Erasing the face after-effect

Highlights

• Face after-effects decay exponentially after a blank delay period.

• Face after-effects are reduced even after brief presentation of an unrelated face.

• Unrelated faces reduce the time constant of face after-effects by 85%.

Abstract

Perceptual after-effects decay over time at a rate that depends on several factors, such as the duration of adaptation and the duration of the test stimuli. Whether this decay is accelerated by exposure to other faces after adaptation is not known. Our goal was to determine if the appearance of other faces during a delay period after adaptation affected the face identity after-effect. In the first experiment we investigated whether, in the perception of ambiguous stimuli created by morphing between two faces, the repulsive after-effects from adaptation to one face were reduced by brief presentation of the second face in a delay period. We found no effect; however, this may have been confounded by a small attractive after-effect from the interference face. In the second experiment, the interference stimuli were faces unrelated to those used as adaptation stimuli, and we examined after-effects at three different delay periods. This showed a decline in after-effects as the time since adaptation increased, and an enhancement of this decline by the presentation of intervening faces. An exponential model estimated that the intervening faces caused an 85% reduction in the time constant of the after-effect decay. In conclusion, we confirm that face after-effects decline rapidly after adaptation and that exposure to other faces hastens the re-setting of the system.

Introduction

Exposure to a stimulus can alter perception of a following one. For example, seeing a line with a clockwise tilt causes a subsequent vertical line to appear tilted counter-clockwise (Gibson and Radner, 1937, Harris and Calvert, 1989). Such after-effects have been observed after adaptation for many types of visual properties, such as curvature (Bales and Follansbee, 1935, Gheorghiu and Kingdom, 2007, Gibson, 1933), motion (Hershenson, 1989) and contour shape (Prins, 2009).

After-effects also occur for more complex stimuli such as faces (Webster and MacLeod, 2011). For example, adaptation to a distorted face results in a subsequently viewed ‘normal’ face being perceived as distorted in the opposite direction (MacLin and Webster, 2001). Face after-effects have been shown not just for shape but also for a number of complex facial properties, such as identity (Fox et al., 2008, Leopold et al., 2001), expression (Fox and Barton, 2007), gender (Oruc et al., 2011), and age (Lai et al., 2012, Lai et al., 2013). Although face after-effects are reduced by off-setting the location of the test stimuli by 5° relative to the adaptor stimuli (Afraz and Cavanagh, 2008), substantial after-effects remain even if the retinotopic contribution of low-level image properties are reduced by varying the size (Rhodes et al., 2007) or retinal location (Leopold et al., 2001, Roach and Webb, 2013) of the adaptor in relation to the test stimulus. This is consistent with other findings that indicate that face after-effects are not derived from low-level image properties, but from high-level representations (Butler et al., 2008).

After-effects are evidence of a short-term experience-dependent form of plasticity in the visual system. The benefits of these effects to perception are not entirely clear. In motion perception it has been suggested that after-effects may be evidence of a re-calibration or gain control, to make operation under varying conditions of motion more effective (Verstraten et al., 1994). For face after-effects, there is evidence that facilitation from brief exposure is accompanied by a lateral inhibition of perception of other faces (Oruç and Barton, 2010), and that this inhibition has a center-surround organisation (Rostamirad et al., 2009), which could enhance perception of the adapted face over similar faces. With longer adaptation, there is sharpened tuning of shape discrimination specific for the adapted face (Oruc and Barton, 2011), which could again benefit the distinction of the adapted face from other competing faces. Thus, face after-effects could also be manifestations of a type of gain control to enhance processing of the face being viewed.

To be effective in daily experience such benefits should be transient, as recently viewed faces continue to be replaced by new faces in new encounters. For visual adaptation in general, there has long been evidence that after-effects decay over time. This has been shown for displacement after-effects (Hammer, 1949), for example, and has been particularly well studied for the motion after-effect, whose decay has been modelled with an exponential function (Hershenson, 1989, Keck and Pentz, 1977, Taylor, 1963). Recent studies also show a rapid decline in face identity after-effects in the first few seconds after adaptation (Leopold et al., 2001, Leopold et al., 2005, Rhodes et al., 2007). On the other hand, there are also reports that face distortion after-effects may persist at a modest level for 1 to 7 days (Carbon and Ditye, 2011, Carbon et al., 2007).

One of the curious aspects of these latter studies of long-term after-effects is that these were found despite the fact that subjects continued to be exposed to other faces in the course of their daily routines. This contrasts with the study of effects over several seconds in the laboratory, in which the time between adaptation and the subject׳s response is filled by either a blank interval of varying duration (Leopold et al., 2001) or merely by prolonging the duration of the test stimulus (Rhodes et al., 2007). Although there are ‘de-adaptation’ studies that employ a strategy of using an opposing stimulus to generate after-effects to counter-act those from the first stimulus (Mesik et al., 2013), it is not known whether even brief introduction of other face stimuli in the period between the adaptor and test stimuli affects the dynamics of the face after-effect. However, notwithstanding the findings from studies of long-term after-effects (Carbon and Ditye, 2011, Carbon et al., 2007), it is plausible that the effects of adaptation to one face would be altered and possibly reduced by viewing of another face. The goal of this study was to test the hypothesis that face identity after-effects would be reduced by exposure to other face stimuli.