Elsevier

Brain and Cognition

Volume 114, June 2017, Pages 1-10
Brain and Cognition

Raising the alarm: Individual differences in the perceptual awareness of masked facial expressions

https://doi.org/10.1016/j.bandc.2017.01.010Get rights and content

Highlights

  • Backward masking effectiveness differs between subjective and objective measures.

  • Participants can reliably access masked facial expressions of happiness.

  • High trait anxiety enhances threat detection of masked facial expressions of anger.

  • Low trait anxiety enhances perceptual access to masked happy facial expressions.

Abstract

A theoretical concern in addressing the unconscious perception of emotion is the extent to which participants can access experiential properties of masked facial stimuli. Performance on a two alternative forced choice (2AFC) task as a measure of objective awareness was compared with a new measure developed to access experiential phenomena of the target-mask transition, the perceptual contrast-change sensitivity (PCCS) measure in a backward-masking paradigm with angry, happy and neutral facial expressions. Whilst 2AFC performance indicated that the targets were successfully masked, PCCS values were significantly higher in the happy-neutral face condition than in the angry-neutral face and the neutral-neutral face conditions (Experiment 1). Furthermore, objective measures of awareness were more readily displayed by individuals with high trait anxiety, whereas individuals with low trait anxiety showed greater access to the experiential quality of happy faces (Experiment 2). These findings provide important insights into the methodological considerations involved in the study of non-conscious processing of emotions, both with respect to individual differences in anxiety and the extent to which certain expressions can be successfully masked relative to others. Furthermore, our results may be informative to work investigating the neural correlates of conscious versus unconscious perception of emotion.

Introduction

Ubiquitous in our daily lives, emotional experiences are ingrained in the evolutionary process (Al-Shawaf et al., 2016, Ekman, 1992, Ekman and Cordaro, 2011, Nesse and Ellsworth, 2009, Öhman, 2006). Thus firmly established in our neurobiological system, emotions are perceived automatically and unconsciously (Dimberg et al., 2000, Dimberg et al., 2002, Tamietto and de Gelder, 2010, Tracy and Robins, 2008). According to Tamietto and de Gelder (2010), the unconscious perception of emotion has been implicated in a multitude of subcortical brain regions, which can broadly be divided into a network involved in the visual processing of emotional cues and a network centered on non-visual emotion-oriented processes. The former network, as the authors note, includes the substantia innominate, the superior colliculus, the nucleus accumbens, the pulvinar and the amygdala, whereas the latter network involves the basal ganglia, the locus coeruleus, the hypothalamus, the periaqueductal grey, the hippocampus and the nucleus basalis of Meynert. Conscious perception of emotional cues, in turn, usually also extends across the cingulate, occipitotemporal and frontal areas, although such activity can sometimes be found in studies rendering emotional cues unconscious as well, possibly due to direct and indirect links between cortical and subcortical structures (Brooks et al., 2012, Tamietto and de Gelder, 2010). Not surprisingly then, there has been an avid debate concerned with the extent to which brain activation during conscious versus unconscious perception of emotion relies on common or distinct neural substrates (Balconi and Bortolotti, 2013, Jiang and He, 2006, Phillips et al., 2004, Tamietto and de Gelder, 2010, Tamietto et al., 2015, Yang et al., 2012). Part of this debate rests on the crucial assumption that the paradigms used to test unconscious perception of emotion uniquely measure unconscious, but not conscious, perception of emotion.

