Unconscious Detection of One’s Own Image

Abstract

A key mechanism behind preferential processing of self-related information might be an early and automatic capture of attention. Therefore, the present study tested a hypothesis that one’s own face will attract bottom-up attention even without conscious identification. To test this, we used a dot-probe paradigm with electrophysiological recordings, in which participants (N = 18) viewed masked and unmasked pairs of faces (other, self) presented laterally. Analysis of the sensitivity measure d ′ indicated that faces were not consciously identified in the masked condition. A clear N2 posterior-contralateral (N2pc) component (a neural marker of attention shifts) was found in both the masked and unmasked conditions, revealing that one’s own face automatically captures attention when processed unconsciously. Therefore, our study (a) demonstrates that self-related information is boosted at an early (preconscious) stage of processing, (b) identifies further features (beyond simple physical ones) that cause automatic attention capture, and (c) provides further evidence for the dissociative nature of attention and consciousness.

Keywords

self-face attention event-related potential N2pc unconscious processing open data

Faces are special types of stimuli that, because of their extreme relevance and preferential neurophysiological processing, constitute an interesting model for investigating unconscious perception (Axelrod, Bar, & Rees, 2015). This might be particularly true in the case of one’s own face (the self-face), as multiple previous studies have demonstrated that such self-related information is preferentially processed at various stages of the processing hierarchy (e.g., Wójcik, Nowicka, Kotlewska, & Nowicka, 2018). Nevertheless, a mechanistic, model-based explanation of the self-preference effect is still missing.

The mechanisms boosting the processing of self-relevant information at later stages could be driven by an early and automatic capture of attention. This hypothesis can be directly tested within the context of visual selection models (Itti & Koch, 2001; Theeuwes, 2010). Both models assume that the feedforward, stimulus-driven sweep of information through the brain is sufficient for specifying visual field locations in which local attributes significantly differ from the surroundings. At this stage, visual selection can also be driven by other factors such as emotional content (Holmes, Bradley, Kragh Nielsen, & Mogg, 2009) or the familiarity of an object (Awh, Belopolsky, & Theeuwes, 2012). Importantly, an object’s identity will become available for top-down processes only after it has captured attention and been selected. Therefore, both models assume that detection of a salient object takes place before it is identified.

On the basis of these models, we hypothesized that the saliency property is assigned to the image of our own face in the preattentive stage of processing. Consequently, we expected a self-face to attract attention in an automatic, exogenous fashion. We tested this hypothesis in a recently published study using a dot-probe paradigm, which allows investigating attention-grabbing properties of to-be-ignored stimuli (Luck & Kappenman, 2012). We demonstrated that presentation of a self-face results in an automatic attention capture, as indicated by a robust N2 posterior-contralateral (N2pc) component (Wójcik et al., 2018). The N2pc is an event-related potential (ERP) defined by a greater negativity at the posterior electrodes contralateral to attended stimuli (with respect to the ipsilateral electrodes). It occurs approximately 200 to 300 ms after stimulus onset and is considered a neuronal marker of the attention-switching mechanism (e.g., Eimer & Kiss, 2008).

Although our previous study revealed an automatic directing of attention toward the image of one’s own face, it did not directly address the unconscious and preidentification nature of the attentional capture. Therefore, we designed a dot-probe experiment in which faces were processed either consciously or unconsciously. A backward-masking procedure was used to render faces invisible, and its validity was demonstrated in an identification test. Electrophysiological recordings were used to estimate the N2pc component and thus investigate automatic shifts of attention. We expected to observe a reliable N2pc in response to conscious and unconscious presentation of the self-face.

Method

Participants

Eighteen participants (10 female) between the ages of 19 and 32 years (M = 25 years, SD = 3) completed the study. All participants had normal or corrected-to-normal vision and reported no history of mental or neurological disorders. Seventeen of these participants were right handed, as assessed by the Edinburgh Handedness Inventory (Oldfield, 1971). The study was conducted with the approval of the human ethics committee of the SWPS University of Social Sciences and Humanities (Warsaw, Poland). All participants gave written informed consent prior to the experiments. In the unconscious dot-probe task, we expected to observe the same effect as in the conscious dot-probe task. The necessary sample size was estimated on the basis of the data acquired in our previous study, in which we investigated attention shifts to the self-face in the conscious dot-probe task (Wójcik et al., 2018). By pooling mean differences and standard deviations of the electrophysiological responses across participants, we estimated Cohen’s d using paired-samples two-tailed t tests on the N2pc amplitude (d = −0.82). We found that a sample size of 18 participants would be sufficient to detect an N2pc effect of the same magnitude with 95% power at the .05 significance level. The estimation was run in G*Power software (Faul, Erdfelder, Lang, & Buchner, 2007).

