a neutral stimulus (conditioned stimulus; CS) with a motivationally relevant stimulus (unconditioned stimulus; US), the CS comes to elicit a response (conditioned response; CR) even in the absence of the US (Mackintosh, 1983; Pavlov, 1927). In humans and nonhuman animals, classical conditioning is
considered a crucial form of learning, which allows predicting future events and modifying behaviour accordingly. This form of learning is thought to be particularly relevant in the face of ¨ hman & Mineka, 2001). threat (O According to several authors, nonhuman animals as well as humans possess an affective learning system that can operate without the
Correspondence should be addressed to An Raes, Ghent University, Department of Experimental-Clinical and Health Psychology, Henri Dunantlaan 2, B-9000 Gent, Belgium. E-mail:
[email protected] An Raes is research fellow of the Scientific Research Foundation, Flanders (FWO), Belgium. The authors would like to thank Jan De Houwer for his useful comments on an earlier draft of the manuscript.
2336 http://www.psypress.com/qjep
# 2010 The Experimental Psychology Society DOI:10.1080/17470218.2010.492995
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
CONDITIONING AND AWARENESS
involvement of higher cognitive functions (Hamm ¨ hman, 2005) & Weike, 2005; LeDoux, 2003; O and that is distinguishable from the propositional learning of contingencies between the CS and the US (Baeyens, Eelen, & Crombez, 1995; Hamm & Weike, 2005). This viewpoint is often referred to as a dual-process perspective. One of the most important predictions from this perspective is that a CR can be produced in the absence of knowledge of the CS2US relationship (contingency awareness; CA). A vast number of studies have claimed to demonstrate human affective learning in the absence of propositional learning by investigating the relation between conditioned responding and CA. However, careful examination of the literature in this domain reveals that this claim is not unambiguously supported. In most review articles on this topic, it is concluded that propositional learning processes can account for the acquisition of both conditioned responding and CA (Dawson & Schell, 1985, 1987; Lipp & Purkis, 2005; Lovibond & Shanks, 2002; Mitchell, De Houwer, & Lovibond, 2009). Most previous studies in this research area have investigated the association between CA and CR using eyeblink conditioning (e.g., Clark & Squire, 1998), autonomic conditioning and/or startle responses (e.g., Hamm & Vaitl, 1996; Purkis & Lipp, 2001), neurophysiological ¨ hman, & Dolan, 1998; responses (e.g., Morris, O Wong, Bernat, Snodgrass, & Shevrin, 2004; Wong, Shevrin, & Williams, 1994), or CS valence ratings (in evaluative conditioning; e.g., Baeyens, Eelen, & Van den Bergh, 1990; as in Lovibond & Shanks, 2002). In most cases, these approaches have yielded inconsistent results or failed to provide evidence for a dual-learning system. It may be that the methods used in these studies were not sensitive to unaware conditioning effects. Therefore, we revisited the question of conditioning without CA by using a different approach. In the present study, subliminal conditioning was used to prevent CA, and conditioned responding was indexed with an implicit behavioural measure. In subliminal conditioning, the CSs are
presented below the perceptual threshold. It is implausible that a conscious (conditioning) process is established based on unconscious input (as in Moors & De Houwer, 2006). Therefore, it is generally accepted that CA is outright prevented in this approach, at least when CS unawareness is thoroughly checked. Unfortunately, the awareness checks of most foregoing subliminal conditioning studies are invalid or unreliable (Lovibond & Shanks, 2002). Most awareness checks consist of a recognition task, in which the subliminally presented CS (e.g., a snake) has to be fully identified (Lovibond & Shanks, 2002). However, the fact that participants cannot identify a masked stimulus (e.g., a snake) correctly, does not rule out the possibility that they can discriminate between this stimulus and another stimulus (e.g., a spider) based on differences in low-level features. And of course, the crucial condition for valid subliminal (differential) conditioning is that participants cannot discriminate between the CS+ and the CS2(Lovibond & Shanks, 2002). Therefore, the results of many of the studies that have claimed to find evidence for unaware aversive conditioning ¨ hman & Soares, 1998; Soares & O ¨ hman, 1993; (O Wong et al., 1994) should be interpreted with caution. In the present study, efforts are made to ensure that participants are unaware of the CS2US contingencies. Also, CA is checked by using various relevant and sensitive awareness checks. Another relevant issue in awareness research in the context of conditioning is the distinction between contingency and expectancy awareness. Following the description of Lovibond and Shanks (2002), CA is the knowledge of the CS2US contingencies, whereas expectancy awareness (EA) refers to the expectancy of threat (US) at CS presentation. In most cases, both forms of awareness co-occur. However, it is possible that participants expect the US but cannot explain why. In that case, there is EA but no CA. EA can then be conceptualized as a “gut feeling”: People experience a feeling of danger when they are presented with the CS, but cannot explain why. Most studies do not distinguish between expectancy and contingency awareness.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
2337
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
RAES ET AL.
This might confound the interpretation of their results, because it is possible that EA correlates with CR, while CA does not (Katkin, Wiens, & ¨ hman, 2001). In the present study, we investiO gate whether there is an association between EA and conditioned responding in the absence of CA. Another modification in comparison with most previous studies (except for Dawson, Rissling, Schell, & Wilcox, 2007; Pleyers, Corneille, Luminet, & Yzerbyt, 2007; on evaluative conditioning) is the use of an implicit behavioural measure to index CR. More specifically, a spatial cueing task is used to index subliminal aversive conditioning. This approach has several advantages. First, the use of this measure enables us to assess the effect of aversive conditioning on early orienting responses (Fox, Russo, Bowles, & Dutton, 2001). Second, participants are unable to strategically control their response pattern on this measure. As such, we can establish whether subliminal aversive conditioning effects manifest themselves on an automatic level of responding. The spatial cueing task (SCT; Posner, 1980) has repeatedly been used to measure aversive conditioning effects on an implicit behavioural level. However, in these studies, contingency awareness was not restricted (Koster, Crombez, Van Damme, Verschuere, & De Houwer, 2004; Van Damme, Crombez, Hermans, Koster, & Eccleston, 2006). The present study comprises three experiments. The basic design of these experiments is as follows. Participants are not merely exposed to the CS2US contingencies. Instead, the contingencies are embedded in an SCT from the outset. In this task, a visual cue that is presented on the right or left side of the computer screen, correctly (valid trial; 75% of the trials) or incorrectly (invalid trials; 25% of the trials) predicts the spatial location of a target. With short stimulus onset asynchronies, responses to valid trials are typically faster than responses to invalid trials. This is referred to as the cue validity effect (CVE). Several studies have found that emotional cues presented for short durations (,500 ms) enhance the CVE compared to neutral cues, which indicates increased attention to emotional
2338
cues (Fox et al., 2001; Koster, Crombez, Verschuere, Vanvolsem, & De Houwer, 2007). In a modified version of the SCT, one of two cues (CS+) is paired with an intrinsically threatening stimulus (US), while another cue (CS2) is associated with a neutral stimulus or with the absence of the US (Stormark & Hugdahl, 1996). Most former studies have demonstrated a larger CVE for the CS+ than for the CS2, indicating enhanced attentional processing of the CS+ cue (Koster et al., 2004; Van Damme et al., 2006). The current study aims at exploring spatial cueing effects under masked presentation of the CSs. In the present study, two fear-relevant stimuli serve as CSs. We use male angry faces as CSs, as these have been used most frequently in related research (e.g., Armony & Dolan, 2002; Morris et al., 1998). The angry faces are masked by a neutral face, because previous research has suggested that faces adequately mask other faces (Carlson & Reinke, 2008; Costen, Shepherd, Ellis, & Craw, 1994). Across our experiments, several stimuli are used as USs. In Experiment 1, we use aversive pictures as USs and neutral pictures as control stimuli. In the subsequent experiments, bursts of white noise serve as USs, with soft white noises as control stimuli. Control stimuli for the USs are used to ensure that both the CS+ and the CS2 hold a signal value for a subsequent stimulus, with only the CS+ predicting an aversive stimulus (i.e., the US). In each experiment, valence ratings are administered to check for the valence of the USs and the control stimuli. Awareness of the CSs during the SCT is prevented by presenting them subliminally (16.7 ms) in combination with a backward masking procedure. Furthermore, participants are not presented with the unmasked versions of the CSs, nor are they informed about the presence of masked stimuli or conditioning to these stimuli. After the SCT, awareness of the CSs is thoroughly checked. First, CA is assessed by explicitly asking participants about the CS2US contingencies. Second, to index EA, participants rate US expectancy for both subliminal and supraliminal presentations of the CSs. Third, perceptual
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
CONDITIONING AND AWARENESS
awareness (PA) of the CSs is gauged via a forcedchoice task, in which participants are asked to discriminate between the CS+ and the CS2 based on low-level features of these stimuli. If participants are unable to discriminate between the CSs, it is unlikely that they are aware of any process operating upon the difference between the CSs (e.g., conditioning; Moors & De Houwer, 2006). Therefore, this can be considered a valid and conservative awareness check. In sum, the present study tests the hypothesis that aversive conditioned responding can be established in the absence of awareness, using a subliminal conditioning paradigm that is embedded in a spatial cueing paradigm. Postconditioning assessments of CA, EA, and PA are provided.