One of the principal paradigms for the study of unconscious emotions in healthy individuals is the backward-masking procedure (Tamietto & de Gelder, 2010). The backward-masking procedure has made a significant contribution to the amygdala’s status as the brain’s silent ‘alarm’ system, alerting us to impending dangers, such as fearful and angry facial expressions, often with relatively little conscious appraisal on our part (Liddell et al., 2005, Öhman and Mineka, 2001). Procedurally, it entails the brief presentation of a visual stimulus, referred to as the target, followed by the subsequent presentation of another visual stimulus, in the same (or nearby) spatial location, referred to as the mask. Presentation rates of the target stimulus, which is usually a picture of an emotional face such as an angry or fearful expression, are very brief, usually in the order of 30msec or less. The masking stimulus, usually a picture of a neutral face has a slightly longer duration, typically 100msec or more. Whilst participants are often able to report the presence of the mask, they are unable to identify or even detect the presence of the target. Thus, the participant is deemed to be unaware of the target, even though the physiological and neuroimaging changes are observed in the participant during the target’s presentation in the backward-masking task (e.g., Dimberg et al., 2000, Whalen et al., 2004). Furthermore, manipulating the temporal interval between the presentation of the target and mask, most commonly expressed in terms of stimulus onset asynchrony (SOA), appears to play a key role in influencing the detection of emotional targets, with longer SOAs facilitating target detection performance (e.g., Esteves & Öhman, 1993).

Outside of the neuroscientific investigation of the threat detecting capacities of the amygdala, cognitive psychologists have been concerned for quite some time as to whether participants are truly unaware of the masked threatening emotional expression (Maxwell and Davidson, 2004, Milders et al., 2008, Pessoa, 2005, Pessoa et al., 2005). A particular problem in classifying whether or not a participant is unaware of the masked emotional expression partially derives from the criteria used in defining subjective vs objective levels of awareness. According to subjective criteria, awareness is assessed on the basis of participants’ self-reports of their conscious experiences; if participants can report that they have ‘seen’ the target, it is assumed that the item was perceived with awareness, and if the participants report that they have not ‘seen’ the target, it is assumed that they are unaware of the critical (i.e., masked) item (e.g.,Dimberg et al., 2000, Morris et al., 1998, Merikle, 1992, Tsuchiya and Adolphs, 2007). A more sensitive approach over such binary responses (e.g., seen vs. not seen) involves the additional use of confidence ratings (e.g., Esteves and Öhman, 1993, Phillips et al., 2004) during the target/mask pairings to establish when participants become fully conscious or aware of the presence of the target (i.e., ‘extremely confident’).

According to objective criteria, awareness is assessed on the basis of setting performance thresholds, typically measured in a forced-choice decision task. Participants are deemed unaware of the target if they cannot discriminate the presence or absence of a stimulus or categorize the emotionality of the target (e.g., fear vs disgust) with above-chance accuracy (e.g., Liddell et al., 2005, Merikle et al., 2001, Phillips et al., 2004). Objective perception of the target face is assessed using a signal detection framework in which the detection threshold of d = 0 or its nonparametric analogue, A = 0.50 is used as a measure of chance performance (e.g., Hanley and McNeil, 1982, Liddell et al., 2005, Macmillan and Creelman, 1991, Maxwell and Davidson, 2004, Milders et al., 2008, Szczepanowski and Pessoa, 2007). Studies that have utilized A′ measures have reported above-chance detection even at 17msec target presentation times (e.g., Pessoa et al., 2005) thus contrasting with previous findings with longer, yet seemingly ‘unconscious’ thresholds (e.g., Morris et al., 1998, Whalen et al., 1998).

