Ethics Statement

This research was approved by the University of Otago Human Ethics Committee, and all participants provided written informed consent prior to participation.

Participants

Ninety young adults recruited at the University of Otago participated either in exchange for NZ$12.50 or in association with a course. Thirty completed the version with non-face objects (mean age = 22, SD = 5, range = 18–42; 14 males; 27 right handed), thirty completed the version with upright faces (mean age = 21, SD = 2, range = 18–27; 15 males; 26 right handed) and thirty completed the version with inverted faces (mean age = 21, SD = 3, range = 18–31; 10 males; 24 right handed). All participants reported no previous neurological history and normal or corrected-to-normal vision.

Stimuli and procedure

All stimuli appeared on a white background. For the version with non-face objects, two grey-scale images, one of a butterfly and the other of a wheel, served as the target and distractor stimuli. Each non-face object subtended 1° both vertically and horizontally, and 2° separated the target and distractor (edge to edge). For the versions with face stimuli, two grey-scale images portraying faces of famous people (George Clooney—actor and Helen Clark—New Zealand Prime Minister) served as the target and distractor stimuli. Each face subtended 3° vertically and 2.4° horizontally, and 2.5° separated the target and distractor (edge to edge). Relative to the non-face objects, the size of the face stimuli had to be increased because pilot testing showed that 1° faces did not elicit any compatibility effects (i.e., the distractors were entirely ineffective). In the versions with non-face objects and upright faces, both the distractor and the target always appeared right-side up. In the version with inverted faces, both the distractor and the target always appeared upside down. In all versions, the two images were each assigned to one of the two buttons on a DirectN Response Box (Empirisoft, New York), with the stimulus-response mapping counterbalanced across participants.

At the start of each trial (see Figure 1), a black fixation dot with a diameter extending .3° of visual angle appeared at the center of the screen. After 500 ms elapsed, one of the images appeared either above or below the fixation dot. This initial image served as the distractor. After a variable interval (50, 350, 650, or 950 ms), another image appeared at the center of the screen (occluding the fixation dot). This central image served as the target. The distractor was either the same as the target (compatible) or different than the target (incompatible). The distractor was positioned above or below the target in order to prevent spatial compatibility effects based on the side of the distractor and the side of the response [13], [14]. The position of the distractor (above or below), the distractor-target onset asynchrony (50, 350, 650 or 950 ms), the distractor identity, and the target identity were randomly selected before each trial with the constraint that each occurred equally often and all conditions were counterbalanced within each block of 32 trials. The code for the experiment relied on MATLAB (The MathWorks, Natick, MA) and The Psychophysics Toolbox [15], [16].

Participants sat 57 cm from the screen in a dimly lit room. At the start of the experiment, the computer displayed the stimulus-response mapping. The experimenter instructed participants to fixate on the center of the screen throughout the experiment and, when a stimulus appeared at center, to press the assigned button as quickly as they accurately could using the index and middle fingers of their dominant hand. Note that the experimenter never referred to the identities of the images. The trial ended when the computer recorded either a correct response or an error, at which time the distractor and target disappeared, leaving the screen blank for 2000 ms before the next trial started. An error tone sounded if the participant depressed the wrong button, responded within 100 ms after target onset, or failed to respond within 2000 ms after target onset. All participants completed 32 practice trials followed by 320 test trials, which were divided into 10 blocks. Between blocks, the stimulus-response mapping display reappeared and participants were given the opportunity to rest. For the versions with face stimuli, after completing the experiment, participants were shown the faces in the upright orientation and asked whether they recognized them. Correct identification of the two faces was an inclusion criterion; ten additional participants did not meet this criterion and were excluded.

