Elsevier

Acta Psychologica

Volume 163, January 2016, Pages 124-134
Acta Psychologica

When do fearful faces override inhibition of return?

https://doi.org/10.1016/j.actpsy.2015.11.002Get rights and content

Highlights

  • Neutral and fearful target faces were presented in exogenous spatial cueing tasks.

  • Fear had no impact on IOR in localization and face/non-face discrimination tasks.

  • IOR was delayed for fearful faces in emotion and gender discrimination tasks.

  • The impact of fear on IOR depends on the depth of facial processing.

Abstract

Inhibition of return (IOR) occurs when more than about 300 ms elapses between the cue and the target in atypical peripheral cueing task: reaction times (RTs) become longer when the cue and target locations are the same versus different. IOR could serve the adaptive role of optimizing visual search by discouraging the re-inspection of previously attended locations. As such, IOR should not reduce our chances of noticing relevant event information and emotional stimuli, in particular. However, previous studies have led to inconsistent results. The present study offers a systematic investigation of the conditions under which target fearful faces can modulate either the magnitude or the time course of the IOR effect. Notably, we manipulated the depth of facial processing required to perform the task and/or the task relevance of the facial expressions. When participants localized target faces (Experiment 1) or discriminated them from non-face stimuli (Experiment 2), their emotional expression had no impact on IOR whatsoever. However, IOR occurred later for fearful versus neutral faces when the participants performed emotion (Experiment 3) or gender (Experiment 4) discrimination tasks. These findings are discussed with regard to the mechanisms responsible for IOR and to the processing of emotional facial expressions.

Introduction

Our visual environment is prodigiously rich and our processing capacities regrettably limited. Different attentional mechanisms are needed to select which information will undergo elaborate processing and access consciousness. Even if visually salient stimuli are in general particularly prone to capture attention, higher order processes (e.g., the expectancies or intentions of the observer) can modulate this effect (e.g., Folk, Remington, & Johnston, 1992). Moreover, the emotional nature of the competing information may also weigh in the contest. Indeed, stimuli which could have an impact on the observer's well-being or survival should in principle be subject to rapid and efficient selection. The present study is part of a recent and active effort aimed at understanding how these so-called bottom-up, top-down and emotional factors interact to promote flexible and adaptive behavior (Pourtois, Schettino, & Vuilleumier, 2013).

One of the most used methods to study attentional mechanisms is the cost and benefit paradigm (Posner, 1980). In this paradigm, a peripheral onset-cue is presented, followed after a variable temporal interval (or stimulus onset asynchrony, SOA) by a target requiring a speeded detection response. The participants are informed that the position of the cue is not predictive of the position of the target. Yet, at short SOAs (i.e., about 100–300 ms), reactions times (RTs) are usually shorter when the target appears at the same position as the cue, which would indicate that attention has been involuntary oriented in accordance with the (uninformative) peripheral cue. Interestingly, at longer SOAs (i.e., more than about 300 ms, Posner & Cohen, 1984), detection RTs (or discrimination RTs, see Lupiáñez, Milán, Tornay, Madrid, & Tudela, 1997) become longer to targets appearing at cued versus uncued locations.

This latter effect was coined “inhibition of return” (IOR) by Posner, Rafal, Choate, and Vaughan (1985). Its canonical interpretation is that attention is initially involuntarily captured by the cue, then disengaged, and finally inhibited to return to the position previously occupied by the cue (e.g., Berlucchi, 2006). This inhibition would be associated with impaired perceptual processes, affecting the detection of stimuli appearing at cued locations (e.g., Prime & Ward, 2006). However, even if this original explanation is still acknowledged by many researchers in the field, recent evidence indicates that IOR may rather result from multiple mechanisms, or from a single mechanism that impacts multiple stages of processing depending on the task parameters (see Berlucchi, 2006, Lupiáñez, 2010, for reviews). For example, Lupiáñez (2010) proposes that peripheral cues produce three effects (detection cost, spatial selection benefit and spatial orienting benefit), each following a different time course, and having a different contribution to performance as a function of the task set and the nature of the target. IOR could also reflect a bias against making saccades towards the location of the cue (e.g., Abrams and Dobkin, 1995, Rafal et al., 1989; Chica, Taylor, Lupiáñez, & Klein, 2010) or a reluctance to respond (a criterion shift) to stimuli appearing at cued locations (Klein & Taylor, 1994; see Klein, 2000, for a review).