EXPERIMENT 1 Method Participants A total of 41 undergraduate students took part in this experiment in exchange for course credits or payment. All were tested individually. A total of 3 participants were excluded from further data analyses because their error rate deviated more than 2.3 standard deviations from the group mean. We did not include participants reporting a lack of sleep (less than 5 hours), medication use, being ill, or having used drugs or alcohol on the day of the experiment; 4 participants were excluded based on these criteria. Mean age of the remaining 34 participants was 19.71 years (SD ¼ 2.18). Most participants (82.4%) were female. Stimuli All stimuli were presented on a black background on a 17-inch screen with a resolution of 1,024 × 768 pixels. A total of 3 pictures of male faces of 261 × 261 pixels served as CSs and masking face. These pictures were selected from a large database of emotional faces, the Karolinska Directed Emotional Faces (KDEF; Lundqvist, ¨ hman, 1998). Prior to the experiment, Flykt, & O 80 pictures were selected from this database.
These pictures had been categorized on four valence dimensions (angry, happy, sad, neutral) according to the norms of Goeleven, De Raedt, Leyman, and Verschuere (2008). Subsequently, these 80 pictures were rated by 73 volunteers (students, colleagues, and acquaintances) on valence. Based on these ratings, we could determine which of the angry male faces were most consistently rated as negative (selected as CSs) and which male face came out as most neutral (selected as mask). The allocation of the faces to the function of CS+ and CS2 was counterbalanced across subjects. Four pictures of 640 × 480 pixels served as USs and control stimuli for the USs. These pictures were taken from the International Affective Picture System (IAPS; Lang, Bradley, & Cuthbert, 1997). Two threatening pictures (1050: snake; 1300: pitbull) and two neutral pictures (7175: lamp; 7009: mug) were selected and were paired with the CS+ and CS2, respectively. A white square of 1 × 1 cm served as target stimulus. A Cedrus RB-730 response box (Cedrus Corporation, San Pedro, CA) was used for response collection. Procedure Participants were seated approximately 50 cm from the computer screen. Before starting with the SCT, they gave their written informed consent. The spatial cueing task. General information and trial description. The spatial cueing task (SCT; Posner, 1980) was programmed in the E-prime software package (Psychology Software Tools, Inc., 2001). The experiment was run on a Fujitsu Siemens Amilo Pro V3505 laptop coupled with a Philips 107P4 CRT screen that displayed the stimuli. Screen refresh rate was 60 Hz. Before the experiment was performed, timing accuracy was checked using independent hardware. More specifically, an optocoupler circuit, which was connected to the CRT screen by glass fibre, was coupled with a digital oscilloscope on which the number of
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
2339
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
RAES ET AL.
refresh rates of each stimulus presentation could be assessed. The sequence of events in the SCT is depicted in Figure 1. Each trial started with a black screen presentation of variable interval (700 ms to 2,000 ms). Subsequently, a white fixation cross was presented in the middle of the screen, accompanied by two white rectangles, right and left of fixation. The peripheral presentations subtended 9 degrees of the visual field. After 500 ms, either the CS+ or the CS2 replaced the left or the right rectangle during one refresh rate (16.7 ms). Both CSs were presented on both sides of the screen equally often. Then, the CS+/CS2 was masked by the neutral face, which was displayed at the same location for 183 ms. The neutral mask face was followed by a shortly presented black screen (16.7 ms) in order to prevent reduced target detection because of visual after-effects.
The target was presented after the black screen mask. Allocation of target position was dependent on allocation of CS position: In 75% of the trials, CS and target appeared on the same side of the screen (valid trials), while in 25% of the trials, CS and target were presented on opposite sides of the screen (invalid trials). Participants responded to the target as rapidly and accurately as possible. They responded with both index fingers, pressing the left key of the response box in response to left targets and the right key in response to right targets. On reinforced trials (see below) the USs or control stimuli were presented immediately after the participant’s response to the target for 500 ms. Their presentation covered the whole screen. On nonreinforced trials, a black screen was presented for the same duration. If the participant failed to respond, the US or control stimulus was presented 600 ms after target onset. Allocation
Figure 1. Overview of an invalid reinforced CS+ trial (above) and a valid reinforced CS2 trial (below). CS ¼ conditioned stimulus. US ¼ unconditioned stimulus. To view a colour version of this figure, please see the online issue of the Journal.
2340
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
CONDITIONING AND AWARENESS
of trial type (CS+/CS2; right/left cue position; valid/invalid; reinforced/nonreinforced) was randomized across all phases of the SCT. The SCT consisted of three phases. The aim of the practice phase was to get participants acquainted with the task. During the baseline phase, a baseline measurement of responding was assessed for each participant. No USs were presented during these phases. In the acquisition phase, USs and control stimuli were implemented in the design. Throughout baseline and acquisition phases, catch trials were presented occasionally to prevent automatic response tendencies. On these trials, no target was presented. As such, participants had to refrain from responding. In addition, digit trials were incorporated to keep participants’ attention allocated to the middle of the screen. On these trials, a centrally presented digit appeared at the beginning of a trial for 200 ms, and participants had to name the digit out loud. The experimenter kept track of participants’ responses to digit trials (absent/present; correct/incorrect). Practice phase. Participants were instructed about trial course, including fixation and cue and target presentation. They were not informed about the presence of masked CSs. Participants were asked to focus their attention on the fixation cross before the start of the practice trials. There were 16 practice trials on which participants had to attain an accuracy rate of 85% to be able to proceed. If a participant failed to attain this rate, the practice phase was repeated automatically. Baseline phase. This phase consisted of 70 trials: 32 CS+ trials, 32 CS2 trials, 3 catch trials, and 3 digit trials. Acquisition phase. Before the start of this phase, information about picture presentation was presented on the screen. Participants were informed that either a threatening or a neutral picture could follow their responses but that they had to continue the task exactly as before. It was stressed that the pictures would be presented randomly and not as a punishment for mistakes or any other reason. Participants were not informed about
the masked CSs or about the CS2US contingencies. The acquisition phase consisted of 146 trials. At first, 6 buffer trials were presented, of which 3 were CS+ trials followed by a threatening picture (US), and 3 were CS2 trials followed by a neutral picture (control stimulus). Thereafter, participants were presented with 140 trials including 64 CS+ and 64 CS2 trials, 6 catch trials, and 6 digit trials. We used a 2:1 reinforcement rate, which means that one of the USs followed the participants’ responses on half of the CS+ trials and that one of the control stimuli followed their responses on half of the CS2 trials. Awareness measurement. Contingency awareness. Immediately after the SCT, participants were asked verbally (a) whether they had noticed anything particular about the experiment; (b) whether they had seen anything particular about the neutral face; (c) whether they had noticed any pictures preceding the neutral face. If participants answered yes to any of these questions, they were asked to explain what exactly they had noticed. Their answers were written down by the experimenter. Perceptual awareness. Subsequently, participants performed a 40-trial forced-choice discrimination task. Apart from target presence, trials in this task were identical to the SCT trials. Participants were informed that two male angry faces preceded the presentation of the neutral mask. They were verbally instructed to differentiate between these two angry faces based on any criterion. Examples of such criteria (shades, flashes, . . . ) were given to all participants. Participants used two response buttons on the keyboard (keys: “a”/“1”), one for each angry face. Because the angry faces were presented subliminally, we could not assign a response button to a specific angry face beforehand. Participants were asked to consistently press the same button for the same angry face, or any criterion associated with this face. It was stressed that they should not focus on discriminating angry faces, but rather on discriminating low-level features or slight visual differences in colour or shade. No
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
2341
RAES ET AL.
USs or control stimuli for the USs were presented during this task.
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
Subjective CS and US ratings. To obtain subjective valence ratings of the CSs and the USs, participants rated the USs and US control stimuli and the CS and mask faces on anchored 10-point rating scales ranging from 0 (not aversive at all) to 100 (very aversive). All participants completed the ratings of the USs and the US control stimuli. However, only 29 out of the 34 remaining participants completed the CS and mask ratings. Data-analytic strategy Awareness data. Participants were considered as contingency aware if they could verbalize the CS2US contingencies on the CA measurement. Participants were classified as perceptually aware if they performed significantly better or worse than chance level on the perceptual discrimination task. This corresponded to ≤14 or ≥26 correct responses on the 40-trial task (a2 ,2.05 according to the binomial distribution). Response times on the SCT. To examine whether spatial cueing effects were modulated by subliminal conditioning, a 2 (phase: baseline, acquisition)2 × 2 (signal: CS+, CS2) × 2 (cueing: valid, invalid) analysis of variance (ANOVA) with repeated measures was conducted upon response times (RTs).