Non-face Objects

Upright Faces

Inverted Faces

Upright Faces versus Inverted Faces

Face processing

The extent to which faces demand attention over and above other classes of objects has attracted considerable interest over recent years [2]. Previous research attributed this phenomenon to enhanced processing specific to faces; however, the potential role of inhibition in this prioritized processing has remained largely uninvestigated until now. Our data clearly demonstrate that less inhibition was used when the distractor engaged face processing (upright distractor face) compared to when the distractor face was inverted, thereby disrupting face processing. For upright faces, although the amount of distractor inhibition was insufficient for the compatibility effect to reverse at the long SOA, the fact that the compatibility effect weakened as the SOA increased converges with previous reports that attention can modulate the neural response to face stimuli [20], [21] and also that the attentional bias toward faces is subject to voluntary control [22]. Note, however, that our face stimuli included hair and hence the attenuation of the compatibility effect as the SOA increased may reflect inhibitory processing of non-facial features.

Superficially, the lack of a negative compatibility effect for upright faces seems inconsistent with a previous report of an inhibitory effect for face distractors [23]. Their task involved unfamiliar faces. The results showed that when the target face had served as the distractor during the previous trial, responses were slower, indicating that the distractor was inhibited (an effect referred to as negative priming). This seems to indicate that distractors depicting faces are subjected to inhibition; however, as noted by the authors, external features (especially hair) were not removed from the faces and thus the negative priming may reflect inhibitory processing of non-facial features. A similar argument can be made regarding the evidence of inhibition of previously cued unfamiliar faces [24], [25]. In addition, it may be the case that unfamiliar faces are subjected to more inhibition during selective attention than famous faces. Consistent with this possibility, Gazzaley et al. [26] reported evidence of inhibitory processing of unfamiliar distractor faces.

One seemingly odd aspect of our results is that the positive compatibility effect elicited by upright distractor faces was no more robust than that elicited by inverted distractor faces. Given that faces are particularly demanding of attention, one might have expected upright distractor faces to trigger stronger compatibility effects, especially given the famous status of the faces. However, considering that faces uniquely suffer from stimulus-specific capacity limits such that attending to a face can exhaust resources and limit processing of additional faces [27], [28], we suggest that face-specific capacity limits attenuated the influence of upright distractor faces, resulting in positive compatibility effects of similar magnitudes for upright and inverted faces.

A limitation of the current study is that the degree of familiarity was not matched across the stimuli used as non-face objects and upright faces; thus, we cannot rule out the possibility that the reduced inhibitory processing reported here for upright faces reflects the familiarity of the stimuli rather than engagement of face processing. However, it is worth noting that the stimuli used as non-face objects were familiar (the images depicted a butterfly and a wheel). Furthermore, negative compatibility effects occurred when consonants served as the stimuli (manuscript under review). Together, these experiments with familiar non-face stimuli suggest that familiarity does not prevent negative compatibility effects from mounting.

Distractor inhibition during selective attention

Our data revealed that the contribution of distractor inhibition to focused attention depended on whether the distractor engaged face processing. This result provides novel evidence that the use of inhibition during selective attention is stimulus specific, which highlights the flexibility of inhibitory processing during focused attention. Importantly, this result also sheds new light on previous suggestions that distractor inhibition does not contribute to our ability to selectively attend, given that the supporting evidence came from a task that used famous faces as distractors [17], [29]. Egner and Hirsch [17] measured activation of the face-sensitive brain region (fusiform face area, FFA) while participants categorized names that appeared superimposed on a distractor face portraying a famous actor or politician. The results showed that FFA activity was not suppressed, which led to the suggestion that inhibition does not contribute to selective attention. Given the unique resilience of faces to selective attention (evidenced in these authors' data by a 41 ms compatibility effect for face distractors versus a 14 ms compatibility effect for name distractors), we suggest that the absence of evidence of inhibition may be specific to face distractors. Moreover, we recommend that future studies investigating mechanisms of selective attention avoid using face stimuli unless face processing is the specific topic of investigation.

Conclusions

The current research provides novel evidence that distractors that portray faces are subject to less inhibitory processing, and this reduced inhibition could contribute to the difficulty people experience ignoring faces. Future research is required to determine whether reduced inhibitory processing is specific to famous faces, or whether faces in general are subjected to less inhibitory processing regardless of the status of the person portrayed.