Nonetheless, regardless of the mechanisms involved, it seems widely assumed that IOR subserves adaptive behavior. IOR would operate to encourage orienting towards novel objects and events (Posner & Cohen, 1984) and discourage wasteful re-inspections of previously attended locations (Klein, 1988). IOR would thus act as a “foraging facilitator” (Klein and MacInnes, 1999, Wang and Klein, 2010), making visual search more efficient. More conservative response criterion for cued locations would also provide the chance to gather extra information from other locations (Klein & Taylor, 1994) and allow the adjustment of decisions and behavior, precluding inaccurate or needless responses to already examined locations or objects (Ivanoff & Taylor, 2006).

If IOR is indeed ascribed evolutionary significance, its size and/or time course should be affected by the nature (the meaning, the emotional content) of the cue and/or the target: IOR should not reduce our chances of noticing event information that could be relevant for our well-being or survival, and in particular, human faces or threatening events. Indeed, faces are particularly salient stimuli, conveying crucial information for social interactions, and due to their biological and social significance, faces may enjoy a privileged processing status: detecting facial configurations is usually fast and efficient (e.g., Pegna, Khateb, Michel, & Landis, 2004) and faces would more likely attract attention to their location than other more common objects (e.g., Ro et al., 2001, Theeuwes and Van der Stigchel, 2006; see Palermo & Rhodes, 2007 for a review). Visual detection, perceptual sensory analysis and attention are also typically heightened (attention being more easily captured, and/or more difficultly disengaged) for threatening (angry or fearful faces, spiders, snakes…) relative to neutral stimuli in various tasks (e.g., dot probe tasks, visual search, attentional blink…; see Pourtois et al., 2013, for a review). Several findings suggest these effects do not reflect faster recognition or response selection once attention has been focused, but rather depend on a coarse perceptual analysis which can operate outside or before attentive fixation (see Domínguez-Borràs & Vuilleumier, 2013).At the brain level, the amygdala, a subcortical structure central to emotion appraisal and learning, could be at least partly responsible for the emotional enhancement of visual perception. The amygdala could act through direct feedback projections to visual areas (including V1) or indirect projections to the dorsal frontoparietal attentional network. Importantly, amygdala may activate to emotional stimuli without explicit attention in many (though not all) situations and this activation may occur before or in parallel with the recruitment of endogenous and exogenous attentional systems (Pourtois et al., 2013).Since its discovery, IOR has been subject to a great number of studies (see Lupianez, Klein, & Bartolomeo, 2006), but the impact of the meaning or emotional content of the cues and/or targets on IOR has received little attention so far.

A few studies have measured the effects of emotional cues on IOR, usually with the stated purpose of exploring the specifics of spatial orienting towards socially or biologically significant stimuli (e.g., neutral or emotional faces, snakes, spiders…; Fox et al., 2002, Stoyanova et al., 2007, Lange et al., 2008, Hu et al., 2014, Experiment 1). The rationale was that if those stimuli catch and hold attention more efficiently than less relevant ones in the location where they appear, the IOR effect should be abolished, reduced, or delayed when they serve as cues: targets appearing in their location could rather be advantaged compared to targets appearing somewhere else in the visual field. However, contrary to what had been expected, the results rather indicated that manipulations of the emotional nature of the cue have no conspicuous impact on the magnitude and time course of IOR.