Stimuli

In the dot-probe task, the stimuli were bilaterally presented pairs of gray-scale face photographs. Each pair contained a self-face and a randomly selected other face from a photograph set (4 male and 4 female) taken from the A series of the Karolinska Directed Emotional Faces database (Lundqvist, Flykt, & Öhman, 1998). To avoid the effects of facial expression, we selected actors on the basis of the unbiased hit rates for neutral-expression detection (Goeleven, De Raedt, Leyman, & Verschuere, 2008). Self-face photographs were taken prior to the experiment. All stimuli were cropped to include only the face, resized to subtend 6.9° × 8.9° of visual angle and equaled for mean luminance using Adobe Photoshop. Cue faces were positioned on the screen with their inner edge 3° to the left and right of the fixation cross. This range is sufficient to detect horizontal eye movements and, as a consequence, to reject trials contaminated with these artifacts (Meyberg, Sommer, & Dimigen, 2017). The mask set consisted of 16 pictures (8 male and 8 female) also taken from the A series of the Karolinska Directed Emotional Faces database. They were created by randomly relocating elements critical for face recognition (eyes, nose, mouth) to disturb the classic feature configuration. For each participant, the gender of other faces and faces used to prepare masks were the same as the participant’s gender, which allowed us to control for the between-category variability of attentional effects.

Procedure and apparatus

Electrode caps (ActiCAP; Brain Products, Munich, Germany) were placed on participants’ heads. Participants were seated in a comfortable chair in a dimly lit and sound-attenuated room. During the experimental tasks, an adjustable chin rest maintained head position and a constant viewing distance of 72 cm. The experimental tasks were presented on a Flex Scan EV-2450 (Hakusan, Ishikawa, Japan) computer screen through an Intel Core i3 computer running Presentation software (Neurobehavioral Systems, Albany, CA). The electroencephalogram (EEG) signal was amplified using QuickAmp and digitized using BrainVision Recorder software (Brain Products, Munich, Germany). The experimental tasks were presented in the following fixed order: masked identification task, masked dot-probe task, unmasked identification task, unmasked dot-probe task. We used this fixed sequence, rather than fully randomizing the order of masked and unmasked trials, because presenting a self-face in unmasked trials would result in participants expecting this stimulus and, consequently, in a lower perception threshold in the masked trials (see Lamy, Carmel, & Peremen, 2017). However, the impact of the tasks sequence was investigated in a control behavioral experiment (see the Results section).

Dot-probe task

Each trial started with a fixation cross (subtending 0.4° × 0.4° of visual angle, positioned in the center of the screen), which remained on-screen for the duration of the trial (see Fig. 1a). After 1,000 ms, a pair of faces was presented bilaterally for 32 ms. In the masked task, faces were followed by backward masks, which remained on-screen for 50 ms. In the unmasked task, no mask was presented. A probe (an asterisk subtending 0.3° × 0.3° of visual angle) was then displayed for 150 ms in the visual field previously occupied by the self-face (congruent condition) or in the opposite visual field (incongruent condition). The other face was presented contralateral to the self-face. Participants were instructed to indicate (as quickly and as accurately as possible) the side on which the probe appeared by pressing the button corresponding to the probe’s location with their left or right index finger. The participants were requested to maintain fixation on the cross and ignore the face cues. Each of the four trial types (the combination of one of two cue-presentation sides and one of two target-presentation sides) was displayed 40 times. Each dot-probe task consisted, therefore, of 160 trials. To test for the presence of an attentional bias, we compared reaction times (RTs) to targets that appeared at the prior location of the prominent stimulus (i.e., congruent trials) with RTs to targets that appeared at the prior location of the neutral stimulus (i.e., incongruent trials). Note that the self-face was presented four times more frequently than a single other face (for further information, see the Supplemental Material available online).

Fig. 1.

Example sequences for masked trials from the (a) dot-probe task and (b) identification task. Trials always started with a fixation cross. Next, two faces (cues) were presented laterally, followed by masks. In the dot-probe task, the self-face was presented in each trial, and the side on which the self-face appeared was chosen randomly. A target (an asterisk) was presented on the left or right side (chosen randomly), and participants’ task was to ignore faces and focus on indicating (with a button press) the side on which the target appeared. In the identification task, which was introduced to test visibility of faces, the self-face was presented either on the left or on the right side, but only on half of the trials (on the other half, two other faces were presented). Participants were asked to indicate whether their own face was presented in a given trial. The only difference between masked and unmasked tasks was the presence or absence of a mask. Note that the letters “S” and “O,” indicating self-face and other face, respectively, are for visualization purposes only and were not part of the experimental stimuli. The stimuli were taken from the A series of the Karolinska Directed Emotional Faces database (images AF19NES, AF34NES, and AF26NES; Lundqvist, Flykt, & Öhman, 1998).

Identification task

The procedure and parameters of the identification task (see Fig. 1b) were similar to those in the dot-probe task. Each trial started with a fixation cross, which stayed on screen for the whole trial. Then, two bilaterally presented faces appeared for 32 ms. In half of the trials, the display consisted of two other faces, and in the remaining trials, one of the faces was a self-face. Participants had to report whether they saw their own face or not. In the masked identification task, a backward mask was presented after the cue faces for 50 ms. Each identification task consisted of 40 trials presenting two other faces and 40 trials presenting an other face and a self-face. To test whether stimuli were presented below or above the threshold of awareness, we calculated the sensitivity measure d ′.