Self-report methodologies that focus on binary ‘seen/unseen’ responses or confidence ratings may not sensitively capture all relevant aspects of participants’ conscious experiences of the backward-masking methodology (e.g., Maxwell and Davidson, 2004, Merikle et al., 2001). Usually, these experiences stem from the perceptual changes during the transition between target and mask, resulting in apparent motion phenomena which are likely to be intensified in some emotional expressions on account of the perceptual discrepancy in the localized facial features. For instance, happy facial expressions are reliably identified from neutral faces (i.e., 70% of raters agree) on the basis of the presence of the bags under eyes, cheeks raised, upper lip raised and exposure of the upper teeth, whilst angry facial expressions are reliably identified by the presence of a pronounced frown (Calvo & Marrero, 2008). Asking participants to explicitly report their experience of such phenomena through questionnaire formats and/or funnel interview techniques can yield important individual differences in detecting the emotionality of masked faces. For example, Maxwell and Davidson (2004) divided their participant pool into those participants who could verbally report the presence of apparent motion (e.g., flickering and movement) in the backward-masking task (explicitly aware) and those who maintained, despite persistent prompting, not to have experienced any apparent motion phenomena (explicitly unaware). The two groups differed in performance in a target identification forced choice procedure, such that the explicitly aware group performed better than the unaware group in identifying happy and neutral targets, whereas the unaware group outperformed their explicitly aware counterparts in the identification of anger. Thus, the contrasting effects in setting subjective vs objective measures of awareness indicates how facial expressions of emotion perceived without awareness can both bias which stimuli are perceived with awareness and influence how stimuli are consciously experienced (e.g., Maxwell & Davidson, 2004).

The perceptual awareness of emotional stimuli may be further modulated by individual differences in trait levels of self-reported anxiety, which can affect an individual’s response to impending situational (i.e., state anxiety) stressors (Eysenck, 1992, Eysenck et al., 2007, Spielberger et al., 1983). Above threshold presentation of facial stimuli tends to trigger greater levels of visual spatial attention towards threatening facial expressions in individuals with high trait anxiety (Byrne and Eysenck, 1995, Mogg and Bradley, 1999). Anxiety-related difficulties also emerge when it comes to disengaging attention away from threat-relevant facial stimuli (Fox et al., 2001, Georgiou et al., 2005) and can also interfere in the processing of task-irrelevant threat distracters (e.g., Damjanovic, Pinkham, Clarke, & Phillips, 2014).

Recent investigations with backwardly masked emotional expressions are also consistent with the view of a finely tuned threat detection mechanism in anxiety, such that high performing participants on fear detection trials are likely to belong to the high end of the trait anxiety continuum (Japee, Crocker, Carver, Pessoa, & Ungerleider, 2009). This bias can emerge as early as 115–145 ms post-stimulus (Li, Zinbarg, Boehm, & Paller, 2008), whilst spatial markers show increases in right amygdala activation which correlates strongly with symptom severity in individuals with generalized anxiety disorder (Monk et al., 2008).

Successful masking of facial expressions poses a complex challenge for researchers. The aim of the current study is to provide a detailed comparison between traditional approaches in determining successful masking, such as above-chance detection rates, and a less explored measure utilizing participants’ subjective experiences of the target-mask transition process. In order not to prime the participants to the emotional content of the targets, we followed Dimberg et al’s (2000) backward masking methodology whereby angry-neutral, happy-neutral and neutral–neutral, target-mask pairings were presented to participants in an independent groups design for 30msec followed by a 5 s neutral face mask. Priming was minimized in Dimberg et al’s (2000) task by not instructing participants to decide whether an emotional target was presented to them during each target-mask trial. This is an important methodological detail to consider in order to arrive at a more accurate evaluation of the role of participant awareness (Pessoa et al., 2005) within the contextual demands of the testing protocol (i.e., instructions, participant’s mode of responding, etc.,). Thus, a more robust argument in favour of an unconscious processing bias towards threat could be established under procedural requirements, which are designed to render participants consciously unaware (Pessoa et al., 2005) of the masked target. Previous research had overlooked this procedural requirement in their methodologies by explicitly priming participants to expect and categorize the emotional expressions of the targets, often on a trial-by-trial basis (e.g., Japee et al., 2009, Li et al., 2008, Maxwell and Davidson, 2004, Milders et al., 2008, Pessoa et al., 2005). We address this important methodological concern in the current study by focusing on the incidental processing of emotional expressions (e.g., Monroe et al., 2013) by concealing our backward-masking procedure within the context of a sex discrimination task. The success of this procedure will be assessed by a 2AFC task to establish an objective outcome measure of the degree of awareness of the masked target faces as indicated by significant levels of above-chance detection performance (e.g., Pessoa et al., 2005).