In another group of studies, the emotional nature of the targets has been manipulated, following the rationale that if emotionally relevant stimuli are particularly prone to capture or attract attention in an automatic manner, then they should be able to override the IOR effect. Studies using simple target detection and localization tasks have found only partial evidence of IOR modulation by the emotional nature of the target. For example, in one study IOR was reduced in a detection task for schematic target faces bearing sad versus happy expressions, but only when they appeared in the left visual field (Baijal & Srinivasan, 2011). In another study, IOR was smaller when localizing negative (pictures of spiders or angry faces) versus neutral (objects or neutral faces) targets, but only after sustained exposure to these stimuli (i.e., when presented in blocks), not if the target type varied pseudo randomly within blocks (Rutherford & Raymond, 2010). The authors therefore suggested that the magnitude of the IOR effect depends more on the affective context set up (by repeated exposure to negative stimuli) before attentional orienting is initiated than on the emotional content of targets on current trials. Finally, in Hu et al.'s (2014) study (Experiment 2), threatening faces (compared to scrambled faces in a detection task) completely abolished IOR in schizophrenic patients but produced a very small non-significant IOR modulation in healthy participants.

Interestingly, in previous studies the emotional nature of the cue or the target was always task irrelevant: the tasks didn't require any explicit processing of the emotional dimensions of the stimuli. As far as we know, only two studies have tested the impact of emotional targets on IOR with experimental designs in which emotion was task-relevant. Pérez-Dueñas, Acosta, and Lupiáñez (2009) compared IOR for neutral, positive and negative (threat) words presented as targets in an emotional categorization task (emotional vs. neutral).They found that only participants with high trait anxiety failed to show IOR for negative words while no IOR emotional modulation occurred for participants with low trait anxiety. Thus, once again, the effect was restricted to participants with emotional disorders. Yet, the same authors have recently reported evidence that IOR selectively disappeared for angry faces (randomly presented among neutral and happy faces) when the participants had to categorize the faces as emotional or neutral (Pérez-Dueñas, Acosta, & Lupiáñez, 2014). Importantly, this latter effect was independent of the participants' state or trait anxiety levels.

Therefore, based on the preceding results, one general conclusion could be that IOR seems not to be modulated by emotional cues, but can be modulated by emotional targets, especially when emotion becomes relevant for the task and/or for the person. This proposal fits well with a growing body of evidence suggesting that attentional biases towards emotional stimuli might not be as unconditional as ordinarily thought, but instead might depend on the cognitive nature of the task (Carretié, 2014), on the task-relevance of the emotional information (Everaert, Spruyt, & De Houwer, 2013), as well as on its personal relevance (Brosch & Van Bavel, 2012).

However, previous experiments differ in several other methodological aspects which might undermine this conclusion. Firstly, they mostly included a single SOA (1000 ms for Pérez-Dueñas et al., 2009, Pérez-Dueñas et al., 2014 and Rutherford & Raymond, 2010; 550 ms for Baijal & Srinivasan, 2011). Yet, non-emotional task manipulations have demonstrated that two forms of IOR modulation can coexist. For example, target discrimination in comparison to target detection usually produces a reduction of IOR together with a later onset of it (e.g., Lupiáñez et al., 1997). As a consequence, finding an IOR effect of similar size for emotional and neutral targets at one given SOA is not sufficient to conclude that emotional targets are unable to modulate IOR at shorter or longer SOAs. Secondly, the previous experiments were designed with a diversity of emotional stimuli (words, drawings or pictures of faces, pictures of spiders or objects). These various stimuli could have influenced IOR in dissimilar ways because they clearly have different emotional and ecological value (Okon-Singer, Lichtenstein-Vidne, & Cohen, 2013) and because the time courses of their emotional processing can differ, especially during the earlier stages of processing (e.g., Frühholz, Jellinghaus, & Herrmann, 2011). Finally, the experiments with facial stimuli involved different emotional expressions (i.e., angry or sad faces) which may also have a different impact on IOR insofar as distinct expressions may not have the same attentional status (e.g., Whalen et al., 2001, Vuilleumier, 2002, Palermo and Rhodes, 2007).

A second important confound across studies is that even if they obviously differ regarding the task relevance of the emotional nature of the target, detection and localization tasks also differ from emotional discrimination tasks on the depth of feature processing required. Apart from emotional relevance, Pérez-Dueñas et al., 2009, Pérez-Dueñas et al., 2014 studies were also the only studies requiring facial fine-grained processing (e.g., Atkinson & Adolphs, 2011). Therefore, one important unresolved question is whether emotional relevance is a key condition for emotional targets to modulate IOR, or if fine-grained facial processing is the critical factor.

