Natalie Mestry

Identifying sources of configurality in three face processing tasks

Participants performed three feature-complete face processing tasks involving detection of changes in: (1) feature size and (2) feature identity in successive matching tasks, and (3) feature orientation. In each experiment, information in the top (eyes) and bottom (mouths) parts of faces were manipulated. All tasks were performed with upright and inverted faces. Data were analyzed first using group-based analysis of signal detection measures (sensitivity and bias), and second using analysis of multidimensional measures of sensitivity and bias along with probit regression models in order to draw inferences about independence and separability as defined within general recognition theory (Ashby and Townsend, 1986). The results highlighted different patterns of perceptual and decisional influences across tasks and orientations. There was evidence of orientation specific configural effects (violations of perceptual independence, perceptual seperability and decisional separabilty) in the Feature Orientation Task. For the Feature Identity Task there were orientation specific performance effects and there was evidence of configural effects (violations of decisional separability) in both orientations. Decisional effects are consistent with previous research (Wenger and Ingvalson, 2002, 2003; Richler et al., 2008; Cornes et al., 2011). Crucially, the probit analysis revealed violations of perceptual independence that remain undetected by marginal analysis.

The Role of Configurality in the Thatcher Illusion: An ERP Study

The Thatcher illusion (Thompson in Perception, 9, 483–484, 1980) is often explained as resulting from recognising a distortion of configural information when ‘Thatcherised’ faces are upright but not when inverted. However, recent behavioural studies suggest that there is an absence of perceptual configurality in upright Thatcherised faces (Donnelly et al. in Attention, Perception & Psychophysics, 74, 1475–1487, 2012) and both perceptual and decisional sources of configurality in behavioural tasks with Thatcherised stimuli (Mestry, Menneer et al. in Frontiers in Psychology, 3, 456, 2012). To examine sources linked to the behavioural experience of the illusion, we studied inversion and Thatcherisation of faces (comparing across conditions in which no features, the eyes, the mouth, or both features were Thatcherised) on a set of event-related potential (ERP) components. Effects of inversion were found at the N170, P2 and P3b. Effects of eye condition were restricted to the N170 generated in the right hemisphere. Critically, an interaction of orientation and eye Thatcherisation was found for the P3b amplitude. Results from an individual with acquired prosopagnosia who can discriminate Thatcherised from typical faces but cannot categorise them or perceive the illusion (Mestry, Donnelly et al. in Neuropsychologia, 50, 3410-3418, 2012) only differed from typical participants at the P3b component. Findings suggest the P3b links most directly to the experience of the illusion. Overall, the study showed evidence consistent with both perceptual and decisional sources and the need to consider both in relation to configurality.

Discriminating Thatcherised from typical faces in a case of prosopagnosia

We report data from a prosopagnosic patient (PHD), and aged-matched control participants, from experiments where participants categorised individually presented emotional faces (experiment 1) and Thatcherised (from typical) faces (experiment 2). In experiment 2 participants also discriminated between simultaneously presented Thatcherised and typical faces. PHD was at chance categorising Thatcherised from typical faces. He was, however, able to discriminate between Thatcherised and typical faces, and partially able to categorise emotional faces. The results are discussed in terms of a loss of configural processing but preserved feature processing in PHD. The loss of configural processing impacts his categorisation of Thatcherised and typical faces, and his emotion processing, while his preserved feature processing supports his ability to categorise some emotional faces and his ability to discriminate between Thatcherised and typical faces.

Dual-target cost in visual search for multiple unfamiliar faces.

The efficiency of visual search for one (single-target) and either of two (dual-target) unfamiliar faces was explored to understand the manifestations of capacity and guidance limitations in face search. The visual similarity of distractor faces to target faces was manipulated using morphing (Experiments 1 and 2) and multidimensional scaling (Experiment 3). A dual-target cost was found in all experiments, evidenced by slower and less accurate search in dual- than single-target conditions. The dual-target cost was unequal across the targets, with performance being maintained on one target and reduced on the other, which we label “preferred” and “non-preferred” respectively. We calculated the capacity for each target face and show reduced capacity for representing the non-preferred target face. However, results show that the capacity for the non-preferred target can be increased when the dual-target condition is conducted after participants complete the single-target conditions. Analyses of eye movements revealed evidence for weak guidance of fixations in single-target search, and when searching for the preferred target in dual-target search. Overall, the experiments show dual-target search for faces is capacity- and guidance-limited, leading to superior search for 1 face over the other in dual-target search. However, learning faces individually may improve capacity with the second face.