Experiment 1 reports the outcome of a new, sensitive measure developed in our laboratory, the perceptual contrast-change sensitivity (PCCS) measure, which aims to quantify for the first time participants’ subjective awareness of perceptual phenomena (e.g., flickering and movement) for angry-neutral, happy-neutral and neutral-neutral, target-mask pairings. Earlier work limited participants’ responses to such aspects of the target-mask transition to a yes/no response format (e.g., Dimberg et al., 2000, Maxwell and Davidson, 2004). We addressed this issue in our measure by asking participants to use a discrete response scale and to further quantify the frequency in which they experienced each phenomenon. These responses would then allow for a score to be calculated for each of the target-mask conditions, thus providing more detailed information about the perceptual transition process than that obtained in other studies. In Experiment 2, we apply these measures to an individual differences framework to assess how differences in self-reported trait anxiety (STAI-T) could modulate subjective and objective awareness of the target face.

Based on the perceptual saliency of the characteristics of happy and angry facial expressions and the magnitude of their perceptual differences compared to neutral faces, we hypothesized that the happy target condition would likely trigger greater perceptual phenomena, and in turn yield higher PCCS scores, than in the angry-neutral and neutral-neutral conditions (Maxwell & Davidson, 2004). This is because there are greater localized differences in the facial features between happy and neutral faces, than angry and neutral faces (e.g., Calvo & Marrero, 2008). Furthermore, the intensity of the eye region in negative facial expressions such as fear (Whalen et al., 2004) and anger (Fox & Damjanovic, 2006) may also trigger a perceptual discrepancy in the transition between the angry-neutral mask, which in turn would be higher than baseline levels, as represented by the neutral-neutral condition. This prediction is further supported by the automatic vigilance hypothesis (Pratto & John, 1991), which stipulates an attentional processing bias towards negative information, especially in situations where participants are unaware. We hypothesized that both the angry and happy targets would yield higher PCCS than the baseline condition, as represented by the neutral-neutral condition. Experiment 1 also provides participants with a two alternative forced choice (2AFC) task to establish whether sensitivity to such low-level information as measured by the PCCS is sufficient to lead to objective awareness of the expressive targets (e.g., Pessoa et al., 2005).

In Experiment 2, we assess whether the repeated exposure of angry facial expressions across experimental trials may tune the amygdala into ‘alarm’ mode (Etkin et al., 2004, Liddell et al., 2005) much more readily in high trait anxious participants due to their greater tendency to constantly scan their visual environment for threat (Blanchette and Richards, 2010, Laretzaki et al., 2011) coupled with an inability to incorporate positive feedback relative to low trait anxious participants (Moser, Huppert, Duval, & Simons, 2008). As such, we hypothesized that this heightened threat priming in participants with high levels of self-reported trait anxiety in the current study may result in increased awareness of the angry target, coupled with a decrease awareness of the happy target, relative to the participants with low levels of self-reported trait anxiety. Whether these group differences are more pronounced under different measures of awareness (i.e., subjective vs objective) will be investigated systematically for the first time in the current study as previous attempts to link anxiety mechanisms to unconscious emotion processing have largely been explored on a post-hoc basis (Japee et al., 2009, Szczepanowski and Pessoa, 2007) or confined to ambiguous threat signals such a fearful expressions rather than direct threat signals such as angry expressions (Brosch et al., 2008, Capitão et al., 2014, Ewbank et al., 2009, Japee et al., 2009, Li et al., 2008, Satterthwaite et al., 2011, Whalen, 1998). Thus, it is unclear what role subjective and objective awareness measures play in the detection of other, more direct facial gestures of threat such as anger (Whalen, 1998) and how trait anxiety levels of the observer can impact on such measures.