EEG and electrooculogram (EOG) recordings

The EEG was continuously recorded with 64 Ag-AgCl electrically shielded electrodes mounted on an elastic cap (ActiCAP) and positioned according to the extended 10-20 system. For ocular artifact scoring, vertical and horizontal EOGs were recorded from bipolar electrodes placed at the supra- and suborbit of the right eye and at the external canthi. EEG electrode impedances were kept below 10 kΩ. The data were amplified using a 128-channel amplifier (QuickAmp; Brain Products, Enschede, The Netherlands) and digitized at a 500-Hz sampling rate. The EEG signal was recorded against an average of all channels calculated by the amplifier hardware. The 62 channels were rereferenced off-line to the algebraic average of the left and right earlobes, notch-filtered at 50 Hz, and band-pass-filtered from 1 to 30 Hz using a zero-phase Butterworth filter (12 dB/octave).

Behavioral analysis

All trials with RTs shorter than 100 ms and longer than 1,000 ms were excluded from the analysis for the following reasons. RTs shorter than 100 ms could not be attributed to the appropriate reactions to presented visual stimuli; rather, they reflected reflexive responses. There may be many reasons for very long RTs (e.g., eye blinks at the time of a stimulus presentation, disturbed attention). These RTs did not constitute a good measurement of the researched mechanism and therefore were rejected. Rejection rates were as follows: masked incongruent trials: 0.6%, masked congruent trials: 0.7%, unmasked incongruent trials: 0.6%, unmasked congruent trials: 0.5%. The results of a two-sample Kolmogorov-Smirnov test indicated that this exclusion procedure did not change the distributions. Mean RTs for trials with correct responses were computed for each participant for both dot-probe tasks.

ERP analysis

Occipital-temporal channels PO8 and PO7 were chosen for the ERP analysis of data obtained in the dot-probe tasks. These electrodes are frequently reported as exhibiting maximal N2pc amplitudes (Eimer & Kiss, 2008; Wójcik et al., 2018). The EEG signal was segmented into 800-ms-long epochs, from 200 ms before to 600 ms after the cue onset. These epochs were baseline-corrected against the mean voltage during the 200-ms prestimulus period. We rejected epochs contaminated with vertical eye movements and blinks (a change in voltage in the vertical EOG channel exceeding 100 µV within a 200-ms period), horizontal eye movements (a change in voltage in the horizontal EOG channel exceeding 50 µV within any 200-ms period), or other artifacts (in all channels: voltage steps exceeding 50 µV, voltage change exceeding 100 µV within any 200-ms period, amplitudes greater than 200 µV and lower than −200 µV, activity in 100-ms intervals lower than 0.5 µV). The mean number of segments that passed the artifact-rejection procedure in the masked and unmasked dot-probe tasks were 131 (SD = 18) and 136 (SD = 14), respectively. The number of epochs used to compute ERPs did not differ significantly between tasks. The rejection rates were as follows—masked dot-probe task: 18% (SD = 11%), unmasked dot-probe task: 15% (SD = 9%).

The ipsilateral waveform was calculated as the average of signals recorded at the PO7 electrode when the self-face stimulus was presented on the left side of the display and at the PO8 electrode when the self-face stimulus was presented on the right side of the display. The contralateral waveform was computed as the average of signals recorded at the PO7 electrode when the self-face stimulus was presented on the right side of the display and at the PO8 electrode when the self-face stimulus was presented on the left side of the display. To clearly visualize the N2pc component and isolate it from other overlapping components, we calculated the difference between the collapsed contralateral and ipsilateral waveforms.

On the basis of visual inspection of the grand-average contralateral-ipsilateral waveforms obtained in the masked and unmasked dot-probe tasks (see Fig. 2), the N2pc component was quantified as mean amplitudes within two successive time windows (early N2pc: 200–300 ms after stimulus presentation; late N2pc: 300–400 ms after stimulus presentation). This procedure is in line with previous studies on electrophysiological markers of attentional capture (Eimer & Kiss, 2008; Holmes et al., 2009).

Fig. 2.

Grand-average event-related potentials (ERPs) for electrodes placed at locations contralateral and ipsilateral to the self-face stimulus (top row) and contralateral-ipsilateral difference waveforms (bottom row), separately for the masked and unmasked dot-probe tasks. ERPs were time-locked to the onset of the cue faces (self vs. other). The differences between contralateral and ipsilateral waveforms were revealed by cluster-based permutation tests. Gray shading indicates the time windows of the early N2 posterior-contralateral (N2pc) component (200–300 ms after stimulus presentation) and the late N2pc component (300–400 ms after stimulus presentation). Asterisks indicate statistically significant differences between contralateral and ipsilateral waveforms (*p < .05, **p < .01).

Statistical analysis

The sensitivity measure d ′ was used to assess the visibility of stimuli. Hits and false alarms of 0 and 1 were replaced using the log-linear rule, as it is the least biased method of correcting extreme values (Stanislaw & Todorov, 1999). For t-test results, 95% confidence intervals (CIs) are given for the mean difference between the experimental condition and the baseline.