Results Awareness data Contingency awareness. A total of 7 participants (20.5%) reported that they had noticed something particular about the experiment. Of these, 5 stated that they had noticed something about the neutral face. Of these 5, 2 participants reported changes in this face on some trials (e.g., a different nose or the appearance of a beard). A total of 3 participants thought that the neutral face had sometimes shown an angry expression. However, after
further questioning, none of the participants could specify that there were two angry faces involved in the experiment or that the faces had been associated in any way with the presentation of IAPS pictures. Therefore, none of the participants was considered contingency aware. Perceptual awareness. The accuracy level of 2 participants deviated significantly from chance level. Of these participants, 1 scored significantly above chance level (28/40 correct responses), whereas the other participant scored below chance level, with 13 out of 40 correct responses. Both these participants were excluded from the SCT response time analysis. SCT response times Data preparation. Neither practice nor buffer trials were included in the response time analyses. An outlier analysis was performed for each participant. RTs that deviated more than 2.3 standard deviations from the mean were excluded from analysis. Trials with errors (2%) were also omitted. In the acquisition phase, US trials were excluded because of interference with response time measurements1 (as in Raes, De Raedt, Fias, Koster, & Van Damme, 2009). Statistical analyses were performed on 65% of the data. Overall effects. The 2 (phase) × 2 (signal) × 2 (cueing) ANOVA yielded a strong effect of cueing, F(1, 32) ¼ 48.18, p , .001. This illustrated the basic spatial cueing effect with slower responding on invalid trials (M ¼ 330, SD ¼ 42) than on valid trials (M ¼ 295, SD ¼ 33). Also the main effect of signal was significant, F(1, 32) ¼ 7.33, p , .05. This main effect was qualified by a significant Phase × Signal interaction, F(1, 31) ¼ 5.19, p , .05. No other effects reached significance, p . .08. We followed up on the significant Phase × Signal interaction by analysing the effect of signal for each experiment phase separately. During baseline, the effect of signal was not
1 The USs were loaded on the moment that the participants responded to the target. The loading of large bmp files probably caused interference with the registration of RTs on reinforced trials. Therefore, these trials were excluded from further analyses.
2342
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
CONDITIONING AND AWARENESS
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
significant, t , 1. As shown in Table 1, RTs on CS+ and CS2 trials were very similar in this phase. In the acquisition phase, the effect of signal was significant, t(31) ¼ 3.07, p , .005. Participants responded slower on CS+ trials than on CS2 trials (see Table 1). Subjective CS and US ratings The USs were rated as more aversive (M ¼ 66, SD ¼ 13) than the control stimuli (M ¼ 7, SD ¼ 15), t(31) ¼ 15.14, p , .001. There were no differences in the postconditioning aversiveness ratings of the CS+ (M ¼ 62, SD ¼ 18) and CS2 (M ¼ 59, SD ¼ 16), t , 1. Both the CS+ and the CS2 faces were rated as significantly more aversive than the neutral mask face (M ¼ 34, SD ¼ 15), t(31) ¼ 5.36, p , .001, and t(31) ¼ 6.34, p , .001.
Discussion The results of Experiment 1 suggest conditioned modulation of spatial cueing effects in participants who are not aware of the CS2US contingencies. Although several participants reported that they had noticed something about the masking face (possibly features of the masked CSs that were “leaking through” the mask), none of them were aware of the presence of the CS+ and the CS2 faces or able to report the CS2US contingencies. Moreover, the 2 participants whose scores on the
postconditioning PA check deviated significantly from chance level were removed from further analysis. Analysis of the SCT response times revealed that participants responded significantly slower on targets cued by the CS+ than on targets cued by the CS2. This finding is inconsistent with previous studies that manipulated threat in an SCT via conditioning of supraliminal cues (Koster et al., 2004; Van Damme et al., 2006). These studies demonstrated attentional capture by threat, illustrated by faster responses on valid CS+ trials, besides decelerated responding on invalid CS+ trials (Koster et al., 2004; Van Damme et al., 2006). Recent findings of Mogg and colleagues could help explaining this inconsistency. That is, Mogg, Holmes, Garner, and Bradley (2008) demonstrated that the emotionally modified SCT indexes two components of responding to threat: attentional capture by threat and a threatinduced response slowing effect. It is possible that, when using subliminal cues, the first component (i.e., attentional capture by threat) is less prominent because subliminal CS presentations do not add to the spatial localization of the neutral mask. This would imply that only the response-slowing effect remains in subliminal threat cueing. In that sense, it could be that the present SCT results mainly reflect a threatinduced slowing effect on motor responding
Table 1. Mean RTs as a function of phase, cueing, and signal, Experiments 1 and 2 Signal
Experiment Experiment 1
Experiment phase
Cueing
CS+ M (SD)
CS2 M (SD)
Baseline
Valid Invalid Valid Invalid
289 (29) 328 (43) 301 (40) 341 (54)
291 (32) 326 (46) 297 (44) 327 (47)
Valid Invalid Valid Invalid
306 (44) 336 (41) 296 (36) 340 (36)
307 (46) 341 (40) 292 (40) 336 (39)
Acquisition Experiment 2
Baseline Acquisition
Note: RT ¼ response time, in ms. CS ¼ conditioned stimulus. THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
2343
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
RAES ET AL.
(e.g., Algom, Chajut, & Lev, 2004; Azevedo et al., 2005). In contrast, the results of SCT studies using supraliminal threat cues also reflect attentional capture by a visible threat cue that is contrasted with a neutral cue. Most importantly, however, the difference in effect between the present study and previous SCT studies investigating supraliminal conditioning effects (e.g., Koster et al., 2004; Van Damme et al., 2006) seems to indicate that different processes are going on in conditioning with and without CA. Because the results of Experiment 1 provide only preliminary evidence for unaware conditioning effects, subsequent experiments were conducted in an attempt to replicate and extend these findings. In the following experiments, a measurement of US expectancy awareness (EA) was added to the awareness measurement, to establish whether there is an association between the strength of subliminal conditioning and the level of EA.
EXPERIMENT 2 In Experiment 2, several changes were made to the design. First, a different US was used—that is, a loud burst of white noise instead of aversive pictures. A soft white noise served as the US control stimulus. Loud white noise has proven to be a successful US in classical conditioning studies (Armony & Dolan, 2002; Koster et al., 2004). Second, we programmed a larger number of acquisition trials. Third, to prevent US habituation, we adapted the reinforcement rate to a 3:1 ratio. Thus, a loud white noise followed responses on a third of the CS+ trials, and a neutral white noise followed responses on a third of the CS2 trials. Fourth, we added measures of expectancy awareness (EA). Participants rated US expectancy for both CSs for different presentation durations and gave confidence scores for each of these ratings. Last, the instructions for the perceptual awareness measurement were adapted in the sense that participants were now instructed about the preceding conditioning procedure.
2344
Method Participants A total of 24 undergraduate students participated in this study. All were tested individually. A total of 2 participants were not included in further data analysis based on high error rates (error rates deviating more than 2.3 standard deviations from the group mean). Mean age of the remaining participants was 19.09 years (SD ¼ 2.20). A total of 6 participants (27.3%) were male. Stimuli The US consisted of a 170-ms 100-dB(A) white noise. The US control stimulus was a 170-ms 67-dB(A) white noise. Both these stimuli were presented binaurally via a Sony MDR-XD100 headphone. Procedure Participants were seated approximately 50 cm from the screen. Head position was stabilized by a chin support device. The experiment started with one presentation of the US (loud white noise). Participants could decide to stop the experiment after this if they felt uncomfortable. After the confrontation with the US, participants gave their informed consent. Subsequently, the SCT started. The spatial cueing task. General information and trial description. In Experiment 1, USs were presented at the participant’s response on the target stimulus. As such, only nonreinforced acquisition trials could be analysed. In order to be able to analyse all trials, all USs in Experiment 2 were presented 600 ms after target onset. Participants’ responses that exceeded 600 ms were excluded as outliers (0.05% of all trials), because noise presentation might have interfered with execution or registration of RTs on these occasions. The target itself was presented for 300 ms instead of 600 ms. Another modification in comparison with Experiment 1 concerned the blank screen presentation at the beginning of each trial, which was presented for a variable interval of 200 – 1,200 ms.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
CONDITIONING AND AWARENESS
Practice phase. This phase was similar to the practice phase in Experiment 1.