Our methodological approach of embedding the backward-masking paradigm within a sex discrimination task in order to render the participants consciously unaware (Pessoa et al., 2005) of the target-mask is a step never before taken in cognitive studies of this kind, and should help meet the aims of subjective unawareness of the emotional targets and in doing so advance our understanding of the mechanisms that underpin facial expression processing, as well as the expression processing deficits in special populations where individuals may be particularly high or low in trait anxiety.

Section snippets

Participants

Fifty-four undergraduate and postgraduate students from the University of Essex took part in the study with a mean age of 25.1 years. The sample consisted of 37 females and 17 males. Participants received £6.00 for their participation. All participants had normal to normal-to-corrected vision. Informed consent was obtained from all participants.

Stimuli and apparatus

Angry, happy and neutral expressions were selected from nine different posers from Ekman and Freisen’s (1976) database. Identity across all the

Results and discussion

Mean performance on the objective (2AFC) and subjective measures (PCCS) for the three target-mask conditions are presented in Table 1. The alpha level for all statistical analyses was set at 0.05. For each participant the number of correct and incorrect responses were converted into proportions and treated as hits and false alarms for subsequent signal detection analysis (Macmillan & Creelman, 1991), thus generating an A′ value for each participant as a measure of their sensitivity to the

Participants

One hundred and eight undergraduate and postgraduate students from the University of Essex took part in the study. As per Fox (2002) participants with STAI-T scores 35 (n = 54) or below were recruited for the low anxiety group and participants with STAI-T scores 45 (n = 54) or above were recruited for the high anxiety group. The age range of the entire sample was 20–59, with a mean age of 26.6 years. The sample consisted of 57 females and 51 males. STAI-T scores for the high trait anxiety group (M = 

Results and discussion

The alpha level for statistical analyses was set at 0.05. For the assessment of above-chance detection performance, six separate one-sample t-tests were performed and are reported with a Bonferroni adjusted alpha level of p < 0.008. Post-test scores on the STAI-T for the high anxiety (M = 53.52, SD = 6.57) and low anxiety (M = 30.89, SD = 3.63) groups were significantly different, t(106) = 22.15, p < 0.001, r = 0.91. Mean A′ performance on the 2AFC task for the two anxiety groups is displayed in Fig. 1.

A 2

General discussion

Whilst both objective and subjective measures have been used in the past to determine participants’ levels of awareness of masked facial expressions, these have largely been based on binary yes/no responses from tasks in which participants have been primed to expect or to categorize facial expressions of emotion (e.g., Japee et al., 2009, Li et al., 2008, Maxwell and Davidson, 2004, Milders et al., 2008, Pessoa et al., 2005). This raises the important consideration of whether participants

Conclusions

In conclusion, our study highlights the importance of taking into consideration the experiential properties of the backward-masking procedure, and proposes that subjective measures as revealed by the PCCS should be used in conjunction with objective measures when drawing conclusions about the ‘silent’ threat detecting capacities of the amygdala and other subcortical regions. Crucially, such a combination of objective and subjective outcome measures can usefully contribute to resolving the

Acknowledgments

This project was funded by the Nuffield Foundation, but the views expressed are those of the authors and not necessarily those of the Foundation. The authors wish to acknowledge the assistance of Jodie Halsey for facilitating with pilot data collection.

References (84)

  • B.J. Liddell et al.

    A direct brainstem-amygdala-cortical ‘alarm’ system for subliminal signals of fear

    Neuroimage

    (2005)
  • R.P. Mattick et al.

    Development and validation of measures of social phobia scrutiny fear and social interaction anxiety

    Behaviour Research and Therapy

    (1998)
  • P.M. Merikle et al.

    Perception without awareness: Perspectives from cognitive psychology

    Cognition

    (2001)
  • J.S. Moser et al.

    Face processing biases in social anxiety: An electrophysiological study

    Biological Psychology

    (2008)
  • L. Pessoa

    To what extent are emotional visual stimuli processed without attention and awareness?