A Shapiro-Wilk normality test conducted on amplitude- and visibility-measure distributions showed that they did not deviate from normality. Because two RT distributions were skewed, a Box-Cox power transformation was applied to all the RT variables. Nevertheless, the RTs in the masked dot-probe task did not meet the assumption of normality of the paired-samples t test; therefore, a Wilcoxon signed-rank test was used.

The traditional null-hypothesis significance-testing approach was complemented with Bayesian analysis methods, which allow testing for lack of differences between variables. Bayes factors (BFs) were computed using JASP software (Wagenmakers et al., 2018). The BF is the ratio of the probability of observing the data given that the alternative hypothesis is true to the probability of observing the data given that the null hypothesis is true. To test for differences and the lack of differences between mean amplitudes, we used the Bayesian equivalent of a paired-samples t test with a medium prior scale (Cauchy scale: .707). A Bayesian one-sample t test with a medium prior scale (Cauchy scale: .707) was used to test whether participants performed the identification task at or above the chance level. A possible relationship between the sensitivity measure (d ′) and the amplitude of the N2pc component was explored with a Bayesian correlation test with a stretched beta prior width equal to 1. We interpreted the BF according to Lee and Wagenmakers (2013). Briefly, a BF between 0.1 and 0.33 indicates moderate evidence for lack of differences, whereas a BF between 0.33 and 1 implies anecdotal evidence for lack of differences. A BF between 3 and 10 gives moderate evidence for the presence of an effect, and a BF between 10 and 100 indicates strong evidence.

Finally, to establish the precise timing of the difference between contralateral and ipsilateral waveforms in both masked and unmasked dot-probe tasks, we computed a nonparametric cluster-level one-sample t test (Maris & Oostenveld, 2007). This technique was run on all time points obtained after the ERP averaging procedure.

Results

Behavioral results

Dot-probe tasks

A Wilcoxon signed rank test indicated no significant differences between RTs for congruent and incongruent trials in the masked dot-probe task, W = 129.0, p = .06, BF = 2.13 (alternative hypothesis 2.1 times more likely, anecdotal evidence for the alternative hypothesis). RTs obtained in the unmasked dot-probe task did not differ between the congruent and incongruent trials, t(17) = −0.71, p = .487, BF = 0.30 (null hypothesis 3.3 times more likely, moderate evidence).

Identification tasks

Participants performed at chance in identifying their own face in the masked identification task, t(17) = 1.57, p = .135, 95% CI for the mean difference = [−0.07, 0.5], BF = 0.68 (null hypothesis 1.5 times more likely, anecdotal evidence). In contrast, d ′ values were significantly different from 0 in the unmasked identification task, t(17) = 3.85, p = .001, 95% CI = [0.73, 2.51], d = 0.91, BF = 23.98 (alternative hypothesis 24 times more likely, strong evidence).

ERP results

For the masked dot-probe task, a paired-samples t test (contralateral vs. ipsilateral) was performed to assess the presence of an early N2pc, t(17) = −2.34, p = .031, 95% CI for the mean difference between contralateral and ipsilateral amplitudes = [−0.48, −0.03], d = −0.55, BF = 4.07 (alternative hypothesis 4.1 times more likely, moderate evidence). It indicated that the overall mean amplitudes of the contralateral waveform were more negative (M = −0.02 µV) than the mean amplitudes of the ipsilateral waveform (M = 0.23 µV).

For the unmasked dot-probe task, a paired-samples t test (contralateral vs. ipsilateral) indicated a clear early N2pc component, as amplitudes of the contralateral waveforms were more negative (M = 0.23 µV) than the amplitudes of the ipsilateral waveforms (M = 0.74 µV), t(17) = 2.91, p = .01, 95% CI = [−0.87, −0.14], d = −0.67, BF = 10.71 (alternative hypothesis 10.7 times more likely, strong evidence). A paired-samples t test was run (contralateral vs. ipsilateral) to test whether a late N2pc component was elicited. It showed that the amplitudes of the contralateral waveforms significantly differed (M = 3.24 µV) from the amplitudes of the ipsilateral waveforms (M = 3.61 µV), t(17) = 2.54, p = .021, 95% CI = [−0.68, −0.06], d = −0.6, BF = 5.66 (alternative hypothesis 5.7 times more likely, moderate evidence).

The nonparametric cluster-level one-sample t tests revealed a difference between contralateral and ipsilateral waveforms in the time window of 188 to 262 ms in the masked dot-probe task (p = .003). In the unmasked dot-probe task, two time windows differentiated the contralateral and ipsilateral waveforms: 214 ms to 282 ms (p = .015) and 306 ms to 364 ms (p = .022), respectively. These clusters are denoted in the lower panels of Figure 2 with difference waves (contralateral – ipsilateral).

The validity of the masking procedure

The visibility measure (d ′) observed in the masked identification task was not different from zero, but the Bayesian analysis provided only anecdotal evidence for the null hypothesis (i.e., no difference). Yet this can be explained by the fact that we used a relatively small sample of participants (N = 18). To further test the validity of the masking procedure, we conducted the following analyses.