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
Baseline phase. This phase consisted of 106 trials, including 10 catch trials and 6 digit trials. The remaining 45 CS+ and 45 CS2 trials were further analysed in order to detect spatial cueing effects. Acquisition phase. Information about noise presentation was presented on the screen: Participants were informed that either loud or soft white noises could follow their responses but that they had to continue the task as before. It was stressed that these auditory stimuli were presented randomly and not as a punishment for mistakes or any other reason. The acquisition phase consisted of 222 trials. At first, 6 buffer trials were presented, 4 of which were CS+ trials followed by white noise to optimize conditions for acquisition. Next, 216 trials, including approximately 10% catch trials (at random), were presented. A digit trial preceded the fixation cross on 1 out of 15 trials. Awareness measurement. Contingency awareness. Immediately after conditioning, participants were asked the same questions as those in Experiment 1 (i.e., whether they had noticed anything particular about the experiment or the neutral face; whether they had seen anything preceding the neutral face, and what in particular they had seen). However, this time these questions appeared on the computer screen, and participants had to answer “yes” or “no” by pressing one out of two response box keys. Participants who answered “yes” on the last question were asked to elaborate further on this answer by describing in detail what they had seen to the experimenter . Expectancy awareness. Both the CS+ and the CS2 were presented on the computer screen three times in fully randomized order: once for 16.7 ms (masked by the neutral CS face), once for 500 ms, and once for 5,000 ms. For each of these six presentations, participants provided US expectancy
ratings using paper and pencil. These ratings were indexed on a 10-point anchored rating scale ranging from 0 (I do not expect the loud white noise at all) to 100 (I very strongly expect the loud white noise). All US expectancy ratings were accompanied by confidence ratings, which were also indexed on a 10-point anchored ratings scale ranging from 0 (I feel very uncertain about my answer) to 100 (I feel very certain about my answer). Perceptual awareness. The forced-choice task was similar to that of Experiment 1 except that participants were informed about the conditioning and masking procedure before they started the task. Each participant completed two forcedchoice tasks of 60 trials. The first task was identical to that of Experiment 1: Participants were asked to discriminate between the CSs based on any possible criterion. Before the start of the second forced-choice task, participants were handed cardboard pictures of the CSs so they could picture both the CS+ and the CS2 stimuli. On both tasks, they were instructed to respond to the CS+ with one of two response buttons and to the CS2 with another response button. Subjective CS and US ratings. Participants rated the auditory stimuli on aversiveness and painfulness. These were paper-and-pencil ratings, using 10point anchored ratings scales ranging from 0 (not aversive at all; not painful at all) to 100 (very aversive; very painful). They also rated the valence of the CS+, CS2, and mask faces on similar anchored ratings scales. The CSs were presented on the computer screen for three different durations (16.7 ms + masked, 500 ms, and 5,000 ms). The mask face was presented twice (500 ms and 5,000 ms). Order of presentations (stimulus type and presentation duration) was randomized. Participants completed these eight ratings using paper and pencil. Data-analytic strategy SCT response times and data on contingency and perceptual awareness were analysed similarly to those in Experiment 1.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
2345
RAES ET AL.
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
Expectancy awareness. Participants were considered expectancy aware if their US expectancy ratings for the CS+ were higher than their ratings for the CS2. Data were analysed separately for the three different presentation durations of the CSs. Participants were categorized as either aware or unaware for each of these durations. Associations between EA and SCT response patterns. Additional analyses were conducted on the SCT response patterns, using expectancy awareness (EA) as a dichotomous between-subject variable. This 2 (phase: baseline, acquisition) × 2 (signal: CS+, CS2) × 2 (cueing: valid, invalid) × 2 (EA: expectancy aware, expectancy unaware) repeated measures ANOVA was performed three times—that is, once for each of the three EA scores (16.7-ms, 500-ms, and 5,000-ms presentations). The only difference between these three analyses was the difference in distribution of participants between expectancy aware and expectancy unaware groups.
Results Awareness data Contingency awareness. A total of 9 participants (40.9%) noticed something particular about the neutral face mask. Of these, 7 had seen changes in this face over the course of the experiment, and 2 reported to have seen another stimulus preceding the neutral face; 1 of the latter 2 participants described this preceding stimulus as a smiling face, and the other person had correctly identified an angry face. Importantly, neither of them was exactly sure of what they had seen, and they had seen the preceding stimulus only once. None of the participants was aware of the presence of two angry faces or was aware of a relationship between the faces and the noises. Therefore, none of the participants was considered contingency aware. Perceptual awareness. Both forced-choice tasks consisted of 60 trials. Following the binomial distribution, all scores equal to or above 37 and scores equal to or below 23 deviate significantly from
2346
chance level (a , .05). On the first forced-choice task (without cardboard pictures), 1 participant scored significantly above chance level (38/60 correct responses), and 1 participant scored significantly below chance level (23/60 correct responses). On the second forced-choice task (with cardboard pictures), 1 participant scored significantly above chance (41/60 correct responses), and another participant scored significantly below chance level (16/60 correct responses). The latter participant was the same participant who scored above chance level on the first task. As such, 3 participants were removed from further data analysis because their performance on the PA measure deviated significantly from chance level. Expectancy awareness. Mean scores across the remaining 19 participants did not show meaningful differences in US expectancy ratings for the CS+ and the CS2 (see Table 2). Frequency analyses show that, for the 16.7-ms CS presentations, 42.1% of the participants reported higher US expectancy for the CS+ than for the CS2. For the 500-ms CS presentations, the percentage of expectancy aware participants was 26.3%. More than a third of participants (36.8%) reported higher US expectancy for the 5,000-ms CS+ presentation than for the 5,000-ms CS2 presentation. Only 1 participant was expectancy aware across all three presentation durations. Confidence ratings with regard to reported US expectancy were similar for the CS+ and the CS2 at all three presentation durations of the CSs. SCT response times Data preparation. RTs deviating more than 2.3 standard deviations from the mean were excluded from analysis. Trials with errors (2%) were omitted. Analyses were performed on 97% of the data. Overall effects. A 2 (phase) × 2 (signal) × 2 (cueing) repeated measures ANOVA was conducted. We found a significant main effect of cueing, F(1, 18) ¼ 73.49, p , .001, indicating slower RTs on invalid trials (M ¼ 338, SD ¼ 35) than on valid trials (M ¼ 300, SD ¼ 40). The Phase × Cueing interaction effect was
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
CONDITIONING AND AWARENESS
Table 2. US expectancy ratings for the CS+ and the CS2 accompanied by their confidence scores as a function of presentation duration, Experiments 2 and 3 Signal
Experiment Experiment 2
Presentation duration (ms) 16.7 500 5,000
Experiment 3
16.7
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
500 5,000
CS+ M (SD)
CS2 M (SD)
t-testa
US expectancy Confidence US expectancy Confidence US expectancy Confidence
49 (26) 49 (20) 60 (27) 63 (18) 55 (28) 65 (23)
47 (30) 56 (20) 59 (26) 58 (23) 52 (30) 59 (27)
,1 1.89 ,1 ,1 ,1 ,1
US expectancy Confidence US expectancy Confidence US expectancy Confidence
45 (20) 41 (19) 63 (26) 48 (24) 55 (28) 44 (26)
47 (21) 43 (23) 53 (21) 43 (27) 43 (22) 45 (27)
,1 ,1 2.36∗ 1.06 2.21∗ ,1
Note: CS ¼ conditioned stimulus. US ¼ unconditioned stimulus. adf ¼ 18 in Experiment 2; df ¼ 21 in Experiment 3. ∗ p , .05.
significant, F(1, 18) ¼ 5.65, p , .05. Also the Phase × Signal interaction effect reached significance, F(1, 18) ¼ 6.75, p , .05. No other main or interaction effects were significant, ps . .19. We followed up on the significant two-way interaction effects. The significant interaction of phase and cueing indicated a stronger cueing effect during acquisition, t(18) ¼ 10.37, p , .001, than during baseline, t(18) ¼ 5.49, p , .001. Follow-up of the significant Phase × Signal interaction showed that the effect of signal was not significant during baseline, t(18) ¼ 1.10, ns. For the acquisition phase, the effect of signal was significant, with slower RTs on CS+ trials than on CS2 trials (see Table 1), t(18) ¼ 2.64, p , .05. EA and SCT response pattern Repeated measures ANOVA. The 2 (phase) × 2 (signal) × 2 (cueing) × 2 (EA) repeated measures ANOVA with expectancy awareness as betweensubjects variable for the 16.7-ms CS presentation yielded a significant Phase × Signal interaction, F(1, 17) ¼ 7.17, p , .05. However, the Phase × Signal × EA interaction was not significant, F , 1. Also, no other main or interaction effects involving EA reached significance, ps . .09.
The same analysis was repeated for the 500-ms and 5,000-ms CS presentations. For both analyses, the Phase × Signal interaction effect was significant, Fs . 6.80, but the Phase × Signal × EA interactions failed to reach significance, Fs ,1.15. In neither of these analyses did other main or interaction effects involving the variable EA reach a significant value, ps . .11. Subjective CS and US ratings As expected, participants rated the 100-dB(A) white noise as more aversive (M ¼ 81, SD ¼ 11) than the 67-dB(A) white noise (M ¼ 45, SD ¼ 18), t(18) ¼ 9.98, p , .001. With respect to painfulness, the 100-dB(A) white noise was rated as neutral (M ¼ 54, SD ¼ 23), while the 67-dB(A) white noise was rated as nonpainful (M ¼ 9, SD ¼ 11), t(18) ¼ 9.09, p , .001. With regard to the valence ratings of the faces, 1 participant failed to complete the valence ratings for the 500-ms mask face. There were no significant differences between the CS+ and CS2 valence ratings (see Table 3). Participants rated both CS faces as more negative than the mask face (M ¼ 51, SD ¼ 23) for the 500-ms presentations, t(17) ¼ 2.78, p , .05, and t(17) ¼ 3.42, p , .005. For the 5,000-ms
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
2347
RAES ET AL.