In the light of the above considerations, the current study aimed at providing a more systematic investigation of the potential modulations of both the size and time course of the IOR effect as a function of the affective nature of the target stimuli. One key question here was whether the absence of emotional IOR modulation reported in some of the previous studies was merely due to the use of one single SOA or whether some IOR modulations could have been spotted if a larger set of SOAs had been explored. A second important aim was to further investigate the separate roles of emotional task relevance and depth of facial processing on the modulation of IOR by emotional faces.

For these purposes, in four experiments, we randomly presented various pictures of neutral and fearful faces as targets in standard spatial cueing procedures suitable to observe IOR. The experiments were designed to be as similar as possible but each involved different task requirements: target localization (left/right), face discrimination (face/non-face), emotion discrimination (fearful/neutral), or gender discrimination (male/female). Different aspects of the targets were therefore made task-relevant (location, global face configuration, emotional expression and gender), allowing us to track the impact of emotional task relevance and depth of facial processing on the modulation of IOR by fearful faces. Also, three different SOAs (500, 750 and 1000 ms) were used in order to better characterize the time course of the IOR modulation for these different task requirements. Finally, we opted for random presentations and the use of a wide range of stimuli with the idea of increasing the ecological validity of the experiments, and fearful expressions were chosen due to their special processing status as social signals of threat (e.g., Whalen et al., 2001, Anderson et al., 2003).

Section snippets

Experiment 1: localization task

In Experiment 1, non-face stimuli, neutral faces and fearful faces were randomly presented as targets in a target localization (left or right) task. Using a similar task and design (without fearful faces), Taylor and Therrien (2005, 2008) showed that emotionally neutral faces do not escape the effects of IOR more readily than non-face configurations at various SOAs. On the contrary, Rutherford and Raymond (2010) reported that IOR could be reduced when localizing negative targets (pictures of

Experiment 2: face discrimination task

Experiment 2 was very similar to Experiment 1 except that the participants were asked to perform a face/non-face discrimination task.

Experiment 3: emotion discrimination task

The design of Experiment 3 was exactly the same as the one of Experiments 1 and 2 except that non-face trials were removed and that the participants were required to categorize the target faces as neutral or fearful. If explicit emotional processing is truly a decisive factor, these changes in the task requirements may be sufficient to unveil IOR modulations by emotion.

Experiment 4: gender discrimination task

In this last experiment, participants judged the facial gender of the targets (e.g., Gorno-Tempini et al., 2001, Critchley et al., 2000). If emotional relevance is a key condition for emotional targets to override IOR, results of Experiment 4 might be more similar to the ones of Experiments 1 and 2, where emotion was also task irrelevant. On the contrary, if fine-grained facial processing is the critical factor for emotional faces to modulate IOR, then the results of Experiment 4 might resemble

General discussion

In four experiments we tested under which task conditions fearful target faces might or might not modulate the magnitude and/or the time course of IOR, using a range of different SOAs (500, 750 and 1000 ms). Experiment 1, showed that emotional faces have no impact on IOR whatsoever when the participants perform a target localization task. These first results are consistent with those of previous experiments also using simple tasks which didn't require any explicit face processing (detection and

Conclusions

To conclude, the present study aimed at isolating the conditions under which fearful facial targets are able to overcome the robust IOR effect. Its main contribution is that it unequivocally demonstrates that IOR is not imperatively modulated (neither in magnitude nor in time course) by the presence of negative emotional target faces. Indeed, while this kind of target produced similar IOR effects as emotionally neutral faces for simple localization and face/non-face discrimination tasks, they

Acknowledgments

This study was supported by a research grant funded by the Spanish Ministerio de Economía y Competitividad (ref. PSI2012-34158) to María J. Funes.

References (59)

  • G. Pourtois et al.

    Brain mechanisms for emotional influences on perception and attention: What is magic and what is not

    Biological Psychology

    (2013)
  • J. Rellecke et al.