Subgroup analysis

When we consider interindividual differences, it might be argued that some participants were conscious of the target presentation, and the early N2pc effect was influenced mainly by their data. Therefore, we divided the sample into two subgroups: One (n = 9) performed at chance level in the masked identification task (mean d ′ = −0.24), and another (n = 9) performed above chance (mean d ′ = 0.66). However, the group factor did not differentiate the N2pc amplitude, t(16) = 0.58, p = .573, BF = 0.46 (null hypothesis 2.2 times more likely, anecdotal evidence).

N2pc and d′ correlation

Further, the amplitude of the early N2pc component observed in the masked dot-probe task did not correlate with the d ′ obtained in the masked identification task, r = −.2, p = .424, BF = 0.39 (null hypothesis 2.6 times more likely, anecdotal evidence), indicating no relation between visibility and attention shifts. In contrast, the amplitude of the early N2pc component elicited in the unmasked dot-probe task correlated negatively with the d ′ measured in the unmasked identification task, r = −.58, p = .011, BF = 5.87 (alternative hypothesis 5.9 times more likely, moderate evidence).

Behavioral control experiment

Masked stimuli can theoretically become more visible as a result of repeated exposure. To establish whether the fixed sequence of the experimental tasks (the identification task always occurred before the dot-probe task) had influenced the visibility measure, we conducted a behavioral control experiment (N = 17) with the tasks in reverse order . The visibility measurements acquired in both experiments were compared using a Wilcoxon rank-sum test because the d ′ distribution obtained in the control experiment deviated from normality. The results indicated that the tasks’ order did not influence the visibility measure, W = 166.0, p = .68, BF = 0.39 (null hypothesis 2.6 times more likely, anecdotal evidence).

Discussion

The main aim of the present study was to test whether an unconsciously processed image of one’s own face triggers automatic, exogenous attentional shifts. On the basis of our previous study (Wójcik et al., 2018), we hypothesized that such an attention-capture effect would be present even without conscious identification. The collected data confirmed our predictions, because in both masked (unconscious) and unmasked (conscious) dot-probe tasks, a clear N2pc was elicited by the self-face. These findings are relevant to the ongoing debates regarding both the self-preference effect and the dissociative nature of attention and consciousness.

Whereas multiple previous studies investigated the self-preference effect (for a review, see Humphreys & Sui, 2016), a mechanistic, model-based explanation of the phenomenon is still missing. Here, we aimed to study self-preference within a context of the visual selection models proposed by Itti and Koch (2001) and Theeuwes (2010). We found that the self-face is analyzed preconsciously, resulting in an automatic capture of attention. Therefore, our results clearly contradict studies suggesting that the prioritization of self-relevant information occurs at later, conscious-processing stages (e.g., Tacikowski & Ehrsson, 2016). Nevertheless, our finding is in line with those of several other experiments showing that self-preferential processing occurs even if conscious access is reduced or eliminated (e.g., Alexopoulos, Muller, Ric, & Marendaz, 2012).

From the perspective of the visual selection theories (Itti & Koch, 2001; Theeuwes, 2010), our finding adds to the growing body of evidence demonstrating that bottom-up attention capture, classically thought to be purely exogenous and automatic, might be affected by factors typically considered as top down (for a review, see Awh et al., 2012). Indeed, the observed shifts of attention to the self-face will be difficult to account for by the physical, bottom-up features of an image. However, previous studies indicate that global analysis of faces and recognition of race or gender are not performed unconsciously (Axelrod & Rees, 2014; for a review, see Axelrod et al., 2015). Therefore, we hypothesized that the effect might be explained in terms of extreme familiarity of the self-face. Indeed, Awh et al. (2012) proposed a modification of the classic selection models by adding familiarity as a feature assessed at the preconscious stage. This is in line with previous studies showing unconscious priming by famous (familiar) faces but not other (unfamiliar) faces (Henson, Mouchlianitis, Matthews, & Kouider, 2008; Kouider, Eger, Dolan, & Henson, 2009). If indeed familiarity is the feature attracting attention, then it might be predicted that attention shifts to famous faces will also be observed in the dot-probe task. This would be in line with the results of Devue, Van der Stigchel, Brédart, and Theeuwes (2009), revealing that a famous face interfered with a visual search task in a similar manner to the self-face (however, the authors attributed this effect not to automatic shifts of attention but, rather, to difficulty in disengaging attention when a familiar face was detected). Therefore, investigating both famous faces and self-faces in the dot-probe task would allow addressing whether familiarity or other factors specific to the self-face (affective response or activation of the self-representation) play a key role in the observed attention capture.

It remains to be investigated how general our finding is and, specifically, whether other complex stimuli might also attract attention unconsciously (e.g., if they are extremely familiar). It might be argued that because of their extreme saliency and relevance, faces are processed automatically, and thus the results presented here are specific to faces only. Indeed, whereas other stimuli cause unconscious priming effects only when top-down (spatial or temporal) attention is deployed (Lachter, Forster, & Ruthruff, 2004; Naccache, Blandin, & Dehaene, 2002), faces can be processed unconsciously even outside the scope of top-down spatial attention (i.e., in uncued locations; Finkbeiner & Palermo, 2009). However, the effect observed here cannot be ascribed to mere automaticity of face processing because faces of other people were used as control stimuli.