Table 3. Valence ratings for the CS+ and the CS2 as a function of presentation duration, Experiment 2 Signal
Presentation duration (ms)
CS+ M (SD)
CS2 M (SD)
t-testa
16.7 500 5,000
51 (17) 31 (22) 34 (23)
54 (22) 26 (16) 29 (19)
,1 ,1 ,1
more capable than others in picking up this autonomic feedback (Critchley, Wiens, Rotshtein, ¨ hman, & Dolan, 2004; Katkin et al., 2001). It O might be that the EA measurement of Experiment 2 was not sensitive enough to assess these gut feelings. Therefore, a third experiment was performed, in which the design of the EA measurement was slightly adapted to increase its sensitivity to pick up interoceptive information.
Note: CS ¼ conditioned stimulus. adf ¼ 18. ∗ p , .05.
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
EXPERIMENT 3 presentations, only the CS2 face was rated as more negative than the mask face (M ¼ 44, SD ¼ 22), t(18) ¼ 2.24, p , .05. The difference between the CS+ face and the mask face was not significant, t(18) ¼ 1.30, ns. There were no differences between the valence ratings of the mask face on the one hand and the masked (16.7-ms) CS presentations on the other hand, ts , 1.07.
Discussion In line with Experiment 1, it is demonstrated that unaware conditioning can produce a slowdown of responding on CS+ trials as compared to CS2 trials in an emotionally modified SCT. Consistent with Experiment 1, none of the participants was contingency aware. In addition, apart from 3 participants, none of the participants was perceptually aware. A new feature of this experiment in comparison with Experiment 1 is that participants were asked to report US expectancies for both CSs at three different presentation durations. For each of these durations, some of the participants reported a higher US expectancy for the CS+ than for the CS2. In that sense, some participants could be considered expectancy aware. We were interested whether the level of EA would be associated with the conditioning effects on the SCT. However, the results did not show such association. Previous research has demonstrated that EA can stem from unconscious information that is made available through interoceptive cues or “gut feelings”. Importantly, some individuals seem
2348
In this experiment several adaptations were made in comparison to Experiment 2. First, to strengthen the aversiveness of conditioning, a 104-dB(A) white noise was employed instead of a 100-dB(A) white noise. Second, the instructions preceding the EA measurement were adapted. To maximize the sensitivity of this measurement, participants were instructed about the aim and design of the experiment before the US expectancy ratings. In addition, they were encouraged to assess US expectancy based on their intuition. Furthermore, to increase their motivation, they were told that many participants before them had made a correct judgement on the US expectancy questions through listening to their gut feeling about the stimuli. Third, we added a measurement of EA in the form of a forced-choice task. This task strongly resembled the forced-choice task used to index PA. However, in the EA version of the task, participants were instructed to indicate on each trial whether they expected the aversive 104-dB(A) white noise or not. Before the start of this task, participants were asked again to focus on their intuition while responding to the masked cue presentations.
Method Participants A total of 25 undergraduate students took part in this study in exchange for money or credits. Mean age of the participants was 21.12 years (SD ¼ 2.09). Most participants were female (80%).
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
CONDITIONING AND AWARENESS
Stimuli The US was a 104-dB(A) white noise. In all other aspects, the same stimuli as those in Experiment 2 were used.
of responses (yes/no) to one of two particular keyboard buttons (keys: “a”/“1”) was counterbalanced across participants. Participants were asked to keep on the headphones, but no USs were presented.
Procedure Participants were seated approximately 50 cm from the screen. No chin support device was used in this experiment, but participants were asked not to change position during the experiment.
Perceptual awareness. Participants completed a 60-trial forced-choice task, similar to those in the previous experiments. Participants were handed over cardboard pictures of the CSs, to get an image of what the CSs looked like. They were instructed to press one particular keyboard button (“1” or “a”) when they perceived the CS+ face and another keyboard button (“1” or “a”) when they perceived the CS2 face. Allocation of responses to the two keyboard buttons was counterbalanced across participants.
The spatial cueing task. General information and trial description. Digit trials were adapted. In the previous experiments, RTs to digit trials had not been registered. In the current experiment, participants were instructed to respond to this digit as fast as possible by pressing a centrally situated response box key. Responses to digit trials were analysed separately. Participants failed to respond to these digits on 12.4% of all digit trials. The mean RT on digit trials was 695 (SD ¼ 85). Awareness measurement. Contingency awareness. The procedure was similar to that of Experiment 2. Expectancy awareness. The measurement of EA was preceded by an explanation about the aims and the design of the study. Participants were informed about the subliminal conditioning procedure and were presented with the pictures of the CSs. Also, participants were encouraged to complete the EA ratings based on their initial feeling about the stimuli. Participants were told that many participants before them had been correct about the US expectancies when listening to their gut feelings, whereas it was impossible to answer these questions based on explicit knowledge. After these instructions, participants completed US expectancy ratings and confidence ratings for the CSs for three presentation durations (cf. Experiment 2). Subsequently, participants performed a 108trial forced-choice task. In this task, participants were instructed to indicate US expectancy on each trial by giving yes/no responses. Allocation
Subjective CS and US ratings. This part of the procedure was similar to that in Experiment 2. Data-analytic strategy Awareness data. The data of the forced-choice measure of EA were analysed using detection sensitivity measures (d ′ ). This measure is based on the ability of participants to distinguish signal from noise. In the present EA task, trials containing the CS+ were regarded as signal trials, and trials containing the CS2 as noise trials. The d ′ scores are calculated by subtracting the z score that corresponds to the (signal) false-alarm rate from the z score that corresponds to the (signal) hit rate. Larger values indicate greater sensitivity. In the case of our forced-choice EA task, correct identifications of the CS+ (participants indicate that they expect the US while the CS+ is presented) were regarded as hits, whereas false identifications of the CS+ (participants indicate that they expect the US while the CS2 is presented) were regarded as false alarms. Associations between different levels of EA. As explained in the “Procedure” section above, the measurement of EA included US expectancy ratings for both CSs (at three different presentation durations) on the one hand and a forcedchoice task with subliminal CS presentations on the other hand. In order to perform correlational
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
2349
RAES ET AL.
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
analyses on these measures, we calculated difference scores for each of these measures. For the US expectancy ratings, difference scores were calculated by subtracting the US expectancy ratings for the CS2 from the US expectancy ratings for the CS+ for each of the three presentation durations. For the forced-choice task on EA, the d ′ values were used. Pearson correlations were computed to investigate whether the different measures of EA were associated. Associations between EA and PA. We calculated Pearson correlations to investigate whether measures of EA and PA were associated. Correlations were calculated between each of the four EA measures (three difference scores for the ratings, one score for the forced-choice EA measure) and the PA measure (accuracy level). Associations between EA and SCT response patterns. Consistent with Experiment 2, repeated measures ANOVAs were performed with EA as a dichotomous between-subjects variable. In addition, a repeated measures ANOVA with phase, signal, and cueing as within-subjects variables and the d ′ values of the forced-choice EA measurement as between-subjects variable was performed.
Results Awareness data Contingency awareness. A total of 2 participants reported having noticed something particular about the neutral mask face. Of these, 1 reported having seen different faces, but could not further describe these faces. The other participant had seen the neutral face change throughout the task in the sense that it had sometimes looked like an angry face. However, none of the participants reported having seen faces preceding the neutral face mask, or was able to report the contingencies. Perceptual awareness. The scores of 3 participants deviated significantly from chance level; 1 participant scored significantly above chance level (37/60 correct responses), whereas 2 participants scored
2350
significantly below chance level (22/60 and 15/ 60 correct responses, respectively). These participants were removed from further analysis. Expectancy awareness. Table 2 summarizes the US expectancy ratings of the remaining 22 participants for both CSs at 16.7-ms (masked), 500-ms, and 5,000-ms presentations. For the 16.7-ms masked presentations, there were no significant differences between the CS+ and the CS2. However, for both the 500-ms and the 5,000-ms presentations, US expectancy ratings were significantly higher for the CS+ than for the CS2 (see Table 2). For the 500-ms presentation duration, 45.5% of the participants reported higher US expectancy for the CS+ than for the CS2. With regard to the 5,000-ms CS presentations, this was the case for 63.6% of the participants. The confidence ratings were not significantly different for the CS+ and the CS2. The forced-choice EA measure, in which 16.7ms (masked) CS presentations were used, were analysed using d ′ measures. At the group level, we found a mean sensitivity score of – 0.02 (SD ¼ 0.38), which is not significantly different from zero, t , 1. Associations between different levels of EA. Correlational analyses showed that there were no significant associations between any of the EA ratings (16.7-, 500-, and 5,000-ms presentations), ps . .37. The d ′ values of the forced-choice EA measure did not correlate significantly with the 500-ms, r ¼ .07, ns, or 5,000-ms EA ratings, r ¼ .14, ns. However, there was a significant positive correlation between the forced-choice EA measure and the 16.7-ms (masked) EA ratings, r ¼ .53, p , .01. This implies that participants who showed higher EA on the subliminal ratings also showed higher EA on the forced-choice subliminal measure, and vice versa. Associations between EA and PA. There were no significant correlations between the PA measure and the EA ratings at the three presentation durations of 16.7 ms, r ¼ – .39, ns, p , .08; 500 ms, r ¼ – .39, p ¼ .07; and 5,000 ms, r ¼ – .25, ns.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
CONDITIONING AND AWARENESS
However, the correlation between the PA accuracy score and the forced-choice EA measure did reach significance, r ¼ – .44, p , .05. The negative association implies that participants attaining higher accuracy on the PA measure showed less EA than participants with low PA.