A possible marker of configural processing at the N170: Converging evidence from typical participants and a case of prosopagnosia

Both orientation and Thatcherisation are thought to influence configural processing in faces (Boutsen, Humphreys, Praamastra & Warbrick, 2006). We explored the effect of orientation and level of Thatcherisation (typical face, eyes Thatcherised, mouth Thatcherised, or both features Thatcherised) on early ERP components. Participants showed evidence of inversion effects leading to increased amplitude for inverted faces at the N170 and reduced amplitude to inverted faces at the P2. However, the effect of Thatcherisation was only evident in the right hemisphere N170 where there was a reduction in N170 amplitude with level of Thatcherisation. These data suggest two distinct processes (consistent with Towler, Gosling, Duchaine, & Eimer, 2012). In a follow-up study using the same task we tested PHD, an individual with acquired prosopagnosia known to be unable to perceive the Thatcher illusion (Mestry, Donnelly, Menneer & McCarthy, 2012). We did so to explore whether the markers of orientation and Thatcherisation would survive in the ERP in the absence of sensitivity to the Thatcher illusion. PHD did show a significant effect of inversion at both the N170 and P2. However, PHD produced no effect of Thatcherisation at the N170, in contrast to the effect found with typical participants. The results suggest that the effect of Thatcherisation manifest in the right hemisphere N170 underpins the perception of the Thatcher illusion. As the effects of orientation at the N170 and P2 are also found in an individual who does not perceive the illusion, then they cannot reflect the type of configural processing affected by the Thatcher illusion (Donnelly, Cornes and Menner, 2012). In conclusion, the novel N170 Thatcherisation effect is a marker of a kind of configural processing present in typical face processing that is affected by Thatcherisation (Mestry, Menneer, Wenger, & Donnelly, 2012).

The Impacts of Inversion and Thatcherisation on Face Processing: Mapping between ERP and GRT

Prior work on the Thatcher illusion (Cornes et al., 2011; Mestry et al., 2012) has indicated both perceptual and decisional components of the illusion. We provide evidence of neural correlates associated with independent influences of inversion and level of feature manipulation (Thatcherisation) in the Thatcher illusion. We also provide an account of how ERP results can be interpreted within the configural processing framework defined by general recognition theory (GRT). We conducted a study examining inversion and the level of Thatcherisation (typical face, eyes Thatcherised, mouth Thatcherised, or both features Thatcherised) on a set of event-related components (the P1, N170, P2 and P3b) that have previously been considered in relation to the Thatcher illusion. Results reveal three independent effects: (1) inversion effects leading to increased amplitude for inverted faces at the N170 and P2 in the occipito-parietal area; (2) a reduction in N170 amplitude with level of Thatcherisation in the right hemisphere; and (3) inversion effects leading to increased amplitude for inverted faces at P3b in the centro-parietal area. This study reveals effects of Thatcherisation that are independent from effects of inversion in the ERP data. These effects can also be mapped to perceptual and decisional measures of configurality revealed in behavioural data using GRT (Mestry et al., 2012). We suggest the Thatcherisation effect at N170 may relate to violations of perceptual independence, the inversion effect across N170 and P2 may relate to violations of perceptual separability, and the inversion effect at P3b may relate to violations of decisional separability. Thus, we suggest a possible role for ERP data in providing a source of converging evidence for inferences regarding the role of perceptual and decisional factors in configurality.