The mechanism of rapid self-detection might be subcortical and based on the magnocellular, rather than cortical, pathway (Johnson, 2005). The hypothesis that the prioritization of self could take place very rapidly, even without complete recognition of a face, is supported by an analogous line of research regarding the processing of faces with emotional expressions. These stimuli also attract attention in an automatic manner, as evidenced by the presence of the N2pc component (Holmes et al., 2009), and are processed unconsciously (Axelrod et al., 2015). Interestingly, Vuilleumier, Armony, Driver, and Dolan (2003) showed that the amygdala is sensitive to the images of emotional faces presented at low spatial frequencies. In addition, a follow-up study with intracranial recordings (Méndez-Bértolo et al., 2016) showed that this activation takes place very early, around 74 ms after stimulus onset. There is also evidence that patients with completely damaged primary visual cortices have the ability to discriminate emotional faces and that this process involves the right amygdala (Pegna, Khateb, Lazeyras, & Seghier, 2005). We hypothesized that feedforward processing via the subcortical pathway could also be the mechanism of rapid automatic attentional orienting toward one’s own face. This supposition is supported by a study showing that amygdala is indeed sensitive to familiar faces presented at low spatial frequencies (Ramon, Vizioli, Liu-Shuang, & Rossion, 2015). Stimuli comprising either high or low spatial frequencies only might be used in the future to elucidate involvement of the subcortical and cortical pathways in detection of a self-face.

Interestingly, our finding could be of importance for developing computational models of voluntary eye movements and attentional shifts. Recent studies have shown that combining saliency map models based on differences in low-level physical features with face-detection algorithms results in significant improvements in the performance of such models (Cerf, Harel, Einhäuser, & Koch, 2008). Adding a module creating a familiarity feature based on the previous selection history might thus improve the predictive power by incorporating this feature into the saliency computation.

Further, our study is relevant to the extensively debated relation between attention and consciousness (for a review, see Koch & Tsuchiya, 2007). We provide further evidence for the claim that consciousness of a stimulus is not necessary to evoke bottom-up attention capture (in line with the findings of Hsieh, Colas, & Kanwisher, 2011; Lamy, Alon, Carmel, & Shalev, 2015), which suggests that attention and consciousness can be in principle dissociated. Interestingly, whereas consciousness is not necessary for bottom-up attention capture, some researchers argue that the involvement of top-down attention is crucial for triggering such processes (Hsieh et al., 2011). Top-down attention can be understood either as spatial attention (i.e., the attentional spotlight) or as a nonspecific, limited attentional resource, and there is evidence indicating that these two processes are independent (e.g., Johnston, McCann, & Remington, 1995). In our study, faces were task irrelevant because participants were required to detect the dot probe, but we did not systematically manipulate the top-down attention (i.e., we did not compare trials with and without top-down attention). Yet when we consider the automatic unconscious processing of faces (Finkbeiner & Palermo, 2009), it might be hypothesized that, in contrast to feature singletons, the self-face might be processed and capture bottom-up attention even outside of the spotlight of top-down attention and without the involvement of top-down attentional resources.

The majority of previous studies demonstrating unconscious attention capture used simple stimuli (e.g., feature singletons; Lamy et al., 2015). However, stimuli of higher complexity such as body pictures (Jiang, Costello, Fang, Huang, & He, 2006) and faces (the present study) elicited a similar effect. Attention shift to unconscious complex stimuli indicates that these stimuli must be perceptually integrated and processed up to a high (semantic) level outside of consciousness. Whether unconscious integration is possible is currently debated, with some studies providing evidence in favor of integration and semantic analysis of complex stimuli outside of consciousness (for a review, see Mudrik, Faivre, & Koch, 2014) but other studies disproving these effects (e.g., Biderman & Mudrik, 2017). In conclusion, the present study not only confirmed the attention-grabbing properties of one’s own image (see Wójcik et al., 2018), but also revealed the preidentification and unconscious quality of this process.

Supplemental Material

Wojcik_SupplementalMaterial – Supplemental material for Unconscious Detection of One’s Own Image

Supplemental material, Wojcik_SupplementalMaterial for Unconscious Detection of One’s Own Image by Michał J. Wójcik, Maria M. Nowicka, Michał Bola and Anna Nowicka in Psychological Science

Supplemental Material

WójcikOpenPracticesDisclosure – Supplemental material for Unconscious Detection of One’s Own Image

Supplemental material, WójcikOpenPracticesDisclosure for Unconscious Detection of One’s Own Image by Michał J. Wójcik, Maria M. Nowicka, Michał Bola and Anna Nowicka in Psychological Science

Footnotes

Action Editor

Alice J. O’Toole served as action editor for this article.