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
SCT response times Data preparation. Consistent with the previous experiments, RTs deviating more than 2.3 standard deviations from the mean were excluded from analysis. Trials with errors (1%) were omitted. Analyses were performed on 98% of the data. Overall effects. The 2 (phase) × 2 (signal) × 2 (cueing) repeated measures ANOVA revealed a significant main effect of cueing, F(1, 21) ¼ 54.90, p , .001, with slower RTs on invalid trials (M ¼ 312, SD ¼ 31) than on valid trials (M ¼ 287, SD ¼ 28). However, no other main or interaction effects reached significance, ps . .14. EA and SCT response pattern Repeated measures ANOVA. We conducted a 2 (phase) × 2 (signal) × 2 (cueing) × 2 (EA) repeated measures ANOVA with EA as between-subjects variable for each of the CS presentation durations. For the subliminal (16.7ms) presentation duration, there were no significant main or interaction effects involving EA, ps . .20.
For the 500-ms presentations, the Phase × Signal × EA interaction was nearly significant, F(1, 20) ¼ 4.19, p ¼ .05. No other main or interaction effects involving EA reached a significant value, ps . .15. Due to its relevance to our research question, we corroborated the three-way interaction effect by investigating the effects per experiment phase. For the baseline phase, there was neither a significant effect of signal, F , 1, nor a significant interaction of signal and EA, F , 1.03. For the acquisition phase, the main effect of signal was not significant, F , 1, but the Signal × EA interaction was significant, F(1, 20) ¼ 4.45, p , .05. Further analysis of the expectancy aware (N ¼ 10) and unaware (N ¼ 12) participants separately showed that the unaware participants did not show an effect of signal during acquisition, t(11) ¼ 1.07, ns. In contrast, the expectancy aware participants showed slower responses for the CS+ than for the CS2 during the acquisition phase, t(9) ¼ 2.45, p , .05 (see Table 4). For the 5,000-ms presentation, there were no significant interactions with EA, ps . .13. Also the forced-choice measure of EA did not show meaningful interactions with the SCT response pattern, ps . .18. Subjective CS and US ratings The 104-dB(A) white noise was rated as more aversive (M ¼ 90, SD ¼ 10) than the 67-dB(A)
Table 4. Mean RTs as a function of phase, signal, and expectancy awareness, Experiment 3 Signal
Expectancy awareness
Experiment phase
Cueing
CS+ M (SD)
CS2 M (SD)
Expectancy aware
Baseline
Valid Invalid Valid Invalid
290 (29) 314 (28) 285 (29) 316 (30)
291 (26) 315 (32) 284 (27) 307 (28)
Valid Invalid Valid Invalid
291 (36) 320 (46) 282 (26) 306 (37)
292 (40) 312 (32) 284 (31) 311 (41)
Acquisition Expectancy unaware
Baseline Acquisition
Note: RT ¼ response time, in ms. CS ¼ conditioned stimulus. THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
2351
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
RAES ET AL.
white noise (M ¼ 50, SD ¼ 17), t(21) ¼ 9.06, p , .001. It was also rated as more painful (M ¼ 55, SD ¼ 26), than the 67-dB(A) white noise (M ¼ 10, SD ¼ 18), t(22) ¼ 7.66, p , .001. The expectancy aware and unaware participants (based on the EA scores for the 500 ms-CS presentations) did not differ in their aversiveness and pain ratings of the US and US control stimulus, as indicated by a 2 (scale: aversiveness, painfulness) × 2 (signal: CS+, CS2) × 2 (EA: expectancy aware, expectancy unaware) ANOVA with EA as between-subjects variable, F , 1. With regard to valence, no significant differences between the CSs were detected, ps . .21. The 500-ms mask face was rated as more neutral in valence (M ¼ 58, SD ¼ 12) than both the CS+ (M ¼ 27, SD ¼ 17), t(21) ¼ 6.58, p , .001, and the CS2 (M ¼ 31, SD ¼ 16), t(21) ¼ 5.25, p , .001. The same was true for the 5,000ms presentations, with the mask face (M ¼ 60, SD ¼ 19) attaining a higher valence rating than the CS+ (M ¼ 30, SD ¼ 17), t(21) ¼ 4.43, p , .001, and the CS2 (M ¼ 34, SD ¼ 14), t(21) ¼ 4.58, p , .001. The valence ratings for the 16.7ms (masked) CS presentations were lower than those for the 500-ms mask face, although the differences were not significant, t(21) ¼ 1.82, p , .09 (CS+: M ¼ 54, SD ¼ 8), and t(21) ¼ 1.71, ns (CS2: M ¼ 53, SD ¼ 12). As reported in Table 5, there were significant differences in the valence ratings of the expectancy aware and unaware participants (EA defined for
the 500-ms CS ratings). A 3 (presentation duration: 16.7 ms, 500 ms, 5,000 ms) × 2 (signal: CS+, CS2) × 2 (EA: expectancy aware, expectancy unaware) ANOVA with EA as between-subjects variable revealed a marginally significant Signal × EA interaction, F(3, 18) ¼ 2.83, p , .07. Univariate tests show that the between-group difference was not significant for the valence ratings of the 16.7-ms CS presentations, F , 1, but was significant for both the 500-ms, F(1, 20) ¼ 7.33, p , .05, and the 5,000-ms CS presentations, F(1, 20) ¼ 4.61, p , .05. As shown in Table 5, the expectancy aware participants rated the CS+ as more negative than the CS2, whereas there was no difference between the CS+ and CS2 valence ratings in the expectancy unaware participants.
Discussion Unexpectedly, the overall results of Experiment 3 did not show significant subliminal conditioning effects on the SCT. In the current experiment, conditioned responding seemed to depend on EA. The present experiment provides the first evidence that subliminal conditioning effects might be associated with participants’ level of EA. Katkin et al. (2001) previously found that EA was associated with participants’ level of interoceptive awareness, but they did not report the association between EA and the strength of
Table 5. CS valence ratings for the CS+ and the CS2 as a function of presentation duration and expectancy awareness, Experiment 3 Signal
Presentation duration (ms)
CS+ M (SD)
CS2 M (SD)
t-testa
Expectancy aware
16.7 500 5,000
50 (8) 25 (10) 28 (13)
51 (13) 40 (15) 39 (15)
,1 2.42∗ 2.28∗
Expectancy unaware
16.7 500 5,000
57 (8) 29 (22) 32 (20)
54 (12) 24 (14) 29 (12)
,1 1.15 ,1
Expectancy awareness
Note: CS ¼ conditioned stimulus. adf ¼ 9 in expectancy aware participants; df ¼ 11 in expectancy unaware participants. ∗ p , .05.
2352
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
CONDITIONING AND AWARENESS
conditioned responding in their subliminal conditioning paradigm. In this study, only expectancy aware participants showed significant conditioning effects on the SCT, whereas expectancy unaware participants did not. However, this result is preliminary, given the small sample size and the fact that it was restricted to EA as assessed with 500-ms presentations of the CSs. All participants included in the SCT data analysis were contingency and perceptually unaware. Moreover, the association between the level of PA and the level of EA, as indexed by the ratings at 500-ms CS presentations, was not significant and, if anything, in a negative direction. Therefore, it seems that, although none of the participants was aware of the conditioned stimuli or of the conditioning procedure, some participants acquired awareness of the effect of conditioning. Interestingly, these participants showed conditioned responding, whereas the others did not. This could imply that only those participants who are responsive to subliminal conditioning develop some sense of (interoceptive) awareness, through which they respond “correctly” on EA measure¨ hman & Soares, 1998). ments (cf. O
GENERAL DISCUSSION The notion that aversive conditioning effects can be acquired in the absence of contingency awareness is highly debated (Lovibond & Shanks, 2002; Mitchell et al., 2009). The present study reconsidered this question using an implicit behavioural measure to index conditioned responding. Spatial cueing effects were measured as participants were exposed to CS2US pairings using subliminal CS presentations. Contingency as well as perceptual awareness of the CSs was checked through subjective ratings and various forced-choice discrimination tasks. Three experiments were performed. In Experiments 2 and 3, a measurement of expectancy awareness (EA) was added to the postconditioning awareness measurement. The results of Experiments 1 and 2 are in line with the notion that aversive conditioning can
develop without awareness. In both experiments, unperceived CS2US pairings affected conditioned responding on a spatial cueing task (SCT). This finding is inconsistent with many previous studies that failed to demonstrate unaware conditioning effects (e.g., Dawson et al., 2007; Purkis & Lipp, 2001). However, this inconsistency might be explained by the fact that unaware conditioning effects depend strongly on the outcome measure (Klucken et al., 2009) and ¨ hman & Mineka, 2001) that is the type of CS (O used. That is, we used an implicit behavioural measure to index conditioned responding to fearrelevant CSs, whereas many former studies used neutral CSs and/or psychophysiological outcome measures (e.g., skin conductance responses) that seem to be especially sensitive to propositional knowledge (e.g., Klucken et al., 2009; Tabbert, Stark, Krisch, & Vaitl, 2006). It is interesting to note that the nature of the conditioning effects in the present study differs from that of studies investigating supraliminal conditioning with the SCT. In the present study, participants responded slower on CS+ trials than on CS2 trials. Studies using the SCT in the context of supraliminal conditioning demonstrated attentional capture (i.e., relatively faster responses) on valid CS+ trials and difficulties to disengage (i.e., relatively slower responses) on invalid CS+ trials in comparison with valid and invalid CS2 trials (Koster et al., 2004; Van Damme et al., 2006). Importantly, former research has demonstrated that within-task results for the processing of subliminal and supraliminal threat stimuli can be very distinctive (Egloff & Hock, 2003). In that sense, the inconsistency with previous supraliminal studies (Koster et al., 2004; Van Damme et al., 2006) corroborates that the present results truly reflect subliminal processes. Unexpectedly, the results of the first two experiments could not be replicated in Experiment 3. In this experiment, there were no significant aversive conditioning effects at the group level. However, we found an association between the level of EA and conditioning effects. More specifically, participants who displayed postconditioning EA showed conditioned responding on the SCT,
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
2353
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
RAES ET AL.