Tests for configural processing in the Thatcher Illusion

Thatcherisation of facial features is immediately apparent only in upright faces. Detection of Thatcherisation is therefore widely upheld as being dependent on configural processing. Configural processing has clear predictions of perceptual dependence between facial features and of supercapacity processing. Perceptual dependence: Configurality in General Recognition Theory (GRT, Ashby & Townsend, 1986) can be conceptualized in terms of violations of perceptual independence, in which the perceptions of the eyes and mouth of a single face are positively correlated; for example, the more inverted the eyes look, the more inverted the mouth will look. Participants were briefly presented with normal, partially Thatcherised and fully Thatcherised faces, and were asked to report whether the outline, eyes and mouth were upright or inverted. Consistent with the predictions, multi-dimensional probit models (DeCarlo, 2003) revealed that within-stimulus perceptual interactions occurred more frequently for upright than inverted faces. However, for the Thatcherised stimulus itself, there was no overall correlation between the eyes and the mouth, suggesting a lack of configural processing for this type of stimulus. Processing capacity: Supercapacity processing is predicted to accompany configural processing (O’Toole, Wenger & Townsend, 2001; Wenger & Townsend, 2001). Therefore, the detection of Thatcherised features in upright faces should be marked by supercapacity processing of inverted eyes and mouths relative to the inversion of only eyes or only mouths. Response times were used to compute measures of processing capacity (capacity coefficient, Townsend & Wenger, 2004; proportional hazards ratio, Wenger & Gibson, 2006) in Thatcherised upright faces. Results demonstrated limited evidence of supercapacity processing, and only in some individuals, implying that configural processing is not necessary for processing of upright Thatcherised faces. In summary, predictions from GRT and of processing capacity were tested across two separate studies, with little evidence found for perceptual independence or supercapacity processing for Thatcherised face stimuli.

Detecting the Thatcher illusion in a case of prosopagnosia

We explored configural face processing in a prosopagnosic patient (PHD, Eimer and McCarthy, 1999) who does not produce an N170 is response for faces. In two sets of studies he was presented with two versions of the Thatcher illusion. In the first set, he was asked to detect Thatcherized from matched typical faces from successive single presentations of faces. He also performed a simultaneous 2 alternative forced choice (2AFC) discrimination task with the same stimulus set to address the question of whether pairs of faces were the same or different. In the second set he was asked to detect Thatcherized from matched typical faces. He also performed in control conditions where orientation decisions were made to isolated eye and mouth features, as well as eye and mouth features presented alone but within face outlines. The results were analyzed using d-prime and C to facilitate cross condition comparisons. The data showed PHD unable to detect Thatcherized from matched typical faces in either study 1 or 2. However, he was as sensitive as controls in the 2AFC discrimination condition of Study 1. In study 2 he showed evidence of moderate sensitivity to the identification of orientation for isolated features; this sensitivity was much enhanced for eyes by face outlines but hindered for mouths. We interpret these findings as showing intact feature processing that should be sufficient to allow the detection of the Thatcher illusion, as well as some evidence of relational processing for eyes but not mouths. However, simultaneous presentation of features and face outline does not allow selective attention to eyes that would enable detection of Thatcherized from matched typical faces. The results suggest one aspect of successful face categorization is to determine face-specific configural routines that allocate attention within faces and that these are missing in PHD.

Are deficits in emotional face processing preventing perception of the Thatcher illusion in a case of prosopagnosia

Behavioural studies using the Thatcher illusion are usually assumed to demonstrate configurality in upright face processing. Previously, we have reported on PHD, an individual with prosopagnosia, could not discriminate Thatcherized faces but showed some evidence for residual face processing (VSS, 09). Recent functional imaging data suggests a role for emotional expression perception in discriminating Thatcherized from neutral faces (Donnelly & Hadjikhani, in preparation). Here we report on a series of emotion perception tasks were conducted on PHD and control participants. Results for PHD revealed: (1) specific deficits for distinguishing magnitude of anger and disgust; (2) poor sensitivity when discriminating faces as one of two given emotions; (3) a within category deficit for intensity, but no intensity deficit between emotions unless disgust was present; (4) a different solution for PHD relative to controls in respect of a multidimensional scaling study for sameness judgements of faces varying in emotion identity and intensity. We consider possible relationships between PHDs emotion perception and his ability to discriminate Thatcherised from normal faces.