Author Contributions

M. J. Wójcik and A. Nowicka developed the study concept. All the authors contributed to the study design. Testing and data collection were performed by all authors. M. J. Wójcik and M. M. Nowicka analyzed and interpreted the data under the supervision of A. Nowicka. M. J. Wójcik, A. Nowicka, and M. Bola wrote the manuscript. All the authors approved the final manuscript for submission.

Declaration of Conflicting Interests

The author(s) declared that there were no conflicts of interest with respect to the authorship or the publication of this article.

Funding

This research was funded by the National Science Centre Poland (Grant 2015/19/B/HS6/01258 to A. Nowicka and Grant 2015/17/D/HS6/00269 to M. Bola). M. Bola was also supported by a stipend from the Polish Ministry of Science and Higher Education (555/STYP/11/2016).

Supplemental Material

Additional supporting information can be found at

Open Practices

Neither of the experiments reported in this article was formally preregistered. All data (mean amplitudes, reaction times, and visibility measures) and the scripts used to present the task have been made publicly available via the Open Science Framework and can be accessed at osf.io/4e7h9/. The complete Open Practices Disclosure for this article can be found at https://journals-sagepub-com.web.bisu.edu.cn/doi/suppl/10.1177/0956797618822971. This article has received the badge for Open Data. More information about the Open Practices badges can be found at .

References

Alexopoulos

Muller

Ric

Marendaz

(2012). I, me, mine: Automatic attentional capture by self-related stimuli. European Journal of Social Psychology, 42, 770–779. doi:10.1002/ejsp.1882

Awh

Belopolsky

A. V.

Theeuwes

(2012). Top-down versus bottom-up attentional control: A failed theoretical dichotomy. Trends in Cognitive Sciences, 16, 437–443. doi:10.1016/j.tics.2012.06.010

Axelrod

Bar

Rees

(2015). Exploring the unconscious using faces. Trends in Cognitive Sciences, 19, 35–45. doi:10.1016/j.tics.2014.11.003

Axelrod

Rees

. (2014). Conscious awareness is required for holistic face processing. Consciousness and Cognition, 27, 233–245. doi:10.1016/j.concog.2014.05.004

Biderman

Mudrik

(2017). Evidence for implicit—but not unconscious—processing of object-scene relations. Psychological Science, 29, 266–277. doi:10.1177/0956797617735745

Cerf

Harel

Einhäuser

Koch

(2008). Predicting human gaze using low-level saliency combined with face detection. In Platt

J. C.

Koller

Singer

Roweis

(Eds.), Advances in neural information processing systems (Vol. 20, pp. 241–248). Cambridge, MA: MIT Press.

Devue

Van der Stigchel

Brédart

Theeuwes

(2009). You do not find your own face faster; you just look at it longer. Cognition, 111, 114–122. doi:10.1016/j.cognition.2009.01.003

Eimer

Kiss

(2008). Involuntary attentional capture is determined by task set: Evidence from event-related brain potentials. Journal of Cognitive Neuroscience, 20, 1423–1433. doi:10.1162/jocn.2008.20099

Faul

Erdfelder

Lang

A.-G.

Buchner

(2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175–191. doi:10.3758/BF03193146

10.

Finkbeiner

Palermo

(2009). The role of spatial attention in nonconscious processing: A comparison of face and nonface stimuli. Psychological Science, 20, 42–51. doi:10.1111/j.1467-9280.2008.02256.x

11.

Goeleven

De Raedt

Leyman

Verschuere

(2008). The Karolinska Directed Emotional Faces: A validation study. Cognition & Emotion, 22, 1094–1118. doi:10.1080/02699930701626582

12.

Henson

R. N.

Mouchlianitis

Matthews

W. J.

Kouider

(2008). Electrophysiological correlates of masked face priming. NeuroImage, 40, 884–895. doi:10.1016/j.neuroimage.2007.12.003

13.

Holmes

Bradley

B. P.

Kragh Nielsen

Mogg

(2009). Attentional selectivity for emotional faces: Evidence from human electrophysiology. Psychophysiology, 46, 62–68. doi:10.1111/j.1469-8986.2008.00750.x

14.

Hsieh

P.-J.

Colas

J. T.

Kanwisher

(2011). Unconscious pop-out: Attentional capture by unseen feature singletons only when top-down attention is available. Psychological Science, 22, 1220–1226. doi:10.1177/0956797611419302

15.

Humphreys

G. W.

Sui

(2016). Attentional control and the self: The self-attention network (SAN). Cognitive Neuroscience, 7, 5–17. doi:10.1080/17588928.2015.1044427

16.

Itti

Koch

(2001). Computational modelling of visual attention. Nature Reviews Neuroscience, 2, 194–203. doi:10.1038/35058500

17.

Jiang

Costello

Fang

Huang

(2006). A gender- and sexual orientation-dependent spatial attentional effect of invisible images. Proceedings of the National Academy of Sciences, USA, 103, 17048–17052. doi:10.1073/pnas.0605678103

18.

Johnson

M. H.

(2005). Subcortical face processing. Nature Reviews Neuroscience, 6, 766–777. doi:10.1038/nrn1766

19.