whereas participants who were expectancy unaware did not. ¨ hman and These results are in accordance with O Soares’ (1998) suggestion that EA stems from autonomic feedback people receive from the CR they experience when confronted with the CS+. This implies that only participants showing successful conditioning display EA, and that is exactly what we found in Experiment 3. Moreover, these same participants also reported lower valence for the CS+ than for the CS2. As such, this group of participants showed CS+/ CS2 differentiation at three different levels (SCT responding, US expectancy, and CS valence), whereas the expectancy unaware participant did not show differences between the CSs at any of these levels. Importantly, the response pattern of the expectancy aware participants in Experiment 3 is entirely consistent with the conditioned response pattern found in Experiment 1 and Experiment 2 (i.e., slower responses on CS+ than on CS2 trials on the SCT). This corroborates that these participants showed true conditioned responding on the SCT. In addition, we found that the EA and PA measures were, if anything, inversely related. This indicates that our EA measure captures a different aspect of awareness than does the PA measure. Future research is needed to clarify the mechanisms underlying the association between EA and subliminal conditioning. For example, based on ¨ hman and Soares’s (1998) explanation, we O would expect that the association between EA and conditioning would be more pronounced in participants with a high level of interoceptive awareness. Still, alternative explanations are possible. For instance, it could be that a third factor, such as trait anxiety, might explain why some participants are both more sensitive to subliminal conditioning and more prone/able to report EA at supraliminal presentation durations. Several aspects of the present results need further consideration. First, it is difficult to explain why aversive subliminal conditioning was demonstrated at the group level in Experiment 1 and Experiment 2, but not in Experiment 3. One plausible explanation could be related to the fact
2354
that we are looking for a relatively small effect in a limited sample of participants. Variations in the individual characteristics of the participants might have exerted an influence on the results. However, it is important to emphasize that, in Experiment 3, at least a part of the participants showed an aversive subliminal conditioning effect that was highly similar to the effect that was found on group level in Experiments 1 and 2. Second, it should be noted that the effect of EA on conditioning was restricted to Experiment 3 although EA was also indexed in Experiment 2. This might be explained by the difference in the instructions preceding the EA measurement. In Experiment 3, participants were specifically instructed to rely on their initial feeling about the stimuli and not on propositional knowledge about the contingencies. The divergence in results between Experiment 2 and Experiment 3 thus suggests that this “intuition instruction” enhanced the sensitivity of our EA measurement. Third, Experiment 3 included four measures of EA, but the results on these measures were not entirely consistent. At the group level, only the US expectancy ratings for the supraliminal CS presentations revealed a difference between the CS+ and the CS2. Results on the two EA measures involving subliminal CSs (ratings and forced-choice task) were correlated, which indicates that both measures indexed the same construct. However, there was no CS differentiation on these measures at the group level, nor were the results associated with participants’ level of conditioned responding. It thus seems that subliminal CS presentations can exert an effect on the SCT responses, whereas they do not produce differences in EA. It is difficult to explain why EA arises only when the CSs can be detected. According to the results of Katkin et al. (2001), EA is associated with interoceptive awareness, which suggests that participants report higher US expectancies for the CS+ based on the autonomic response it elicits. It could be that this process is time dependent in the sense that participants need some time to become aware of their bodily feedback. This might explain why participants cannot reliably report US expectancies for subliminally presented stimuli.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
CONDITIONING AND AWARENESS
A fourth point that warrants further discussion is related to the previous point. It is remarkable that only one of the four EA measurements (i.e., the EA ratings for the 500-ms CS presentations) was related to the aversive conditioning effect. It seems as if the reliability and validity of the EA measure largely depends on the time frame that is available to register and report EA. Possibly, a time frame of 500 ms offers participants the right amount of time to assess and report their initial feelings about the CSs, which would make this measure the best one to index “true” EA. As discussed previously, it might be that subliminal CS presentations are less suitable because participants need more time to become aware of their initial feelings. On the other hand, a timeframe that is too long (e.g., as in the EA measure for the 5,000-ms CS presentations) might cause participants to doubt their answers more and to rely less on their initial feelings than for the 500-ms CS presentations. The lack of significant correlations between the different EA measures, except for the two measures using subliminal presentations, is consistent with the possibility that these measures do not index the same construct. Future studies should follow up on this matter by investigating whether reports of EA depend on the time frame that is offered. Although several aspects of the present data pattern do not allow straightforward interpretations, the present findings are important in several ways. First, it should be underscored that a basic conditioning effect in the absence of CA was found and replicated. This effect consisted of decelerated responding to CS+ trials compared to CS2 trials. Response slowing to threat is a commonly observed phenomenon (e.g., Algom et al., 2004; Azevedo et al., 2005). The difference with the data pattern derived in supraliminal conditioning studies (e.g., Koster et al., 2004; Van Damme et al., 2006) suggests that difference processes are going on in supraliminal versus subliminal conditioning. The present results are also important for the ongoing discussion on the role of awareness in aversive conditioning. Several authors have postulated that all human learning is propositional (Mitchell
et al., 2009). However, the current findings are difficult to reconcile with a single-system view on learning. It is hard to sustain that propositional processes produced the modulation of spatial cueing effects in the current experiments, given the use of a sensitive discriminatory measure of awareness and subliminal stimulus durations (Milders, Sahraie, & Lohan, 2008). Still, the magnitude of the effects that were found in this study suggests that the effect of nonpropositional processes in human learning is restricted. In that sense, the present results do not challenge the notion that propositional processes exert the largest influence on human learning and behaviour. However, the present results also suggest that aversive conditioning does not necessarily operate in complete independence of awareness. The results of Experiment 3 suggest that although the response to the (subliminal) threat develops in absence of awareness, participants can become aware of the response itself later on (EA). Further research should investigate which propositional processes might come into play after conditioning, for instance with the individual deducing what he is afraid of (CS+) based on interoceptive awareness of the threat response to the CS+. There are several limitations to this study. A first limitation relates to our forced-choice measurement of PA. This task was based on discriminating the CSs, which is a prerequisite for differential conditioning. However, as participants had to discriminate these stimuli by assigning each CS to a different response button, noise might have been introduced. Also, as we are the first to use this task, there is no independent evidence that this task is sensitive to PA. On the other hand, several other aspects of our study indicate that the results of our PA measurement were valid. The absence of EA at the group level in Experiment 3 for the subliminal presentations, in contrast to the supraliminal presentations, suggests that the CSs were truly subliminal. In addition, if the CSs had not been subliminal, we would expect a response pattern on the SCT similar to the results of Koster et al. (2004). A second limitation regards the postexperimental nature of the awareness checks. We cannot
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
2355
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
RAES ET AL.