    Does processing of emotional facial expressions depend on intention? Time-resolved evidence from event-related brain potentials

    Biological Psychology

    (2012)
  • B. Valdés-Conroy et al.

    Following the time course of face gender and expression processing: A task-dependent ERP study

    International Journal of Psychophysiology

    (2014)
  • Z. Wang et al.

    Searching for inhibition of return in visual search: A review

    Vision Research

    (2010)
  • A.A. Wijers et al.

    Foveal and parafoveal spatial attention and its impact on the processing of facial expression: An ERP study

    Clinical Neurophysiology

    (2012)
  • R.A. Abrams et al.

    The gap effect and inhibition of return: Interactive effects on eye movement latencies

    Experimental Brain Research

    (1995)
  • A.K. Anderson et al.

    Neural correlates of the automatic processing of threat facial signals

    Journal of Neuroscience

    (2003)
  • A.P. Atkinson et al.

    The neuropsychology of face perception: Beyond simple dissociations and functional selectivity

    Philosophical Transactions of the Royal Society B

    (2011)
  • S. Baijal et al.

    Emotional and hemispheric asymmetries in shifts of attention: An ERP study

    Cognition and Emotion

    (2011)
  • G. Berlucchi

    Inhibition of return: A phenomenon in search of a mechanism and a better name

    Cognitive Neuropsychology

    (2006)
  • L. Carretié

    Exogenous (automatic) attention to emotional stimuli: A review

    Cognitive, Affective, & Behavioral Neuroscience

    (2014)
  • M. Cheal et al.

    Timing of facilitatory and inhibitory effects of visual attention

    Visual Cognition

    (2002)
  • A.B. Chica et al.

    Two mechanisms underlying inhibition of return

    Experimental Brain Research

    (2010)
  • H. Critchley et al.

    Explicit and implicit neural mechanisms for processing of social information from facial expressions: A functional magnetic resonance imaging study

    Human Brain Mapping

    (2000)
  • J. Domínguez-Borràs et al.

    Affective biases in attention and perception

  • K.R. Dukewich

    Reconceptualizing inhibition of return as habituation of the orienting response

    Psychonomic Bulletin & Review

    (2009)
  • K.R. Dukewich et al.

    Inhibition of return: A phenomenon in search of a definition and a theoretical framework

    Attention, Perception, & Psychophysics

    (2015)
  • T. Everaert et al.

    On the malleability of automatic attentional biases: Effects of feature-specific attention allocation

    Cognition and Emotion

    (2013)
  • C.L. Folk et al.

    Involuntary covert orienting is contingent on attentional control settings

    Journal of Experimental Psychology: Human Perception and Performance

    (1992)
  • Cited by (8)

    • Influence of inhibitory tagging (IT) on emotional and cognitive conflict processing: Evidence from event-related potentials

      2017, Neuroscience Letters
      Citation Excerpt :

      All pictures were programmed by E-prime 2.0 (Psychology Software Tools Inc., Pittsburgh, USA) and were presented on a Lenovo 17-in monitor. Upon arrival at the laboratory, which was quiet and dimly lit, the participants were familiarized with the standard spatial cueing procedure [15]. The procedure for each trial was as follows (Fig. 1).

    • A comprehensive look at phobic fear in inhibition of return: Phobia-related spiders as cues and targets

      2017, Journal of Behavior Therapy and Experimental Psychiatry
      Citation Excerpt :

      Another important factor advantageous for our study compared to the previous ones may be participants’ characteristics. While some studies (Baijal & Srinivasan, 2011; Rutherford & Raymond, 2010; Silvert & Funes, 2016) recruited unselected samples (i.e., no fearful group), others recruited anxious individuals and did not find group differences in behavior (Berdica et al., 2014; Pérez-Dueñas et al., 2013). Spiders have been shown to gain preferential processing by triggering anxious emotional responses (Gerdes & Alpers, 2014) and amygdala activation in phobic patients (Alpers et al., 2009), therefore we believe it is crucial to include fearful individuals in these kind of designs.

    View all citing articles on Scopus
    View full text