    Current Opinion in Neurobiology

    (2005)
  • M.L. Phillips et al.

    Differential neural responses to overt and covert presentations of facial expressions of fear and disgust

    Neuroimage

    (2004)
  • T.D. Satterthwaite et al.

    Opposing amygdala and ventral striatum connectivity during emotion identification

    Brain and Cognition

    (2011)
  • T. Suslow et al.

    Automatic mood-congruent amygdala responses to masked facial expressions in major depression

    Biological Psychiatry

    (2010)
  • M. Tamietto et al.

    Once you feel it, you see it: Insula and sensory-motor contribution to visual awareness for fearful bodies in parietal neglect

    Cortex

    (2015)
  • N. Tsuchiya et al.

    Emotion and consciousness

    Trends in cognitive sciences

    (2007)
  • J. Yang et al.

    The amygdala is involved in affective priming effect for fearful faces

    Brain and Cognition

    (2012)
  • L. Al-Shawaf et al.

    Human emotions: An evolutionary psychological perspective

    Emotion Review

    (2016)
  • N. Amir et al.

    Automatic activation and strategic avoidance of threat relevant information in social phobia

    Journal of Abnormal Psychology

    (1998)
  • A. Bados et al.

    The state-trait anxiety inventory, trait version: does it really measure anxiety?

    Journal of Personality Assessment

    (2010)
  • M. Balsamo et al.

    The state-trait anxiety inventory: Shadows and lights on its construct validity

    Journal of Psychopathology and Behavioral Assessment

    (2013)
  • I. Blanchette et al.

    The influence of affect on higher level cognition: A review of research on interpretation, judgement, decision making and reasoning

    Cognition & Emotion

    (2010)
  • T. Brosch et al.

    Beyond fear: Rapid spatial orienting toward positive emotional stimuli

    Psychological Science

    (2008)
  • A. Byrne et al.

    Trait anxiety, anxious mood, and threat detection

    Cognition and Emotion

    (1995)
  • M.G. Calvo et al.

    Visual search of emotional faces: The role of affective content and featural distinctiveness

    Cognition and Emotion

    (2008)
  • L.P. Capitão et al.

    Anxiety increases breakthrough of threat stimuli in continuous flash suppression

    Emotion

    (2014)
  • C.S. Carver et al.

    Behavioral inhibition, behavioural activation, and affective responses to impending reward and punishment – The BIS/BAS scales

    Journal of Personality and Social Psychology

    (1994)
  • L. Damjanovic et al.

    Enhanced threat detection in experienced riot police officers: Cognitive evidence from the face-in-the-crowd effect

    The Quarterly Journal of Experimental Psychology

    (2014)
  • L. Damjanovic et al.

    Searching for happiness across cultures

    Journal of Cognition and Culture

    (2010)
  • U. Dimberg et al.

    Unconscious facial reactions to emotional facial expressions

    Psychological Science

    (2000)
  • U. Dimberg et al.

    Facial reactions to emotional stimuli: Automatically controlled emotional responses

    Cognition & Emotion

    (2002)
  • W. Donaldson

    The role of decision processes in remembering and knowing

    Memory & Cognition

    (1996)
  • J.D. Eastwood et al.

    Individuals with social phobia are biased to become aware of negative faces

    Visual Cognition

    (2005)
  • P. Ekman

    An argument for basic emotions

    Cognition & Emotion

    (1992)
  • P. Ekman et al.

    What is meant by calling emotions basic

    Emotion Review

    (2011)
  • P. Ekman et al.

    Pictures of facial affect

    (1976)
  • F. Esteves et al.

    Masking the face: Recognition of emotional facial expressions as a function of the parameters of backward masking

    Scandinavian Journal of Psychology

    (1993)
  • M.W. Eysenck

    Anxiety: The cognitive perspective

    (1992)
  • Cited by (0)

    View full text