exclude the possibility that issues of interference or forgetting influenced the results of this assessment (Lovibond & Shanks, 2002). In addition, there was a context shift in the measurement of awareness, as these measurements occurred outside of the SCT. This might have weakened our awareness measurement. However, we relied largely on the perceptual discrimination between the CSs, which can be regarded as a prerequisite for developing CA. It seems unlikely that the postconditioning nature of this awareness measurement influenced the ability of participants to discriminate between the CS+ and the CS2. To conclude, the present study provides evidence for aversive conditioning in the absence of awareness. Although none of the participants were contingency or perceptually aware, some participants reported higher US expectancy for the CS+ than for the CS2 when they were presented with detectable versions of the CSs after conditioning, Moreover, these participants in particular displayed strong subliminal conditioning effects. This finding is intriguing but difficult to explain in a clear-cut manner. Future studies are required to clarify the association between subliminal conditioning effects and expectancy awareness at the supraliminal level. Original manuscript received 08 September 2008 Accepted revision received 29 March 2010 First published online 4 August 2010
REFERENCES Algom, D., Chajut, E., & Lev, S. (2004). A rational look at the emotional Stroop phenomenon: A generic slowdown, not a Stroop effect. Journal of Experimental Psychology: General, 133, 323–338. Armony, J. L., & Dolan, R. J. (2002). Modulation of spatial attention by fear-conditioned stimuli: An event-related fMRI study. Neuropsychologia, 40, 817–826. Azevedo, T. M., Volchan, E., Imbiriba, L. A., Rodrigues, E. C., Oliveira, J. M., Oliveira, L. F., et
2356
al. (2005). A freezing-like posture to pictures of mutilation. Psychophysiology, 42, 255–260. Baeyens, F., Eelen, P., & Crombez, G. (1995). Pavlovian associations are forever: On classical conditioning and extinction. Journal of Psychophysiology, 9, 127–141. Baeyens, F., Eelen, P., & Van den Bergh, O. (1990). Contingency awareness in evaluative conditioning: A case for unaware affective–evaluative learning. Cognition & Emotion, 4, 3–18. Carlson, J. M., & Reinke, K. S. (2008). Masked fearful faces modulate the orienting of covert spatial attention. Emotion, 8, 522–529. Clark, R. E., & Squire, L. R. (1998). Classical conditioning and brain systems: The role of awareness. Science, 280, 77–81. Costen, N. P., Shepherd, J., Ellis, H., & Craw, I. (1994). Masking of faces by facial and non- facial stimuli. Visual Cognition, 1, 227–251. ¨ hman, A., Critchley, H. D., Wiens, S., Rotshtein, P., O & Dolan, R. J. (2004). Neural systems supporting interoceptive awareness. Nature Neuroscience, 7, 189–195. Dawson, M. E., Rissling, A. J., Schell, A. M., & Wilcox, R. (2007). Under what conditions can human affective conditioning occur without contingency awareness? Test of the evaluative conditioning paradigm. Emotion, 7, 755–766. Dawson, M. E., & Schell, A. M. (1985). Information processing and human autonomic classical conditioning. In P. K. Ackles, J. R. Jennings, & M. Coles (Eds.), Advances in psychophysiology (Vol. 1). Greenwich, CT: JAI Press. Dawson, M. E., & Shell, A. M. (1987). Human autonomic and skeletal classical conditioning: The role of conscious cognitive factors. In G. Davey (Ed.), Cognitive processes and Pavlovian conditioning in humans. London: John Wiley & Sons. Egloff, B., & Hock, M. (2003). Assessing attention allocation toward threat-related stimuli: A comparison of the emotional Stroop task and the attentional probe task. Personality and Individual Differences, 35, 475–483. Fox, E., Russo, R., Bowles, R., & Dutton, K. (2001). Do threatening stimuli draw or hold attention in subclinical anxiety? Journal of Experimental Psychology: General, 130, 681–700. Goeleven, E., De Raedt, R., Leyman, L., & Verschuere, B. (2008). The Karolinska Directed Emotional
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
CONDITIONING AND AWARENESS
Faces: A validation study. Cognition and Emotion, 22, 1094–1118. Hamm, A. O., & Vaitl, D. (1996). Affective learning: Awareness and aversion. Psychophysiology, 33, 698–710. Hamm, A. O., & Weike, A. I. (2005). The neuropsychology of fear learning and fear acquisition. International Journal of Psychophysiology, 57, 5–14. ¨ hman, A. (2001). Katkin, E. S., Wiens, S., & O Nonconscious aversive conditioning, visceral perception, and the development of gut feelings. Psychological Science, 12, 366–370. Klucken, T., Kagerer, S., Schweckendiek, J., Tabbert, K., Vaitl, D., & Stark, R. (2009). Neural, electrodermal, and behavioral response patterns in contingency aware and unaware subjects during a picture – picture conditioning paradigm. Neuroscience, 158, 721–731. Koster, E. H. W., Crombez, G., Van Damme, S., Verschuere, B., & De Houwer, J. (2004). Signals for threat modulate attentional capture and holding: Fear-conditioning and extinction during the exogenous cueing task. Emotion, 4, 771–780. Koster, E. H. W., Crombez, G., Verschuere, B., Vanvolsem, P., & De Houwer, J. (2007). A timecourse analysis of attentional cueing by threatening scenes. Experimental Psychology, 54, 161–171. Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1997). International Affective Picture System (IAPS) [CDROM]. Gainesville, FL: NIMH Center for Emotion and Attention (CSEA). LeDoux, J. (2003). The emotional brain, fear, and the amygdala. Cellular and Molecular Neurobiology, 23, 727–738. Lipp, O. V., & Purkis, H. M. (2005). No support for dual process accounts of human affective learning in simple Pavlovian conditioning. Cognition and Emotion, 19, 269–282. Lovibond, P. F., & Shanks, D. R. (2002). The role of awareness in Pavlovian conditioning: Empirical evidence and theoretical implications. Journal of Experimental Psychology: Animal Behavior Processes, 28, 3–26. ¨ hman, A. (1998). The Lundqvist, D., Flykt, A., & O Karolinska Directed Emotional Faces (KDEF). Stockholm: Karolinska Hospital, Department of Neurosciences. Mackintosh, N. J. (1983). Conditioning and associative learning. Oxford, UK: University Press.
Milders, M., Sahraie, A., & Logan, S. (2008). Minimum presentation time for masked facial expression discrimination. Cognition & Emotion, 22, 63–82. Mitchell, C. J., De Houwer, J., & Lovibond, P. F. (2009). The propositional nature of human associative learning. Behavioral and Brain Sciences, 32, 183–246. Mogg, K., Holmes, A., Garner, M., & Bradley, B. P. (2008). Effects of threat cues on attentional shifting, disengagement and response slowing in anxious individuals. Behaviour Research and Therapy, 46, 656–667. Moors, A., & De Houwer, J. (2006). Automaticity: A theoretical and conceptual analysis. Psychological Bulletin, 132, 297–326. ¨ hman, A., & Dolan, R. J. (1998). Morris, J. S., O Conscious and unconscious emotional learning in the human amygdala. Nature, 393, 467–470. ¨ hman, A. (2005). The role of the amygdala in O human fear: Automatic detection of threat. Psychoneuroendocrinology, 30, 953–958. ¨ hman, A., & Mineka, S. (2001). Fears, phobias and O preparedness: Toward an evolved module of fear and fear learning. Psychological Review, 108, 483–522. ¨ hman, A., & Soares, J. J. F. (1998). Emotional O conditioning to masked stimuli: Expectancies for aversive outcomes following nonrecognized fearrelevant stimuli. Journal of Experimental Psychology: General, 127, 69–82. Pavlov, I. P. (1927). Conditioned reflexes. Oxford, UK: University Press. Pleyers, G., Corneille, O., Luminet, O., & Yzerbyt, V. (2007). Aware and (dis)liking: Item-based analyses reveal that valence acquisition via evaluative conditioning emerges only when there is contingency awareness. Journal of Experimental Psychology: Learning, Memory, & Cognition, 33, 130–144. Posner, M. I. (1980). Orientation of attention. Quarterly Journal of Experimental Psychology, 32, 3–25. Purkis, H. M., & Lipp, O. V. (2001). Does affective learning exist in the absence of contingency awareness? Learning and Motivation, 32, 84–99. Raes, A. K., De Raedt, R., Fias, W., Koster, E. H. W., & Van Damme, S. (2009). Does contingency awareness mediate the influence of emotional learning on the cueing of visual attention? Psychological Research– Psychologische Forschung, 73, 107–113.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
2357
RAES ET AL.
Van Damme, S., Crombez, G., Hermans, D., Koster, E. H. W., & Eccleston, C. (2006). The role of extinction and reinstatement in attentional bias to threat: A conditioning approach. Behaviour Research and Therapy, 44, 1555–1563. Wong, P. S., Bernat, E., Snodgrass, M., & Shevrin, H. (2004). Event-related brain correlates of associative learning without awareness. International Journal of Psychophysiology, 53, 217–231. Wong, P. S., Shevrin, H., & Williams, W. J. (1994). Conscious and nonconscious processes: An ERP index of an anticipatory response in a conditioning paradigm using visually masked stimuli. Psychophysiology, 31, 87–101.
Downloaded By: [Van Damme, Stefaan] At: 07:00 30 November 2010
¨ hman, A. (1993). Preattentive procesSoares, J. F., & O sing, preparedness and phobias: Effects of instruction on conditioned electrodermal responses to masked and non-masked fear-relevant stimuli. Behaviour Research and Therapy, 31, 87–95. Stormark, K. M., & Hugdahl, K. (1996). Peripheral cueing of covert spatial attention before and after emotional conditioning of the cue. International Journal of Neuroscience, 86, 225–240. Tabbert, K., Stark, R., Krisch, P., & Vaitl, D. (2006). Dissociation of neural responses and skin conductance reactions during aversive conditioning with and without awareness of stimulus contingencies. NeuroImage, 32, 761–770.
2358
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2010, 63 (12)