Johnston

J. C.

McCann

R. S.

Remington

R. W.

(1995). Chronometric evidence for two types of attention. Psychological Science, 6, 365–369. doi:10.1111/j.1467-9280.1995.tb00527.x

20.

Koch

Tsuchiya

(2007). Attention and consciousness: Two distinct brain processes. Trends in Cognitive Sciences, 11, 16–22. doi:10.1016/j.tics.2006.10.012

21.

Kouider

Eger

Dolan

Henson

R. N.

(2009). Activity in face-responsive brain regions is modulated by invisible, attended faces: Evidence from masked priming. Cerebral Cortex, 19, 13–23. doi:10.1093/cercor/bhn048

22.

Lachter

Forster

K. I.

Ruthruff

(2004). Forty-five years after Broadbent: Still no identification without attention. Psychological Review, 111, 880–913. doi:10.1037/0033-295X.111.4.880

23.

Lamy

Alon

Carmel

Shalev

(2015). The role of conscious perception in attentional capture and object-file updating. Psychological Science, 6, 48–57. doi:10.1177/0956797614556777

24.

Lamy

Carmel

Peremen

(2017). Prior conscious experience enhances conscious perception but does not affect response priming. Cognition, 160, 62–81. doi:10.1016/j.cognition.2016.12.009

25.

Lee

M. D.

Wagenmakers

E.-J.

(2013). Bayesian cognitive modeling: A practical course. Cambridge, England: Cambridge University Press.

26.

Luck

S. J.

Kappenman

E. S.

(2012). ERP components and selective attention. In Luck

S. J.

Kappenman

E. S.

(Eds.), The Oxford handbook of event-related potential components (pp. 329–360). New York, NY: Oxford University Press.

27.

Lundqvist

Flykt

Öhman

(1998). The Karolinska Directed Emotional Faces (KDEF) [CD-ROM]. Stockholm, Sweden: Department of Clinical Neuroscience, Psychology Section, Karolinska Institutet.

28.

Maris

Oostenveld

(2007). Nonparametric statistical testing of EEG- and MEG-data. Journal of Neuroscience Methods, 64, 177–190. doi:10.1016/j.jneumeth.2007.03.024

29.

Méndez-Bértolo

Moratti

Toledano

Lopez-Sosa

Martínez-Alvarez

Mah

Y. H.

. . . Strange

B. A.

(2016). A fast pathway for fear in human amygdala. Nature Neuroscience, 19, 1041–1049. doi:10.1038/nn.4324

30.

Meyberg

Sommer

Dimigen

(2017). How microsaccades relate to lateralized ERP components of spatial attention: A co-registration study. Neuropsychologia, 99, 64–80. doi:10.1016/j.neuropsychologia.2017.02.023

31.

Mudrik

Faivre

Koch

(2014). Information integration without awareness. Trends in Cognitive Sciences, 18, 488–496. doi:10.1016/j.tics.2014.04.009

32.

Naccache

Blandin

Dehaene

(2002). Unconscious masked priming depends on temporal attention. Psychological Science, 13, 416–424. doi:10.1111/1467-9280.00474

33.

Oldfield

R. C.

(1971). The assessment and analysis of handedness: The Edinburgh inventory. Neuropsychologia, 9, 97–113. doi:10.1016/0028-3932(71)90067-4

34.

Pegna

A. J.

Khateb

Lazeyras

Seghier

M. L.

(2005). Discriminating emotional faces without primary visual cortices involves the right amygdala. Nature Neuroscience, 8, 24–25. doi:10.1038/nn1364

35.

Ramon

Vizioli

Liu-Shuang

Rossion

(2015). Neural microgenesis of personally familiar face recognition. Proceedings of the National Academy of Sciences, USA, 112, E4835–E4844. doi:10.1073/pnas.1414929112

36.

Stanislaw

Todorov

(1999). Calculation of signal detection theory measures. Behavior Research Methods, Instruments, & Computers, 31, 137–149.

37.

Tacikowski

Ehrsson

H. H.

(2016). Preferential processing of self-relevant stimuli occurs mainly at the perceptual and conscious stages of information processing. Consciousness and Cognition, 41, 139–149. doi:10.1016/j.concog.2016.02.013

38.

Theeuwes

(2010). Top-down and bottom-up control of visual selection. Acta Psychologica, 135, 77–99. doi:10.1016/j.actpsy.2010.02.006

39.

Vuilleumier

Armony

J. L.

Driver

Dolan

R. J.

(2003). Distinct spatial frequency sensitivities for processing faces and emotional expressions. Nature Neuroscience, 6, 624–631. doi:10.1038/nn1057

40.

Wagenmakers

E.-J.

Love

Marsman

Jamil

Verhagen

. . . Morey

R. D.

(2018). Bayesian inference for psychology. Part II: Example applications with JASP. Psychonomic Bulletin & Review, 25, 58–76. doi:10.3758/s13423-017-1323-7

41.

Wójcik

M. J.

Nowicka

M. M.

Kotlewska

Nowicka

(2018). Self-face captures, holds, and biases attention. Frontiers in Psychology, 8, Article 2371. doi:10.3389/fpsyg.2017.02371

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.80 MB

0.07 MB